emg processing based measures of fatigue assessment during...

13
Review Article EMG Processing Based Measures of Fatigue Assessment during Manual Lifting E. F. Shair, 1,2 S. A. Ahmad, 1 M. H. Marhaban, 1 S. B. Mohd Tamrin, 3 and A. R. Abdullah 2 1 Department of Electrical and Electronic Engineering, Faculty of Engineering, Universiti Putra Malaysia (UPM), Selangor, Malaysia 2 Faculty of Electrical Engineering, Universiti Teknikal Malaysia Melaka (UTeM), Melaka, Malaysia 3 Department of Environmental and Occupational Health, Faculty of Medicine and Health Sciences, Universiti Putra Malaysia (UPM), Selangor, Malaysia Correspondence should be addressed to E. F. Shair; [email protected] Received 28 November 2016; Accepted 31 January 2017; Published 19 February 2017 Academic Editor: Stefan Rampp Copyright © 2017 E. F. Shair et al. is is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Manual liſting is one of the common practices used in the industries to transport or move objects to a desired place. Nowadays, even though mechanized equipment is widely available, manual liſting is still considered as an essential way to perform material handling task. Improper liſting strategies may contribute to musculoskeletal disorders (MSDs), where overexertion contributes as the highest factor. To overcome this problem, electromyography (EMG) signal is used to monitor the workers’ muscle condition and to find maximum liſting load, liſting height and number of repetitions that the workers are able to handle before experiencing fatigue to avoid overexertion. Past researchers have introduced several EMG processing techniques and different EMG features that represent fatigue indices in time, frequency, and time-frequency domain. e impact of EMG processing based measures in fatigue assessment during manual liſting are reviewed in this paper. It is believed that this paper will greatly benefit researchers who need a bird’s eye view of the biosignal processing which are currently available, thus determining the best possible techniques for liſting applications. 1. Introduction Musculoskeletal disorders (MSDs) caused by manual liſting tasks have been perceived for quite some time as one of the primary work-related injuries which influences the personal satisfaction of industrial workers around the world [1, 2]. Common reasons of MSDs in manual liſting are due to improper liſting and muscle fatigue. National Institute of Occupational Safety and Health (NIOSH) has published a technical report entitled Work Practices Guide for Manual Liſting as a reference to perform correct liſting [3]. However, in terms of muscle fatigue monitoring, researchers are still in search of finding the best processing technique to assess the human muscle condition. 2. Manual Lifting Despite the wide utilization of robots in modern industries, there are still numerous assignments in the industry that are performed physically by humans, such as carrying, liſting, pushing, or pulling. ese physical movements of the body oſten cause MSDs that can be subdivided into more particular and perceived regions, such as back disorders (BDs), upper limb disorders (ULDs), and lower limb disorders (LLDs) [4]. ULDs cover distinctive work-related musculoskeletal dissen- sions in the arm, wrist, hand, elbow, neck, and shoulder, while LLDs are associated with the lower areas of the body such as feet, legs, and hips. e anatomical zones of the body mostly influenced by these MSDs issues due to manual handling are in the spine and lumbar (back area), and the arm, and wrist (upper limb area). ese are further supported by various researches done in previous years on muscle areas investi- gated during manual liſting and are summarized in Table 1. Manual errands that are performed improperly by work- ers in the industries may most of the time bring pain in the lumbar spinae [5]. is situation happens when one individ- ual performs forward flexion; the weight of abdominal area and the load affects the lumbar spinae, which is adjusted by Hindawi BioMed Research International Volume 2017, Article ID 3937254, 12 pages https://doi.org/10.1155/2017/3937254

Upload: others

Post on 16-Jul-2020

1 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: EMG Processing Based Measures of Fatigue Assessment during …downloads.hindawi.com/journals/bmri/2017/3937254.pdf · 2019-07-30 · BioMedResearchInternational 3 5. Electromyography

Review ArticleEMG Processing Based Measures of Fatigue Assessment duringManual Lifting

E F Shair12 S A Ahmad1 M H Marhaban1 S B Mohd Tamrin3 and A R Abdullah2

1Department of Electrical and Electronic Engineering Faculty of Engineering Universiti Putra Malaysia (UPM) Selangor Malaysia2Faculty of Electrical Engineering Universiti Teknikal Malaysia Melaka (UTeM) Melaka Malaysia3Department of Environmental and Occupational Health Faculty of Medicine and Health Sciences Universiti PutraMalaysia (UPM)Selangor Malaysia

Correspondence should be addressed to E F Shair ezreenutemedumy

Received 28 November 2016 Accepted 31 January 2017 Published 19 February 2017

Academic Editor Stefan Rampp

Copyright copy 2017 E F Shair et alThis is an open access article distributed under theCreativeCommonsAttribution License whichpermits unrestricted use distribution and reproduction in any medium provided the original work is properly cited

Manual lifting is one of the common practices used in the industries to transport or move objects to a desired place Nowadayseven though mechanized equipment is widely available manual lifting is still considered as an essential way to perform materialhandling task Improper lifting strategies may contribute to musculoskeletal disorders (MSDs) where overexertion contributes asthe highest factor To overcome this problem electromyography (EMG) signal is used to monitor the workersrsquo muscle conditionand to find maximum lifting load lifting height and number of repetitions that the workers are able to handle before experiencingfatigue to avoid overexertion Past researchers have introduced several EMG processing techniques and different EMG features thatrepresent fatigue indices in time frequency and time-frequency domainThe impact of EMG processing based measures in fatigueassessment during manual lifting are reviewed in this paper It is believed that this paper will greatly benefit researchers who needa birdrsquos eye view of the biosignal processing which are currently available thus determining the best possible techniques for liftingapplications

1 Introduction

Musculoskeletal disorders (MSDs) caused by manual liftingtasks have been perceived for quite some time as one of theprimary work-related injuries which influences the personalsatisfaction of industrial workers around the world [1 2]Common reasons of MSDs in manual lifting are due toimproper lifting and muscle fatigue National Institute ofOccupational Safety and Health (NIOSH) has published atechnical report entitled Work Practices Guide for ManualLifting as a reference to perform correct lifting [3] Howeverin terms of muscle fatigue monitoring researchers are still insearch of finding the best processing technique to assess thehuman muscle condition

2 Manual Lifting

Despite the wide utilization of robots in modern industriesthere are still numerous assignments in the industry that are

performed physically by humans such as carrying liftingpushing or pulling These physical movements of the bodyoften causeMSDs that can be subdivided intomore particularand perceived regions such as back disorders (BDs) upperlimb disorders (ULDs) and lower limb disorders (LLDs) [4]ULDs cover distinctive work-related musculoskeletal dissen-sions in the arm wrist hand elbow neck and shoulder whileLLDs are associated with the lower areas of the body such asfeet legs and hips The anatomical zones of the body mostlyinfluenced by these MSDs issues due to manual handling arein the spine and lumbar (back area) and the arm and wrist(upper limb area) These are further supported by variousresearches done in previous years on muscle areas investi-gated during manual lifting and are summarized in Table 1

Manual errands that are performed improperly by work-ers in the industries may most of the time bring pain in thelumbar spinae [5] This situation happens when one individ-ual performs forward flexion the weight of abdominal areaand the load affects the lumbar spinae which is adjusted by

HindawiBioMed Research InternationalVolume 2017 Article ID 3937254 12 pageshttpsdoiorg10115520173937254

2 BioMed Research International

Table 1 Common muscles investigated during manual lifting

Reference Muscle areasKamarudin et al [7] Biceps brachii and triceps brachii

Voge and Dingwell [8] Biceps middeltoid midtrapeziusand postdeltoid

Zawawi et al [9] Biceps brachii

Al-Ashaik et al [10] Biceps brachii erector spinaeanterior deltoid and trapezius

Granata and Marras [11]Erector spinae external abdominal

obliques internal abdominalobliques rectus abdominis and

latissimus dorsi

Roy et al [12] Longissimus thoracis iliocostalislumborum and multifidus

Seroussi and Pope [13] Erector spinae and external oblique

Graham et al [14]Lumbar erector spinae thoracic

erector spinae and rectusabdominis

Gagnon et al [15]Thoracic erector spinae lumbarerector spinae internal oblique

external oblique rectus abdominisand latissimus dorsi

Dolan and Adams [16] Erector spinae

Bonato et al [17] Iliocostalis lumborum longissimusthoracis and multifidus

Potvin [18] Thoracic erector spinae and lumbarerector spinae

Kingma and Van Dieen [19]External oblique internal oblique

rectus abdominis iliocostalislumborum longissimus thoracis

and pars thoracis

Shin and Kim [20]Erector spinae rectus abdominislatissimus dorsi internal oblique

and external oblique

Cholewicki et al [21]Lumbar erector spinae thoracicerector spinae latissimus dorsi

internal oblique external obliqueand rectus abdominis

trunk muscles (predominately by erector spinae muscles)[6] This movement causes the erector spinae muscles tomake incredible expansive strengths for adjusting the bodyEven though there are numerous assistive gadgets such as lifttrucks trolleys and derricks that can help reduce loads on thelumbar spinae however in many cases these instruments arecostly and inconvenient

Lifting strategies are an important measure during liftingand because physically correct lifting is significant theirimpact onmusculoskeletal well-being is broadly talked about[22] The impact of lifting techniques including positionwidth (characterized as the separation between the feet inthe average sidelong heading with the sagittal symmetry ofposition) has been considered recommending the utilizationof a wide position to lessening the burden on the spine [23]Lifting procedures (stoop or squat and semisquat) can alsohave a significant effect on spine stacking and steadinessduring lifting Past studies have researched lifting methods in

biomechanical terms to give some mediation methodologiesand recognize the correct lifting procedures [24 25] Squatlifting seems to have less lumbar shear stretch and place lessweight on noncontractile connective tissues Stoop lifting hasall the earmarks of being more regular and less exhaustingSemisquat lifting might be a decent trade-off lift despitethe fact that early confirmation recommends perhaps higherlumbar minutes [22]

3 Muscle Fatigue Assessment

Muscle fatigue is characterized as the long lasting deterio-ration of the performance of the human operator to createforce [24 25] A personrsquos muscle capabilities may differ fromone another but the characteristics should have some resem-blance Past researchers have outlined several features ofmuscle fatigue such as an increment in amplitude and thetransition from high frequency range to low frequency [26]

To assess this muscle fatigue characteristics various tech-niques are available depending on the usage At thismomenteven though the measurement of muscle fatigue by use ofinvasive means such as blood test (blood lactate level andblood oxygen level) andmuscle biopsies (pHofmuscle) offershigher accuracy compared to noninvasive techniques it isunsuitable for some applications such as sports ergonomicsand occupational therapy [27] The use of noninvasiveapproach to acquire muscle signals such as electromyog-raphy (EMG) sonomyography (SMG) mechanomyography(MMG) and near infrared spectroscopy (NIRS) for researchpurposes is considered to bemore convenient and easy to use[28] Every technique has its own advantages and disadvan-tages based on its applications For the purpose of assessingmuscle physiology during manual lifting EMG is the mostcommonly used tool and has been well reported by manyresearchers [29]

4 Electromyography Signal

EMG is the measure of electrical potential present on theskin in consequence to a muscle contraction that representsthe neuromuscular activities [26] It can be measured by twomethods (1) applying electrodes to the skin surface (non-invasive) or (2) intramuscular (invasive) within the muscleEven though intramuscular EMG (imEMG) provides addi-tional benefits to overcome the drawbacks of surface EMG(sEMG) such as maintaining robust electrode contact withthe skin and the capacity to record from profound muscleswith little EMG crosstalk [27] sEMG and imEMG havebeen proven to have equal classification performance for dataof comparative nature (unmodulated) However the perfor-mance of imEMG diminished compared to surface whentested on modulated data [28 29]

The greater selectivity of imEMG with respect to sEMGis due to wires exposed only at the tip but then again may bea hindrance since the signal may provide local rather thanglobal information where imEMG recordings depend on therecruitment ofmotor units it may be the case that insufficientinformation was captured at low amplitudefrequency [27]

BioMed Research International 3

5 Electromyography Preprocessing

When dealing with EMG signal noise removal is a veryimportant factor to be taken into consideration This is dueto the sensitive nature of EMG signal itself that different typesof noises or artifacts unavoidably contaminate itThese noisesare caused by various sources originates from either the skin-electrode interface hardware that intensifies the signal orother external sourcesTherefore before proceeding with theanalysis and classification of EMG signal a preprocessing isusually introduced to remove these noises Some of the noisesthat may affect the signals are inherent noise in electronicsequipment ambient noise motion artifact ECG artifactsinherent instability of signal and crosstalk However thisreview only covers three types of noises motion artifact ECGnoise and device noise

51 Motion Artifact Cutting edge technology is considerablyinvulnerable to some of the EMG noises but not to themotion artifact as it has frequency spectra that are affectingthe low frequency part of EMG signal This motion artifactoriginates at the electrode-skin interface where there existmuscle movements underneath the skin and force impulsethat goes through the muscle and the skin underlying theelectrodes causing movements [30] These movements willresult in a time-varying voltage across the two electrodes

To remove the low frequency noise caused by the motionartifact a study has been conducted to ascertain a reasonablevalue for the high-pass corner frequency filter The result ofthe study recommends the use of a Butterworth filter with a20Hz corner frequency and a 12 dBoct slope [31]

52 ECG Noise When EMG signal is measure close to theheart ECG bursts may pollute the EMG recording whichcan affect the analysis and result in misinterpretations [32]This is a natural relic that is often unavoidable It can bediminished by proper skin preparation and adjusted positionof the ground electrode To date other than the conventionalECG removal procedures such as the use of high past filteringseveral ECG noise removal techniques have been introducedthat can clean these ECG bursts but still maintaining theregular EMG characteristics

In 2009 Lu et al came out with a study to evaluate theperformance of recursive-least-square adaptive filter in theremoval of ECG interference from EMG signal [33] Theresult of the study shows that due to fast convergence ofthe recursive-least-square algorithm the filter is reliable toeffectively remove ECG noise

Later in 2012 independent component analysis (ICA)wasproposed to remove the ECG contamination The filteringeffects are measured in terms of root mean square relativeerrors and linear envelopes correlation of uncontaminatedand contaminated EMGThis technique excellently producesgood results when EMG and ECG are statistically indepen-dent [34]

53 Device Noise Another type of noise that usually affectsEMG signal is the noise generated inside the EMG device

itself that is commonly referred to as inherent noise inelectronics equipment [35] This noise cannot be completelyeliminated as all electronic equipment generates noise [36]However the severity of the noise can be reduced using high-quality electronic components

6 Electromyography Processing

Biosignal processing is a critical part in biomedical engineer-ing in order to classify the frequency content of a signalThesesignals include EMG electrocardiography (ECG) and elec-troencephalography (EEG)Most of these biosignals are non-stationary signals due to its time-varying characteristics [37ndash39] The problem of this nonstationary signal is the processof assuming it to be stationary over short-time intervalswhere stationary analysis techniques are used However thisassumption is not always suitable and further methods fornonstationary processes are needed

Past researchers have developed several techniques toprocess the nonstationary signal specifically the EMG signaland these include parametric and nonparametric approaches[40 41] The differences between these two approaches aremainly due to the parameters involved Since most of the sig-nals encounters are unknown signals without known param-eters the development of nonparametric approach whichinvolves a variety of transforms is actively researched com-pared to the parametric approach that requires the modelingof the nonstationary process Figure 1 provides an overviewof the EMG signal processing methods that are reviewed inthe manuscript

61 Parametric Approach In the field of EMG processingone type of parametric approach that has been explored isbased on the time-varying linear predictive models Thisincludes time-varying autoregressive (TVAR) model whereits parameters vary with time The time-varying spectrum inTVAR is estimated from the time-varying model parametersand the nonstationary signalrsquos instantaneous frequency (IF)can be extracted [42] IF represents the spectral peak locationof the EMG signal as it varies with time [43] Even thoughthe poles and zeros of the estimated model can be estimateddirectly from the EMG signal it is not guaranteed that thetime-varying poles remain inside the unit circle on the 119911-plane thus making TVAR model temporarily unstable [44]The constraints can be easily imposed by factorizing thedenominator of TVAR in direct form and formulating it incascaded form [45] Computation of the mean frequency(MNF) is taken as the time average of the IF estimation basedon the cascaded model in each interval [46]

Research conducted by Al Zaman et al has comparedTVAR with short-time Fourier transform (STFT) and con-cludes that even though both TVAR and STFThave producedsimilar decreasing muscle fatigue patterns from the MNFTVAR has achieved a better performance with the slope andinterception of the linear regression line is closer [47]

Another parametric approach that has been used is thetime-varying autoregressive moving average (TVARMA)Studies have been made to compare the performance of

4 BioMed Research International

Electromyography processing

Parametric approach

Time distribution Frequency distribution

Linear

transform (STFT)

Spectrogram

Wavelet transform (WT)

(TVAR) average (TVARMA)

Bilinear

(WVD)

(CWD)

Time-varying autoregressive Time-varying autoregressive moving

Time-frequency distribution

Nonparametric approach

B-distributionWigner-Ville distribution

Choi-William distributionS-transform

Short-time Fourier

Figure 1 Overall view of the electromyography signal processing methods that are reviewed in this manuscript

TVAR and TVARMA in analyzing EMG signals and it isconcluded that AR models are more favorable than ARMAbecause of the following reasons (a) estimation of ARmodelsis moderately basic since they are represented by an arrange-ment of direct mathematical statements (b) AR modelsare predominant in displaying the low recurrence sEMGsegment and (c) fluctuation of the AR model estimation onshorter fragments is lower than for ARMAmodels [40]

62 Nonparametric Approach Another EMG processingmethod is the nonparametric approaches which includeSTFT spectrogram andwavelet transformThese approachesappear to be more favorable to researchers since the fatiguesindices can be obtained directly from the EMG signal withoutknowing its parametersThe analysis of the EMG signal basedon the nonparametric approaches can be further dividedinto three types of distribution techniques time domainfrequency domain and time-frequency domain [48ndash50]

621 Time Distribution (TD) Extraction of time domainfatigue indices is very simple and easy and involves lowcomputational complexity compared to the other two tech-niques since it does not involve any transformation [51] Itcan be directly obtained from the time representation of theraw EMG signal just by doing some simple mathematicalstatisticsThe idea of fatigue is approximately connected withthe amount and rate of change of some variables that reflectmuscle alterations amid sustained contractions [52]

Listed in Table 2 are several fatigue indices in the timedomain that have been explored by previous researchers

Even though the analysis based on time domain is alreadyestablished and arewidely used since the last decade howeverit is considered less accurate since their calculations are basedon EMG signal amplitude values where there aremuch inter-ference acquired through the recording especially for fea-tures that are extracted from energy property [53] Anothermajor disadvantage of this time domain features that affectsthe accuracy comes from the nonstationary property of theEMG signal where the statistical properties changes overtime but time domain features assume the data as stationarysignal [54]

Nowadays researches involving time distribution onlyfocus on finding new fatigue indices with higher robustnesswhile most of other researchers nowadays opt for time-frequency analysis which is considered to be more accurate

622 FrequencyDistribution (FD) Inmany studies of fatiguemuscle frequency domain features are usually used to extractinformation from a signal To date there are more than20 fatigue indices in the frequency domain where two ofthe widely used parameters are the mean frequency (MNF)and median frequency (MDF) [75] Other fatigue indices arelisted in Table 3 Several techniques have been used to extractinformation fromEMG signals such as fast Fourier transform(FFT) power spectral density (PSD) and parametric meth-ods (ie AR model)

The EMG signal should undergo Fourier transform inorder to represent the signal in frequency domain The

BioMed Research International 5

Table 2 Fatigue indices in time domain

Reference Fatigue indicesMalinzak et al [55] Integrated EMG (IEMG)Arabadzhiev et al [56] Root mean square (RMS)Suetta et al [57] Mean absolute value (MAV)

Oskoei and Hu [58] Modified mean absolute value type 1(MAV1)

Villarejo et al [59] Modified mean absolute value type2 (MAV2)

Phinyomark et al [60] Simple integral square (SSI)Zardoshti-Kermani et al [61] Variance of EMG (VAR)Tkach et al [62] v-order (V)Zardoshti-Kermani et al [61] Log detector (LOG)Kiguchi et al [63] Waveform length (WL)Al Omari et al [64] Average amplitude change (AAC)

Siddiqi et al [65] Difference absolute standarddeviation value (DASDV)

Kilbom et al [66] Zero crossing (ZC)AlOmari and Liu [67] Myopulse percentage rate (MYOP)Phinyomark et al [68] Willison amplitude (WAMP)Rogers et al [69] Slope sign change (SSC)

Buchenrieder [70] Mean absolute value slope(MAVSLP)

Venugopal et al [71] Multiple hamming windows(MHW)

Du and Vuskovic [51] Multiple trapezoidal windows(MTW)

Phinyomark et al [72] Histogram of EMG (HIST)Al-Quraishi et al [73] Autoregressive coefficient (AR)Chang et al [74] Cepstral coefficient (CC)

definition of Fourier transform in the form of power spec-trum is shown in

119878119909 (119891) = 111987910038161003816100381610038161003816100381610038161003816intinfin

minusinfin119909 (119905) sdot 119890minus2120587119891119905119889119905

100381610038161003816100381610038161003816100381610038162

(1)

where 119909(119905) is the signal in time domainHowever the fluctuations of the EMG frequency compo-

nent due to the adjustments in the muscle force length andcontraction speed throughout time have caused challenges inthe usage of FFT and other traditional processing methodsThe fundamental confinement of a FFT is that it cannot givesimultaneous time and frequency localization [83]Thereforeinvestigation of the EMG signal in dynamic contraction(ie repetitive lifting) utilizing these methods may not besuccessful since it requires the signal to be stationary

623 Time-Frequency Distribution (TFD) Time-frequencyrepresentation (TFR) of a signal maps a one-dimensionalsignal of time into a two-dimensional of time and frequencyIn analyzing modifying and synthesizing nonstationarysignals TFR arewidely used since both representation of timeand frequency are taken into account thus leading to higheraccuracy [84]

Table 3 Fatigue indices in frequency domain

Reference Fatigue indicesMerletti and Lo Conte [76] Mean frequency (MNF)De Luca [77] Median frequency (MDF)Khanam and Ahmad [78] Peak frequency (PKF)Khanam and Ahmad [78] Mean power (MNP)Khanam and Ahmad [78] Total power (TTP)

Altaf et al [79] The 1st 2nd and 3rd spectralmoments (SM1 SM2 and SM3)

Phinyomark et al [80] Power spectrum ratio (PSR)

Gonzalez-Izal et al [81] Instantaneous frequency variance(119865var)

Georgakis et al [75] Averaged instantaneous frequency(AIF)

Dimitrov et al [82] Dimitrovrsquos spectral fatigue index(FInsm5)

In general TFDs consist of linear TFDs and bilinear(quadratic) TFDs The most basic form of TFD technique isthe short-time Fourier transform (STFT) STFT is one of thelinear TFDs that have beenused to analyze EMGsignals alongwith spectrogram wavelet transform S-transform and soforth Examples of the bilinear TFDs include theWigner-Villedistribution (WVD) and Choi-William distribution (CWD)

Linear TFD

Short-Time Fourier Transform (STFT) To overcome thedisadvantages of time distribution and frequency distributionand satisfy the stationary condition it is common to separatelong haul signals into blocks of narrow fragments [85] Thisought to be sufficiently narrow to be viewed as stationary andtake the FT of every segment Every FT gives the spectralinformation of a different time-slice of the signal givingsimultaneous time and frequency estimation

This was proposed by Gabor who had developed theSTFT the extended version of FT [86]

STFT119909 (119905 119908) = intinfin

minusinfin119909 (120591)119908 (120591 minus 119905) 119890minus2120587119891120591119889120591 (2)

where 119909(120591) is the EMG signal 119908(120591 minus 119905) is the observationwindow and 119905 is the variable that slides the window over thesignal 119909(120591)

The choice of the window function in STFT is criticalin order to get accurate results The shape of the windowcan be either rectangular Gaussian or elliptic depending onthe shape of the signal For sampled STFT using a Gaussianwindow it is often called Gabor transform Even though thewindow should be narrow enough to ensure that the portionof the signal that falls within the window is stationary it mustnot be too narrow since it will lead to bad localization in thefrequency domain For infinitely long window 119908(119905) = 1 willeventually cause STFT to turn into FT providing excellentfrequency localization but no time localization In contrastto infinitely short window 119908(119905) = 120575 will result in the time

6 BioMed Research International

signal (with a phase factor) which will provide excellent timelocalization but no frequency localization

Research byMacIsaac et al usesMNFas fatigue index andhas drawn three important conclusions First the nonstation-arities in EMGdo not seem to affect theMNF values obtainedfrom Fourier transform Secondly the STFT method tomeasure MNF has successfully midpoints out the impacts ofnonstationarities in dynamic compressions Third the STFTis capable of identifying a descending pattern with fatigue indynamic contraction as long as the range of motion remainsconsistent across sequential time interval [87]

Since the STFT is basic in idea and execution and isequipped for distinguishing a pattern in muscle fatigue itis an undeniable choice for applications that require simpleanalysis with acceptable results

Spectrogram The squared magnitude of the STFT is calledspectrogram It can be expressed as

119878119909 (119905 119891) =10038161003816100381610038161003816100381610038161003816intinfin

minusinfin119909 (120591)119908 (120591 minus 119905) 119890minus2120587119891120591

100381610038161003816100381610038161003816100381610038162

119889120591 (3)

where 119909(120591) is the EMG signal 119908(120591 minus 119905) is the observationwindow and 119905 is the variable that slides the window over thesignal 119909(120591)

Spectrogram can be used to obtain the power distributionand energy distribution of the signal along the frequencydirection at a given time In EMG processing the instan-taneous energy is useful to separate muscle activation fromthe baseline which is called the segmentation process Seg-mentation is important in reducing the high computationalcomplexity of TFDs The muscle activation segmentationprocess by using spectrogram has been proposed by Shair etal and resulted in a mean absolute percentage error (MAPE)of just 1404 [88]

For both STFT and spectrogram there is a compromisebetween the time-based and frequency-based perspective ofa signal Both time and frequency are represented in limitedprecision where precision is controlled by the span of thewindow and the size of the window chosen will be the samefor all frequencies

In 2010 Jankovic and Popovic have presented in theirpaper the use of spectrogram in the analysis ofmuscle fatigueThey highlighted that even though the traditional MDFtechnique provides good sensitivity for fatigue indicationhowever there is a significant slope error particularly for lowactivity of coinitiated muscles They proposed a hybrid ofalternative methods (spectrogram and scalogram) that cansuccessfully provide complete information but at the sametime produce less error prediction for low-level coinitiatedmuscles [89] As a recommendation for improvement theyalso proposed further investigation on the robustness of thismethod in dynamic exercise contractions

Later in 2016 Zawawi et al came out with a detailedanalysis of EMG signals by using spectrogram for manuallifting application By taking the average instantaneous RMSvoltage (119881rms(119905)) as fatigue index she concludes that asthe lifting height is increased the average 119881rms(119905) is alsoincreased However as the average 119881rms(119905) decreased thenumber of lifting repetitions is increased [90]

Wavelet Transform (WT) A wavelet is a waveform of effec-tively limited duration that has an average value of zero Someof its properties are short-time localized waves with zerointegral value the possibility of time shifting and flexibilityWavelet analysis produces a time-scale view of the signalTheresult of a continuous wavelet transform (CWT) is waveletcoefficients Multiplying each coefficient with the appropri-ately scaled and shifted wavelet yields the constituent waveletof the original signal Equation (4) shows the formula forCWT

CWT119909 = intinfin

minusinfin119909 (120591) 1radic|119886|Ψ (

120591 minus 119905119886 ) 119889120591 (4)

where 119905 is the translation 119886 is the scale parameter and Ψ isthe mother wavelet

Yochum et al have proposed a new fatigue index basedon the continuous wavelet transform (CWT) named 119868CWTand have compared it to other fatigue indices from literaturein terms of their sensitivity to noiseThe results show that thisnew fatigue index quantifies the EMG signal elongation dur-ing a contraction and thus makes it a suitable fatigue index119868CWT is less subjected to noise and less truncation dependentcompared to other fatigue indices based on the frequency likeMNF and MDF [91]

A comparison has been made by Dantas et al betweenSTFT and CWT in evaluating muscle fatigue in isometricand dynamic contractions The after effects of this studyexhibit that CWT and STFT analysis give comparable fatigueestimates (slope of MDF) in isometric and dynamic contrac-tions However the aftereffects of CWT for both contractionsindicate less variability (higher accuracy) in EMG signalanalysis contrasted with STFT [92]

S-Transform Another TFD technique exists is the S-transform This technique has been widely used in diverseareas of telecommunication power quality geophysics andbiomedicine due to its obvious advantages in processingnonstationary signals [93] In the area of EEG and ECGdespite the use of S-transform in the time-frequency analysisthis technique is also a good alternative for denoising andremoval of artifacts that exist in ECG and EEG signals [94]Even though it has been explored in the areas of EEG andECG there are no available researches conducted in utilizingthis technique for EMG signal

S-transform was introduced in 1996 by StockwellMansinha and Lowe It is a hybrid of two advanced signalprocessing techniques which are STFT and WT [95] Dueto this it inherits the good qualities from both techniquesIt provides good resolution in both time and frequency andallows users to assess any frequency component in the time-frequency domain without the need of using any digital filterEven though S-transform uses a variable window length tomaintain a good time-frequency resolutions for all frequen-cies it still retains the phase information using a Fourierkernel

Expression for S-transform is shown in

ST119909 (119905 119891) = intinfin

minusinfin119909 (120591)

10038161003816100381610038161198911003816100381610038161003816radic2120587119890

minus(120591minus119905)212059122119890minus1198952120587119891120591119889120591 (5)

BioMed Research International 7

where 120591 and119891 denote the time of the spectral localization andFourier frequency respectively

The development of the S-transform is led by the urge toovercome the low resolution of STFT and the absence of thephase information in CWT Since the features had existed inthe S-transform it is sometimes viewed as a variable slidingwindow STFT or a phase-corrected CWT [96]

Although S-transform has better time-frequency resolu-tion than STFT the resolution is still far from perfect andneeds improvement To date there are several improved S-transform introduced by researchers such as the modified S-transform [96] and discrete orthonormal S-transform [94]

Bilinear TFD

Wigner-Ville Distribution (WVD) Wigner first introducedWVD in the area of quantum mechanics in the year 1932and later in 1948 Ville developed and applied the sametransformation to signal processing and spectral analysis[97]

Compared to STFT and WT WVD does not contain awindowing function and thus frees WVD from the smearingeffect due to the windowing function As a result it providesbest possible temporal and frequency resolution in the time-frequency plane It possesses several interesting propertiessuch as energy conservation real-valued marginal prop-erties translation covariance dilation covariance instanta-neous frequency and group delay

However because of its quadratic nature when dealingwith signals with several frequency components WVD suf-fers from the so-called cross components (inference terms)which represent significant defects in this method

These interference terms are troublesome since they mayoverlap with autoterms (signal terms) and in this manner itis hard to visually interpret the WVD image It creates theimpression that these terms must be available or the goodproperties of WVD (localization group delay instantaneousfrequency marginal properties etc) cannot be fulfilledThere is also a trade-off between the amount of interferencesand the number of good properties These are known issueswith the WVD spectrum and there are several ways to com-pensate these cross terms that has been discussed by previousresearchers [98ndash100] Equation (6) shows the formula forWVD

119882119909 (119905 119891) = intinfin

minusinfin119909(119905 + 1205912) 119909

lowast (120591 minus 1205912) 119890minus2119895120587119891120591119889120591 (6)

where 119909(119905 + 1205912)119909lowast(120591 minus 1205912) is the instantaneous autocorre-lation function and lowast shows conjugate operation

In the occasion that time smoothing window 119892(119905) anda frequency smoothing window ℎ(119905) are connected toWVDWVDwill then transform into the smoothed-pseudo-Wigner-Ville distribution (SPWVD) as composed in

SPWVD119909 (119905 119891)= ∬ℎ (119905 minus 120591) 119892 (119891 minus 120576)119882119909 (120591 120576) 119889120591 119889120576

(7)

where119882 is the WVD

A research by Subasi and Kiymik in 2010 has comparedthe performance between STFT SPWVD and CWT in termsof accuracy specificity and sensitivity The results suggestedthat all three methods provide almost similar performanceand are found to be satisfactory despite the fact that there arelittle differences between the results [99]

Choi-William Distribution (CWD) CWD is a member ofCohenrsquos class distribution function and was proposed in theyear 1989 by [101] It is able to avoid one of the main problemsof WVD which is the presence of interference in regionswhere one would expect zero power values Adoption of theexponential kernel in the distribution helps to suppress theeffect of cross term [102] CWD is given by

CWD119909 (119905 119891)

= intinfin

minusinfinintinfin

minusinfin119860119909 (120578 120591)Φ (120578 120591) 119890(1198952120587(120578119905minus120591119891))119889120578 119889120591

(8)

where

119860119909 (120578 120591) = intinfin

minusinfin119909(119905 + 1205912) 119909

lowast (119905 minus 1205912) 119890minus1198952120587119905120578119889119905 (9)

and the kernel function is given by

Φ(120578 120591) = 119890minus120572(120578120591)2 (10)

By studying the influence of muscle contraction towardsthe frequency content of EMG signal usingWVD and CWDAlemu et al observed that cross terms existing in WVD aregreatly reduced by CWD thus resulting in easier interpreta-tion with no loss of definition due to cross terms [103] Thereduction of the cross terms is due to the smoothing param-eters introduced in CWD However this still depends on thesmoothing parameter value selected As observed from theresult itself when the smoothing parameter decreased thereduction of the cross terms is better This however affectsthe time and frequency resolutions by reducing it and leadsto more signal loss As the smoothing parameter approachesinfin it resembles WVDmore [103]

B-Distribution Karthick et al introduced B-distribution in2015 to attenuate the cross terms exist in WVD and CWDThis is realized by introducing a smoothing function referredto as quadratic time-frequency distribution (QTFD) [104]Expression of the QTFD is given as follows [105]

119901 [119899 119896] = 2 sum|119898|lt1198722

119866 [119899119898] lowast 119911 [119899 + 119898] 119911

lowast [119899 minus 119898] 119890minus1198952120587119896119898119872(11)

where 119866[119899119898] is the smoothing kernel and operator lowastrepresents convolution

Kernel of the B-distribution is given by

119866 [119899119898] = ( |2119898|cosh2119899)

120573

lowast (sin 119888119898) (12)

where 120573 is a smoothing parameter

8 BioMed Research International

Three fatigue indices namely instantaneous median fre-quency (IMDF) instantaneousmean frequency (IMNF) andinstantaneous spectral entropy (ISPEn) were used alongsidethe B-distribution and it is found that all three features aredistinct in both fatigue and nonfatigue conditions The studyconcludes that B-distribution may be useful in EMG analysisunder various clinical and normal conditions

Modified B-Distribution A year later the same author cameout with a study using modified B-distribution to monitorthe progression of muscle fatigue Since the performance oftime-frequency distribution is assessed based on its abilityto reduce cross term and provide closely spaced frequencycomponents representation this modified version success-fully removes cross term interference while maintaininghigh time-frequency resolution [106] It is demonstratedthat modified B-distribution based time-frequency distribu-tion performs better in suppressing cross terms with goodtime and frequency resolution for multicomponent signalscompared with other above-mentioned techniques [107]Recently this technique has been widely used to analyzenonstationarities related to EEG signals heart rate signalsand accelerometer data based on fetal movements [108 109]

7 Discussion and Conclusion

The use of EMG processing to assess muscle fatigue duringmanual lifting was reviewed and several fatigue indices wereintroduced Research in the area of EMG signal currentlyevolves around the sensitivity variability and repeatability ofthe fatigue indices as well as the best processing techniqueswith high efficiency and less computational complexityDespite the use of traditional methods such as time distribu-tion and frequency distribution time-frequency distributionis found to be more superior in terms of monitoring since itenables the user to observe the progress of the signal in timeand frequency This is very important as at the point whenmuscle fatigue sets in frequency compression can happenUnlike the conventional frequency distribution techniquetime-frequency analysis demonstrates the time when anychanges in frequencies occur

In recent years researchers had identified that the bilinearTFD could perform better than the linear TFD such asSTFT spectrogram and WT due to the fact that it doesnot suffer from the smearing effects cause by windowingfunction Nonetheless on account of its quadratic naturebilinear TFD suffers the cross term effectsWith the existenceof high resolution Cohen class TFD such as the modified B-distribution introduced in 2016 the cross term effects canbe removed and at the same time maintain the high time-frequency resolution

Other than the processing techniques listed in this paperthere are still several techniques such as Hilbert-Huangtransform which have been used to analyze the EMG signalsThere are also already well established techniques in otherresearch areas and are proven to be effective and have yetto be implemented in analyzing EMG signals such as the S-transform This technique may be suitable for nonstationarysignals like EMG itself due to its good nature including

linearity lossless reversibility multiple resolution good time-frequency resolution and simple algorithm In conclusion itcan be seen that there are many gaps to be filled in this areain either finding new and better fatigue indices or constantlydeveloping new and improved processing techniques

Competing Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

The authors acknowledge and thank the Control and SignalProcessing Group of Universiti Putra Malaysia and Rehabili-tation andAssistive Technology ResearchGroup ofUniversitiTeknikal Malaysia Melaka for their continuous support inthe research This research is fully funded by the Ministry ofHigher Education Malaysia (MOHE)

References

[1] S Gallagher and J R Heberger ldquoExamining the interactionof force and repetition on musculoskeletal disorder risk asystematic literature reviewrdquo Human Factors vol 55 no 1 pp108ndash124 2013

[2] E M Eatough J D Way and C-H Chang ldquoUnderstandingthe link between psychosocial work stressors and work-relatedmusculoskeletal complaintsrdquo Applied Ergonomics vol 43 no 3pp 554ndash563 2012

[3] T RWaters V Putz-Anderson A Garg and L J Fine ldquoRevisedNIOSH equation for the design and evaluation ofmanual liftingtasksrdquo Ergonomics vol 36 no 7 pp 749ndash776 1993

[4] Y Roquelaure C Ha A Leclerc et al ldquoEpidemiologic surveil-lance of upper-extremity musculoskeletal disorders in theworking populationrdquo Arthritis Care and Research vol 55 no5 pp 765ndash778 2006

[5] H Arif M S E Kosnan K Jusoff et al ldquoFinite element analysisof conceptual lumbar spine for different lifting positionrdquoWorldApplied Sciences Journal vol 21 pp 68ndash75 2013

[6] C J Colloca and R N Hinrichs ldquoThe biomechanical and clin-ical significance of the lumbar erector spinae flexion-relaxationphenomenon a review of literaturerdquo Journal of Manipulativeand Physiological Therapeutics vol 28 no 8 pp 623ndash631 2005

[7] N H Kamarudin S A Ahmad M K Hassan R M Yusoffand S Z Dawal ldquoMuscle contraction analysis during liftingtaskrdquo in Proceedings of the 3rd IEEE Conference on BiomedicalEngineering and Sciences (IECBES rsquo14) pp 452ndash457 IEEE KualaLumpur Malaysia December 2014

[8] K R Voge and J B Dingwell ldquoRelative timing of changes inmuscle fatigue and movement coordination during a repetitiveone-hand lifting taskrdquo in Proceedings of the A New Beginningfor Human Health Proceddings of the 25th Annual InternationalConference of the IEEE Engineering in Medicine and BiologySociety (EMBS rsquo03) pp 1807ndash1810 IEEE Cancun MexicoSeptember 2003

[9] TN S T Zawawi A R Abdullah E F Shair I Halim and SMSaleh ldquoEMG signal analysis of fatiguemuscle activity inmanualliftingrdquo Journal of Electrical Systems vol 11 no 3 pp 319ndash3252015

BioMed Research International 9

[10] R A Al-Ashaik M Z Ramadan K S Al-Saleh and T M Kha-laf ldquoEffect of safety shoes type lifting frequency and ambienttemperature on subjectrsquos MAWL and physiological responsesrdquoInternational Journal of Industrial Ergonomics vol 50 pp 43ndash512015

[11] K P Granata and W S Marras ldquoAn EMG-assisted model oftrunk loading during free-dynamic liftingrdquo Journal of Biome-chanics vol 28 no 11 pp 1309ndash1317 1995

[12] S H Roy P Bonato and M Knaflitz ldquoEMG assessment ofback muscle function during cyclical liftingrdquo Journal of Elec-tromyography and Kinesiology vol 8 no 4 pp 233ndash245 1998

[13] R E Seroussi andM H Pope ldquoThe relationship between trunkmuscle electromyography and lifting moments in the sagittaland frontal planesrdquo Journal of Biomechanics vol 20 no 2 pp135ndash146 1987

[14] R B Graham M J Agnew and J M Stevenson ldquoEffectivenessof an on-body lifting aid at reducing low back physical demandsduring an automotive assembly task assessment of EMG res-ponse and user acceptabilityrdquo Applied Ergonomics vol 40 no5 pp 936ndash942 2009

[15] D Gagnon C Lariviere and P Loisel ldquoComparative abilityof EMG optimization and hybrid modelling approaches topredict trunk muscle forces and lumbar spine loading duringdynamic sagittal plane liftingrdquoClinical Biomechanics vol 16 no5 pp 359ndash372 2001

[16] P Dolan and M A Adams ldquoThe relationship between EMGactivity and extensor moment generation in the erector spinaemuscles during bending and lifting activitiesrdquo Journal of Biome-chanics vol 26 no 4-5 pp 513ndash522 1993

[17] P Bonato P Boissy U Della Croce and S H Roy ldquoChanges inthe surface EMG signal and the biomechanics of motion duringa repetitive lifting taskrdquo IEEE Transactions on Neural Systemsand Rehabilitation Engineering vol 10 no 1 pp 38ndash47 2002

[18] J R Potvin ldquoMechanically corrected EMG for the continuousestimation of erector spinae muscle loading during repetitiveliftingrdquo European Journal of Applied Physiology and Occupa-tional Physiology vol 74 no 1-2 pp 119ndash132 1996

[19] I Kingma and J H Van Dieen ldquoLifting over an obstacle effectsof one-handed lifting and hand support on trunk kinematicsand low back loadingrdquo Journal of Biomechanics vol 37 no 2pp 249ndash255 2004

[20] H-J Shin and J-Y Kim ldquoMeasurement of trunk muscle fatigueduring dynamic lifting and lowering as recovery time changesrdquoInternational Journal of Industrial Ergonomics vol 37 no 6 pp545ndash551 2007

[21] J Cholewicki S M McGill and R W Norman ldquoComparisonof muscle forces and joint load from an optimization and EMGassisted lumbar spine model towards development of a hybridapproachrdquo Journal of Biomechanics vol 28 no 3 pp 321ndash3311995

[22] L Straker ldquoEvidence to support using squat semi-squat andstoop techniques to lift low-lying objectsrdquo International Journalof Industrial Ergonomics vol 31 no 3 pp 149ndash160 2003

[23] C K Anderson and D B Chaffin ldquoA biomechanical evaluationof five lifting techniquesrdquo Applied Ergonomics vol 17 no 1 pp2ndash8 1986

[24] R Burgess-Limerick ldquoSquat stoop or something in betweenrdquoInternational Journal of Industrial Ergonomics vol 31 no 3 pp143ndash148 2003

[25] S M Hsiang G E Brogmus and T K Courtney ldquoLow backpain (LBP) and lifting techniquemdasha reviewrdquo International Jour-nal of Industrial Ergonomics vol 19 no 1 pp 59ndash74 1997

[26] A Merlo and I Campanini ldquoTechnical aspects of surface elec-tromyography for cliniciansrdquo The Open Rehabilitation Journalvol 3 no 1 pp 98ndash109 2010

[27] E N Kamavuako J C Rosenvang R Horup W Jensen DFarina and K B Englehart ldquoSurface versus untargeted intra-muscular EMG based classification of simultaneous and dyna-mically changing movementsrdquo IEEE Transactions on NeuralSystems and Rehabilitation Engineering vol 21 no 6 pp 992ndash998 2013

[28] L H Smith and L J Hargrove ldquoComparison of surface andintramuscular EMG pattern recognition for simultaneouswristhand motion classificationrdquo in Proceedings of the 35thAnnual International Conference of the IEEE Engineering inMedicine and Biology Society (EMBC rsquo13) pp 4223ndash4226Osaka Japan July 2013

[29] L J Hargrove K Englehart and B Hudgins ldquoA comparisonof surface and intramuscular myoelectric signal classificationrdquoIEEE Transactions on Biomedical Engineering vol 54 no 5 pp847ndash853 2007

[30] A Fratini M Cesarelli P Bifulco and M Romano ldquoRelevanceof motion artifact in electromyography recordings duringvibration treatmentrdquo Journal of Electromyography and Kinesiol-ogy vol 19 no 4 pp 710ndash718 2009

[31] C J De Luca L Donald Gilmore M Kuznetsov and S HRoy ldquoFiltering the surface EMG signal movement artifact andbaseline noise contaminationrdquo Journal of Biomechanics vol 43no 8 pp 1573ndash1579 2010

[32] H L Butler R Newell C L Hubley-Kozey and J W KozeyldquoThe interpretation of abdominal wall muscle recruitment stra-tegies change when the electrocardiogram (ECG) is removedfrom the electromyogram (EMG)rdquo Journal of Electromyographyand Kinesiology vol 19 no 2 pp e102ndashe113 2009

[33] G Lu J-S Brittain P Holland et al ldquoRemoving ECG noisefrom surface EMG signals using adaptive filteringrdquo Neuro-science Letters vol 462 no 1 pp 14ndash19 2009

[34] N W Willigenburg A Daffertshofer I Kingma and J H vanDieen ldquoRemoving ECG contamination from EMG recordingsa comparison of ICA-based and other filtering proceduresrdquoJournal of Electromyography and Kinesiology vol 22 no 3 pp485ndash493 2012

[35] M I Ibrahimy M R Ahsan and O O Khalifa ldquoDesign andoptimization of Levenberg-Marquardt based neural networkclassifier for EMG signals to identify hand motionsrdquo Measure-ment Science Review vol 13 no 3 pp 142ndash151 2013

[36] S N Sidek and A J Haja Mohideen ldquoMapping of EMG signalto hand grip force at varying wrist anglesrdquo in Proceedings ofthe 2nd IEEE-EMBS Conference on Biomedical Engineering andSciences (IECBES rsquo12) pp 648ndash653 IEEE Langkawi MalaysiaDecember 2012

[37] Y Zhan D Halliday P Jiang X Liu and J Feng ldquoDetectingtime-dependent coherence between non-stationary electro-physiological signalsmdasha combined statistical and time-freq-uency approachrdquo Journal of Neuroscience Methods vol 156 no1-2 pp 322ndash332 2006

[38] E R Ferrara and BWidrow ldquoFetal electrocardiogram enhance-ment by time-sequenced adaptive filteringrdquo IEEE Transactionson Biomedical Engineering vol 29 no 6 pp 458ndash460 1982

[39] R H Shumway ldquoTime-frequency clustering and discriminantanalysisrdquo Statistics amp Probability Letters vol 63 no 3 pp 307ndash314 2003

[40] D Korosec ldquoParametric estimation of the continuous non-stationary spectrum and its dynamics in surface EMG studiesrdquo

10 BioMed Research International

International Journal of Medical Informatics vol 58-59 pp 59ndash69 2000

[41] G Morren S Van Huffel I Helon et al ldquoEffects of non-nutritive sucking on heart rate respiration and oxygenationa model-based signal processing approachrdquo Comparative Bio-chemistry and PhysiologymdashAMolecular and Integrative Physiol-ogy vol 132 no 1 pp 97ndash106 2002

[42] R Latif E Aassif M Laaboubi A Moudden and G MazeldquoDetermination of the cut-off frequency of the anti-symmetriccircumferential waves using the time-varying autoregressiveTVARrdquo in Proceedings of the 3rd International Symposium onCommunications Control and Signal Processing (ISCCSP rsquo08)pp 357ndash361 IEEE March 2008

[43] R Latif E Aassif M Laaboubi and G Maze ldquoDeterminationof the group velocity using the time-varying autoregressiveTVARrdquo in Proceedings of the 9th International Symposium onSignal Processing and Its Applications (ISSPA rsquo07) pp 1ndash4 IEEESharjah United Arab Emirates February 2007

[44] A Al Zaman X Luo M Ferdjallah and A Khamayseh ldquoA newTVARmodeling in cascaded form for nonstationary signalsrdquo inProceedings of the IEEE International Conference on AcousticsSpeech and Signal Processing (ICASSP rsquo06) pp 353ndash356 May2006

[45] A Al Zaman M A Ahad M Ferdjallah and J J Wertsch ldquoAnew approach for muscle fatigue analysis in young adults atdifferent MVC levelsrdquo in Proceedings of the IEEE International48th Midwest Symposium on Circuits and Systems (MWSCASrsquo05) pp 499ndash502 August 2005

[46] Z G Zhang H T Liu S C Chan K D K Luk and Y HuldquoTime-dependent power spectral density estimation of surfaceelectromyography during isometric muscle contraction meth-ods and comparisonsrdquo Journal of Electromyography and Kinesi-ology vol 20 no 1 pp 89ndash101 2010

[47] AAl ZamanM Ferdjallah andAKhamayseh ldquoMuscle fatigueanalysis for healthy adults using TVAR model with instanta-neous frequency estimationrdquo in Proceedings of the 38th South-eastern Symposium on SystemTheory pp 44ndash47 March 2006

[48] K Englehart B Hudgins P A Parker and M StevensonldquoClassification of the myoelectric signal using time-frequencybased representationsrdquoMedical Engineering and Physics vol 21no 6-7 pp 431ndash438 1999

[49] N Selvaraj J Lee and K H Chon ldquoTime-varying methods forcharac-terizing nonstationary dynamics of physiological sys-temsrdquo Methods of Information in Medicine vol 49 no 5 pp435ndash442 2010

[50] D Tkach H Huang and T A Kuiken ldquoStudy of stability oftime-domain features for electromyographic pattern recogni-tionrdquo Journal of NeuroEngineering and Rehabilitation vol 7article 21 13 pages 2010

[51] S Du and M Vuskovic ldquoTemporal vs spectral approach tofeature extraction from prehensile EMG signalsrdquo in Proceedingsof the IEEE International Conference on Information Reuseand Integration (IRI rsquo04) pp 344ndash350 Las Vegas Nev USANovember 2004

[52] R M Enoka and J Duchateau ldquoMuscle fatigue what why andhow it influences muscle functionrdquo Journal of Physiology vol586 no 1 pp 11ndash23 2008

[53] W J Williams and J Jeong ldquoKernel design for reduced interfer-ence distributionsrdquo IEEE Transactions on Signal Processing vol40 no 2 pp 402ndash412 1992

[54] M B I Reaz M S Hussain and F Mohd-Yasin ldquoTechniquesof EMG signal analysis detection processing classification and

applicationsrdquo Biological Procedures Online vol 8 no 1 pp 11ndash35 2006

[55] R A Malinzak S M Colby D T Kirkendall B Yu and W EGarrett ldquoA comparison of knee joint motion patterns betweenmen and women in selected athletic tasksrdquo Clinical Biomechan-ics vol 16 no 5 pp 438ndash445 2001

[56] T I Arabadzhiev V G Dimitrov N A Dimitrova and G VDimitrov ldquoInterpretation of EMG integral or RMS and esti-mates of lsquoneuromuscular efficiencyrsquo can bemisleading in fatigu-ing contractionrdquo Journal of Electromyography and Kinesiologyvol 20 no 2 pp 223ndash232 2010

[57] C Suetta P Aagaard A Rosted et al ldquoTraining-inducedchanges in muscle CSA muscle strength EMG and rate offorce development in elderly subjects after long-term unilateraldisuserdquo Journal of Applied Physiology vol 97 no 5 pp 1954ndash1961 2004

[58] M A Oskoei andHHu ldquoSupport vectormachine-based classi-fication scheme for myoelectric control applied to upper limbrdquoIEEE Transactions on Biomedical Engineering vol 55 no 8 pp1956ndash1965 2008

[59] J J Villarejo A Frizera T F Bastos and J F Sarmiento ldquoPatternrecognition of hand movements with low density sEMG forprosthesis control purposesrdquo in Proceedings of the IEEE 13thInternational Conference onRehabilitation Robotics (ICORR rsquo13)Seattle Wash USA June 2013

[60] A Phinyomark C Limsakul and P Phukpattaranont ldquoA novelfeature extraction for robust EMG pattern recognitionrdquo Journalof Computing vol 1 no 1 pp 71ndash80 2009

[61] M Zardoshti-Kermani B C Wheeler K Badie and R MHashemi ldquoEMG Feature Evaluation for Movement Control ofUpper Extremity Prosthesesrdquo IEEE Transactions on Rehabilita-tion Engineering vol 3 no 4 pp 324ndash333 1995

[62] D Tkach H Huang and T A Kuiken ldquoStudy of stability oftime-domain features for electromyographic pattern recogni-tionrdquo Journal of NeuroEngineering and Rehabilitation vol 7 no1 article no 21 2010

[63] K Kiguchi S Kariya K Watanabe K Izumi and T FukudaldquoAn exoskeletal robot for human elbowmotion supportmdashsensorfusion adaptation and controlrdquo IEEE Transactions on SystemsMan and Cybernetics Part B Cybernetics vol 31 no 3 pp 353ndash361 2001

[64] F Al Omari J Hui C Mei and G Liu ldquoPattern recognition ofeight hand motions using feature extraction of forearm EMGsignalrdquo Proceedings of the National Academy of Sciences IndiaSection AmdashPhysical Sciences vol 84 no 3 pp 473ndash480 2014

[65] A R Siddiqi S N Sidek andAKhorshidtalab ldquoSignal process-ing of EMG signal for continuous thumb-angle estimationrdquo inProceedings of the 41st Annual Conference of the IEEE IndustrialElectronics Society (IECON rsquo15) pp 374ndash379 Yokohama JapanNovember 2015

[66] A Kilbom G M Hagg and C Kall ldquoOne-handed loadcarryingmdashcardiovascular muscular and subjective indices ofendurance and fatiguerdquo European Journal of Applied Physiologyand Occupational Physiology vol 65 no 1 pp 52ndash58 1992

[67] F AlOmari and G Liu ldquoAnalysis of extracted forearm sEMGsignal using LDA QDA K-NN classification algorithmsrdquoOpenAutomation and Control Systems Journal vol 6 no 1 pp 108ndash116 2014

[68] A Phinyomark G Chujit P Phukpattaranont C Limsakuland H Hu ldquoA preliminary study assessing time-domain EMGfeatures of classifying exercises in preventing falls in the elderlyrdquoin Proceedings of the 9th International Conference on Electri-cal EngineeringElectronics Computer Telecommunications and

BioMed Research International 11

Information Technology (ECTI-CON rsquo12) pp 1ndash4 IEEE Phetch-aburi Thailand May 2012

[69] D R Rogers and D T MacIsaac ldquoEMG-based muscle fatigueassessment during dynamic contractions using principal com-ponent analysisrdquo Journal of Electromyography and Kinesiologyvol 21 no 5 pp 811ndash818 2011

[70] K Buchenrieder ldquoDimensionality reduction for the controlof powered upper limb prosthesesrdquo in Proceedings of the 14thAnnual IEEE International Conference and Workshops on theEngineering of Computer-Based Systems (ECBS rsquo07) pp 327ndash333IEEE Tucson Ariz USA March 2007

[71] G Venugopal M Navaneethakrishna and S RamakrishnanldquoExtraction and analysis of multiple time window featuresassociated with muscle fatigue conditions using sEMG signalsrdquoExpert Systems with Applications vol 41 no 6 pp 2652ndash26592014

[72] A Phinyomark C Limsakul and P Phukpattaranont ldquoEMGfeature extraction for tolerance of white Gaussian noiserdquo inProceedings of the International Workshop and Symposium onScience and Technology (I-SEEC rsquo08) pp 178ndash183 December2008

[73] M S Al-Quraishi A J Ishak S A Ahmad and M K HasanldquoImpact of feature extraction techniques on classification accu-racy for EMG based ankle joint movementsrdquo in Proceedings ofthe 10th Asian Control Conference (ASCC rsquo15) pp 1ndash5 SabahMalaysia June 2015

[74] G-C Chang W-J Kang L Jer-Junn et al ldquoReal-time imple-mentation of electromyogram pattern recognition as a controlcommand of man-machine interfacerdquoMedical Engineering andPhysics vol 18 no 7 pp 529ndash537 1996

[75] A Georgakis L K Stergioulas and G Giakas ldquoFatigue analysisof the surface EMG signal in isometric constant force con-tractions using the averaged instantaneous frequencyrdquo IEEETransactions on Biomedical Engineering vol 50 no 2 pp 262ndash265 2003

[76] R Merletti and L R Lo Conte ldquoSurface EMG signal processingduring isometric contractionsrdquo Journal of Electromyography andKinesiology vol 7 no 4 pp 241ndash250 1997

[77] C J De Luca ldquoUse of the surface EMG signal for performanceevaluation of backmusclesrdquoMuscle and Nerve vol 16 no 2 pp210ndash216 1993

[78] F Khanam and M Ahmad ldquoFrequency based EMG powerspectrum analysis of Salat associated muscle contractionrdquo inProceedings of the 1st International Conference on Electricaland Electronic Engineering (ICEEE rsquo15) Rajshahi BangladeshNovember 2015

[79] M M Altaf B M Elbagoury F Alraddady and M RoushdyldquoExtended case-based behavior control for multi-humanoidrobotsrdquo International Journal of Humanoid Robotics vol 13 no2 2016

[80] A Phinyomark A Nuidod P Phukpattaranont and C Lim-sakul ldquoFeature extraction and reduction of wavelet transformcoefficients for EMG pattern classificationrdquo Elektronika ir Elek-trotechnika vol 122 no 6 pp 27ndash32 2012

[81] M Gonzalez-Izal A Malanda I Navarro-Amezqueta et alldquoEMG spectral indices and muscle power fatigue duringdynamic contractionsrdquo Journal of Electromyography and Kine-siology vol 20 no 2 pp 233ndash240 2010

[82] G V Dimitrov T I Arabadzhiev K N Mileva J L Bowtell NCrichton andNADimitrova ldquoMuscle fatigue during dynamiccontractions assessed by new spectral indicesrdquo Medicine andScience in Sports and Exercise vol 38 no 11 pp 1971ndash1979 2006

[83] MVitor-Costa H Bortolotti T V Camata et al ldquoEMG spectralanalysis of incremental exercise in cyclists and non-cyclistsusing fourier and wavelet transformsrdquo Revista Brasileira deCineantropometria e Desempenho Humano vol 14 no 6 pp660ndash670 2012

[84] M Wacker and H Witte ldquoTime-frequency techniques in bio-medical signal analysis a tutorial review of similarities anddifferencesrdquoMethods of Information in Medicine vol 52 no 4pp 279ndash296 2013

[85] S Karlsson J Yu and M Akay ldquoTime-frequency analysis ofmyoelectric signals during dynamic contractions a compara-tive studyrdquo IEEE Transactions on Biomedical Engineering vol47 no 2 pp 228ndash238 2000

[86] A L Martinez-Herrera L M Ledesma-Carrillo M Lopez-Ramirez S Salazar-Colores E Cabal-Yepez and A Garcia-Perez ldquoGabor and theWigner-Ville transforms for broken rotorbars detection in induction motorsrdquo in Proceedings of the 24thInternational Conference on Electronics Communications andComputers (CONIELECOMP rsquo14) pp 83ndash87 IEEE CholulaMexico February 2014

[87] D MacIsaac P A Parker and R N Scott ldquoThe short-timeFourier transform and muscle fatigue assessment in dynamiccontractionsrdquo Journal of Electromyography and Kinesiology vol11 no 6 pp 439ndash449 2001

[88] E F Shair A R Abdullah T N S Tengku Zawawi S AAhmad and SMohamad Saleh ldquoAuto-segmentation analysis ofEMG signal for lifting muscle contraction activitiesrdquo Journal ofTelecommunication Electronic and Computer Engineering vol8 no 7 pp 17ndash22 2016

[89] MM Jankovic andD B Popovic ldquoAnEMGsystem for studyingmotor control strategies and fatiguerdquo in Proceedings of the 10thSymposium on Neural Network Applications in Electrical Engi-neering (NEUREL-rsquo10) pp 15ndash18 Belgrade Serbia September2010

[90] T N S T Zawawi A R Abdullah I Halim E F Shair andS M Salleh ldquoApplication of spectrogram in analysing elec-tromyography (EMG) signals of manual liftingrdquo ARPN Journalof Engineering andApplied Sciences vol 11 no 6 pp 3603ndash36092016

[91] M Yochum T Bakir R Lepers and S Binczak ldquoEstimationof muscular fatigue under electromyostimulation using CWTrdquoIEEE Transactions on Biomedical Engineering vol 59 no 12 pp3372ndash3378 2012

[92] J L Dantas T V Camata M A O C Brunetto A C MoraesT Abrao and L R Altimari ldquoFourier and Wavelet spectralanalysis of EMG signals in isometric and dynamic maximaleffort exerciserdquo in Proceedings of the 32nd Annual InternationalConference of the IEEEEngineering inMedicine andBiology Soci-ety (EMBC rsquo10) pp 5979ndash5982 IEEE Buenos Aires ArgentinaSeptember 2010

[93] L Ai J Nan and J Shen ldquoApplication of S-transform to drivingfatigue in EEG analysisrdquo in Proceedings of the IEEE InternationalConference on Intelligent Computing and Intelligent Systems(ICIS rsquo10) pp 814ndash817 IEEE Guilin China October 2010

[94] R Upadhyay P K Padhy and P K Kankar ldquoEEG artifactremoval and noise suppression by discrete orthonormal S-transform denoisingrdquo Computers and Electrical Engineeringvol 53 pp 125ndash142 2016

[95] C Dora and P K Biswal ldquoRobust ECG artifact removalfrom EEG using continuous wavelet transformation and linearregressionrdquo in 2016 International Conference on Signal Process-ing and Communications (SPCOM) pp 1ndash5 Bangalore IndiaJune 2016

12 BioMed Research International

[96] S Assous and B Boashash ldquoEvaluation of themodified S-trans-form for time-frequency synchrony analysis and source locali-sationrdquo Eurasip Journal on Advances in Signal Processing vol2012 article 49 2012

[97] A L Ricamato R G Absher M T Moffroid and J P Tra-nowski ldquoA time-frequency approach to evaluate electromyo-graphic recordingsrdquo in Proceedings of the 5th Annual IEEESymposium on Computer-Based Medical Systems pp 520ndash527IEEE Durham NC USA 1992

[98] M Khezri and M Jahed ldquoAn inventive quadratic time-freq-uency scheme based on Wigner-Ville distribution for classifi-cation of sEMG signalsrdquo in Proceedings of the 6th InternationalSpecial Topic Conference on Information TechnologyApplicationsin Biomedicine pp 261ndash264 IEEE Tokyo Japan November2007

[99] A Subasi and M K Kiymik ldquoMuscle fatigue detection in EMGusing time-frequency methods ICA and neural networksrdquoJournal of Medical Systems vol 34 no 4 pp 777ndash785 2010

[100] W Martin and P Flandrin ldquoWigner-Ville Spectral Analysisof Nonstationary Processesrdquo IEEE Transactions on AcousticsSpeech and Signal Processing vol 33 no 6 pp 1461ndash1470 1985

[101] H-I Choi and W J Williams ldquoImproved time-frequency rep-resentation of multicomponent signals using exponential ker-nelsrdquo IEEE Transactions on Acoustics Speech and Signal Pro-cessing vol 37 no 6 pp 862ndash871 1989

[102] G R Pereira L F de Oliveira and J Nadal ldquoReducing crossterms effects in the Choi-Williams transform of mioelectricsignalsrdquo Computer Methods and Programs in Biomedicine vol111 no 3 pp 685ndash692 2013

[103] M Alemu D K Kumar and A Bradley ldquoTime-frequency anal-ysis of SEMGmdashwith special consideration to the interelectrodespacingrdquo IEEE Transactions on Neural Systems and Rehabilita-tion Engineering vol 11 no 4 pp 341ndash345 2003

[104] P A Karthick G Venugopal and S Ramakrishnan ldquoAnalysis ofsurface EMG signals under fatigue and non-fatigue conditionsusing B-distribution based quadratic time frequency distribu-tionrdquo Journal of Mechanics in Medicine and Biology vol 15 no2 Article ID 1540028 2015

[105] V Sucic B Barkat and B Boashash ldquoPerformance evaluationof the B-distributionrdquo in Proceedings of the 5th InternationalSymposium on Signal Processing and Its Applications (ISSPA rsquo99)pp 267ndash270 Queensland Australia August 1999

[106] PAKarthick and S Ramakrishnan ldquoSurface electromyographybasedmuscle fatigue progression analysis usingmodified B dis-tribution time-frequency featuresrdquo Biomedical Signal Processingand Control vol 26 pp 42ndash51 2016

[107] B Boashash and V Sucic ldquoResolution measure criteria for theobjective assessment of the performance of quadratic time-frequency distributionsrdquo IEEE Transactions on Signal Process-ing vol 51 no 5 pp 1253ndash1263 2003

[108] S Dong G Azemi and B Boashash ldquoImproved characteriza-tion of HRV signals based on instantaneous frequency featuresestimated from quadratic time-frequency distributions withdata-adapted kernelsrdquoBiomedical Signal Processing andControlvol 10 no 1 pp 153ndash165 2014

[109] B Boashash and T Ben-Jabeur ldquoDesign of a high-resolutionseparable-kernel quadratic TFD for improving newborn healthoutcomes using fetal movement detectionrdquo in Proceedings ofthe 11th International Conference on Information Science SignalProcessing and their Applications (ISSPA rsquo12) pp 354ndash359Quebec Canada July 2012

Submit your manuscripts athttpswwwhindawicom

Stem CellsInternational

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MEDIATORSINFLAMMATION

of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Behavioural Neurology

EndocrinologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Disease Markers

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

BioMed Research International

OncologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Oxidative Medicine and Cellular Longevity

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

PPAR Research

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Immunology ResearchHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

ObesityJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Computational and Mathematical Methods in Medicine

OphthalmologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Diabetes ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Research and TreatmentAIDS

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Gastroenterology Research and Practice

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Parkinsonrsquos Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttpwwwhindawicom

Page 2: EMG Processing Based Measures of Fatigue Assessment during …downloads.hindawi.com/journals/bmri/2017/3937254.pdf · 2019-07-30 · BioMedResearchInternational 3 5. Electromyography

2 BioMed Research International

Table 1 Common muscles investigated during manual lifting

Reference Muscle areasKamarudin et al [7] Biceps brachii and triceps brachii

Voge and Dingwell [8] Biceps middeltoid midtrapeziusand postdeltoid

Zawawi et al [9] Biceps brachii

Al-Ashaik et al [10] Biceps brachii erector spinaeanterior deltoid and trapezius

Granata and Marras [11]Erector spinae external abdominal

obliques internal abdominalobliques rectus abdominis and

latissimus dorsi

Roy et al [12] Longissimus thoracis iliocostalislumborum and multifidus

Seroussi and Pope [13] Erector spinae and external oblique

Graham et al [14]Lumbar erector spinae thoracic

erector spinae and rectusabdominis

Gagnon et al [15]Thoracic erector spinae lumbarerector spinae internal oblique

external oblique rectus abdominisand latissimus dorsi

Dolan and Adams [16] Erector spinae

Bonato et al [17] Iliocostalis lumborum longissimusthoracis and multifidus

Potvin [18] Thoracic erector spinae and lumbarerector spinae

Kingma and Van Dieen [19]External oblique internal oblique

rectus abdominis iliocostalislumborum longissimus thoracis

and pars thoracis

Shin and Kim [20]Erector spinae rectus abdominislatissimus dorsi internal oblique

and external oblique

Cholewicki et al [21]Lumbar erector spinae thoracicerector spinae latissimus dorsi

internal oblique external obliqueand rectus abdominis

trunk muscles (predominately by erector spinae muscles)[6] This movement causes the erector spinae muscles tomake incredible expansive strengths for adjusting the bodyEven though there are numerous assistive gadgets such as lifttrucks trolleys and derricks that can help reduce loads on thelumbar spinae however in many cases these instruments arecostly and inconvenient

Lifting strategies are an important measure during liftingand because physically correct lifting is significant theirimpact onmusculoskeletal well-being is broadly talked about[22] The impact of lifting techniques including positionwidth (characterized as the separation between the feet inthe average sidelong heading with the sagittal symmetry ofposition) has been considered recommending the utilizationof a wide position to lessening the burden on the spine [23]Lifting procedures (stoop or squat and semisquat) can alsohave a significant effect on spine stacking and steadinessduring lifting Past studies have researched lifting methods in

biomechanical terms to give some mediation methodologiesand recognize the correct lifting procedures [24 25] Squatlifting seems to have less lumbar shear stretch and place lessweight on noncontractile connective tissues Stoop lifting hasall the earmarks of being more regular and less exhaustingSemisquat lifting might be a decent trade-off lift despitethe fact that early confirmation recommends perhaps higherlumbar minutes [22]

3 Muscle Fatigue Assessment

Muscle fatigue is characterized as the long lasting deterio-ration of the performance of the human operator to createforce [24 25] A personrsquos muscle capabilities may differ fromone another but the characteristics should have some resem-blance Past researchers have outlined several features ofmuscle fatigue such as an increment in amplitude and thetransition from high frequency range to low frequency [26]

To assess this muscle fatigue characteristics various tech-niques are available depending on the usage At thismomenteven though the measurement of muscle fatigue by use ofinvasive means such as blood test (blood lactate level andblood oxygen level) andmuscle biopsies (pHofmuscle) offershigher accuracy compared to noninvasive techniques it isunsuitable for some applications such as sports ergonomicsand occupational therapy [27] The use of noninvasiveapproach to acquire muscle signals such as electromyog-raphy (EMG) sonomyography (SMG) mechanomyography(MMG) and near infrared spectroscopy (NIRS) for researchpurposes is considered to bemore convenient and easy to use[28] Every technique has its own advantages and disadvan-tages based on its applications For the purpose of assessingmuscle physiology during manual lifting EMG is the mostcommonly used tool and has been well reported by manyresearchers [29]

4 Electromyography Signal

EMG is the measure of electrical potential present on theskin in consequence to a muscle contraction that representsthe neuromuscular activities [26] It can be measured by twomethods (1) applying electrodes to the skin surface (non-invasive) or (2) intramuscular (invasive) within the muscleEven though intramuscular EMG (imEMG) provides addi-tional benefits to overcome the drawbacks of surface EMG(sEMG) such as maintaining robust electrode contact withthe skin and the capacity to record from profound muscleswith little EMG crosstalk [27] sEMG and imEMG havebeen proven to have equal classification performance for dataof comparative nature (unmodulated) However the perfor-mance of imEMG diminished compared to surface whentested on modulated data [28 29]

The greater selectivity of imEMG with respect to sEMGis due to wires exposed only at the tip but then again may bea hindrance since the signal may provide local rather thanglobal information where imEMG recordings depend on therecruitment ofmotor units it may be the case that insufficientinformation was captured at low amplitudefrequency [27]

BioMed Research International 3

5 Electromyography Preprocessing

When dealing with EMG signal noise removal is a veryimportant factor to be taken into consideration This is dueto the sensitive nature of EMG signal itself that different typesof noises or artifacts unavoidably contaminate itThese noisesare caused by various sources originates from either the skin-electrode interface hardware that intensifies the signal orother external sourcesTherefore before proceeding with theanalysis and classification of EMG signal a preprocessing isusually introduced to remove these noises Some of the noisesthat may affect the signals are inherent noise in electronicsequipment ambient noise motion artifact ECG artifactsinherent instability of signal and crosstalk However thisreview only covers three types of noises motion artifact ECGnoise and device noise

51 Motion Artifact Cutting edge technology is considerablyinvulnerable to some of the EMG noises but not to themotion artifact as it has frequency spectra that are affectingthe low frequency part of EMG signal This motion artifactoriginates at the electrode-skin interface where there existmuscle movements underneath the skin and force impulsethat goes through the muscle and the skin underlying theelectrodes causing movements [30] These movements willresult in a time-varying voltage across the two electrodes

To remove the low frequency noise caused by the motionartifact a study has been conducted to ascertain a reasonablevalue for the high-pass corner frequency filter The result ofthe study recommends the use of a Butterworth filter with a20Hz corner frequency and a 12 dBoct slope [31]

52 ECG Noise When EMG signal is measure close to theheart ECG bursts may pollute the EMG recording whichcan affect the analysis and result in misinterpretations [32]This is a natural relic that is often unavoidable It can bediminished by proper skin preparation and adjusted positionof the ground electrode To date other than the conventionalECG removal procedures such as the use of high past filteringseveral ECG noise removal techniques have been introducedthat can clean these ECG bursts but still maintaining theregular EMG characteristics

In 2009 Lu et al came out with a study to evaluate theperformance of recursive-least-square adaptive filter in theremoval of ECG interference from EMG signal [33] Theresult of the study shows that due to fast convergence ofthe recursive-least-square algorithm the filter is reliable toeffectively remove ECG noise

Later in 2012 independent component analysis (ICA)wasproposed to remove the ECG contamination The filteringeffects are measured in terms of root mean square relativeerrors and linear envelopes correlation of uncontaminatedand contaminated EMGThis technique excellently producesgood results when EMG and ECG are statistically indepen-dent [34]

53 Device Noise Another type of noise that usually affectsEMG signal is the noise generated inside the EMG device

itself that is commonly referred to as inherent noise inelectronics equipment [35] This noise cannot be completelyeliminated as all electronic equipment generates noise [36]However the severity of the noise can be reduced using high-quality electronic components

6 Electromyography Processing

Biosignal processing is a critical part in biomedical engineer-ing in order to classify the frequency content of a signalThesesignals include EMG electrocardiography (ECG) and elec-troencephalography (EEG)Most of these biosignals are non-stationary signals due to its time-varying characteristics [37ndash39] The problem of this nonstationary signal is the processof assuming it to be stationary over short-time intervalswhere stationary analysis techniques are used However thisassumption is not always suitable and further methods fornonstationary processes are needed

Past researchers have developed several techniques toprocess the nonstationary signal specifically the EMG signaland these include parametric and nonparametric approaches[40 41] The differences between these two approaches aremainly due to the parameters involved Since most of the sig-nals encounters are unknown signals without known param-eters the development of nonparametric approach whichinvolves a variety of transforms is actively researched com-pared to the parametric approach that requires the modelingof the nonstationary process Figure 1 provides an overviewof the EMG signal processing methods that are reviewed inthe manuscript

61 Parametric Approach In the field of EMG processingone type of parametric approach that has been explored isbased on the time-varying linear predictive models Thisincludes time-varying autoregressive (TVAR) model whereits parameters vary with time The time-varying spectrum inTVAR is estimated from the time-varying model parametersand the nonstationary signalrsquos instantaneous frequency (IF)can be extracted [42] IF represents the spectral peak locationof the EMG signal as it varies with time [43] Even thoughthe poles and zeros of the estimated model can be estimateddirectly from the EMG signal it is not guaranteed that thetime-varying poles remain inside the unit circle on the 119911-plane thus making TVAR model temporarily unstable [44]The constraints can be easily imposed by factorizing thedenominator of TVAR in direct form and formulating it incascaded form [45] Computation of the mean frequency(MNF) is taken as the time average of the IF estimation basedon the cascaded model in each interval [46]

Research conducted by Al Zaman et al has comparedTVAR with short-time Fourier transform (STFT) and con-cludes that even though both TVAR and STFThave producedsimilar decreasing muscle fatigue patterns from the MNFTVAR has achieved a better performance with the slope andinterception of the linear regression line is closer [47]

Another parametric approach that has been used is thetime-varying autoregressive moving average (TVARMA)Studies have been made to compare the performance of

4 BioMed Research International

Electromyography processing

Parametric approach

Time distribution Frequency distribution

Linear

transform (STFT)

Spectrogram

Wavelet transform (WT)

(TVAR) average (TVARMA)

Bilinear

(WVD)

(CWD)

Time-varying autoregressive Time-varying autoregressive moving

Time-frequency distribution

Nonparametric approach

B-distributionWigner-Ville distribution

Choi-William distributionS-transform

Short-time Fourier

Figure 1 Overall view of the electromyography signal processing methods that are reviewed in this manuscript

TVAR and TVARMA in analyzing EMG signals and it isconcluded that AR models are more favorable than ARMAbecause of the following reasons (a) estimation of ARmodelsis moderately basic since they are represented by an arrange-ment of direct mathematical statements (b) AR modelsare predominant in displaying the low recurrence sEMGsegment and (c) fluctuation of the AR model estimation onshorter fragments is lower than for ARMAmodels [40]

62 Nonparametric Approach Another EMG processingmethod is the nonparametric approaches which includeSTFT spectrogram andwavelet transformThese approachesappear to be more favorable to researchers since the fatiguesindices can be obtained directly from the EMG signal withoutknowing its parametersThe analysis of the EMG signal basedon the nonparametric approaches can be further dividedinto three types of distribution techniques time domainfrequency domain and time-frequency domain [48ndash50]

621 Time Distribution (TD) Extraction of time domainfatigue indices is very simple and easy and involves lowcomputational complexity compared to the other two tech-niques since it does not involve any transformation [51] Itcan be directly obtained from the time representation of theraw EMG signal just by doing some simple mathematicalstatisticsThe idea of fatigue is approximately connected withthe amount and rate of change of some variables that reflectmuscle alterations amid sustained contractions [52]

Listed in Table 2 are several fatigue indices in the timedomain that have been explored by previous researchers

Even though the analysis based on time domain is alreadyestablished and arewidely used since the last decade howeverit is considered less accurate since their calculations are basedon EMG signal amplitude values where there aremuch inter-ference acquired through the recording especially for fea-tures that are extracted from energy property [53] Anothermajor disadvantage of this time domain features that affectsthe accuracy comes from the nonstationary property of theEMG signal where the statistical properties changes overtime but time domain features assume the data as stationarysignal [54]

Nowadays researches involving time distribution onlyfocus on finding new fatigue indices with higher robustnesswhile most of other researchers nowadays opt for time-frequency analysis which is considered to be more accurate

622 FrequencyDistribution (FD) Inmany studies of fatiguemuscle frequency domain features are usually used to extractinformation from a signal To date there are more than20 fatigue indices in the frequency domain where two ofthe widely used parameters are the mean frequency (MNF)and median frequency (MDF) [75] Other fatigue indices arelisted in Table 3 Several techniques have been used to extractinformation fromEMG signals such as fast Fourier transform(FFT) power spectral density (PSD) and parametric meth-ods (ie AR model)

The EMG signal should undergo Fourier transform inorder to represent the signal in frequency domain The

BioMed Research International 5

Table 2 Fatigue indices in time domain

Reference Fatigue indicesMalinzak et al [55] Integrated EMG (IEMG)Arabadzhiev et al [56] Root mean square (RMS)Suetta et al [57] Mean absolute value (MAV)

Oskoei and Hu [58] Modified mean absolute value type 1(MAV1)

Villarejo et al [59] Modified mean absolute value type2 (MAV2)

Phinyomark et al [60] Simple integral square (SSI)Zardoshti-Kermani et al [61] Variance of EMG (VAR)Tkach et al [62] v-order (V)Zardoshti-Kermani et al [61] Log detector (LOG)Kiguchi et al [63] Waveform length (WL)Al Omari et al [64] Average amplitude change (AAC)

Siddiqi et al [65] Difference absolute standarddeviation value (DASDV)

Kilbom et al [66] Zero crossing (ZC)AlOmari and Liu [67] Myopulse percentage rate (MYOP)Phinyomark et al [68] Willison amplitude (WAMP)Rogers et al [69] Slope sign change (SSC)

Buchenrieder [70] Mean absolute value slope(MAVSLP)

Venugopal et al [71] Multiple hamming windows(MHW)

Du and Vuskovic [51] Multiple trapezoidal windows(MTW)

Phinyomark et al [72] Histogram of EMG (HIST)Al-Quraishi et al [73] Autoregressive coefficient (AR)Chang et al [74] Cepstral coefficient (CC)

definition of Fourier transform in the form of power spec-trum is shown in

119878119909 (119891) = 111987910038161003816100381610038161003816100381610038161003816intinfin

minusinfin119909 (119905) sdot 119890minus2120587119891119905119889119905

100381610038161003816100381610038161003816100381610038162

(1)

where 119909(119905) is the signal in time domainHowever the fluctuations of the EMG frequency compo-

nent due to the adjustments in the muscle force length andcontraction speed throughout time have caused challenges inthe usage of FFT and other traditional processing methodsThe fundamental confinement of a FFT is that it cannot givesimultaneous time and frequency localization [83]Thereforeinvestigation of the EMG signal in dynamic contraction(ie repetitive lifting) utilizing these methods may not besuccessful since it requires the signal to be stationary

623 Time-Frequency Distribution (TFD) Time-frequencyrepresentation (TFR) of a signal maps a one-dimensionalsignal of time into a two-dimensional of time and frequencyIn analyzing modifying and synthesizing nonstationarysignals TFR arewidely used since both representation of timeand frequency are taken into account thus leading to higheraccuracy [84]

Table 3 Fatigue indices in frequency domain

Reference Fatigue indicesMerletti and Lo Conte [76] Mean frequency (MNF)De Luca [77] Median frequency (MDF)Khanam and Ahmad [78] Peak frequency (PKF)Khanam and Ahmad [78] Mean power (MNP)Khanam and Ahmad [78] Total power (TTP)

Altaf et al [79] The 1st 2nd and 3rd spectralmoments (SM1 SM2 and SM3)

Phinyomark et al [80] Power spectrum ratio (PSR)

Gonzalez-Izal et al [81] Instantaneous frequency variance(119865var)

Georgakis et al [75] Averaged instantaneous frequency(AIF)

Dimitrov et al [82] Dimitrovrsquos spectral fatigue index(FInsm5)

In general TFDs consist of linear TFDs and bilinear(quadratic) TFDs The most basic form of TFD technique isthe short-time Fourier transform (STFT) STFT is one of thelinear TFDs that have beenused to analyze EMGsignals alongwith spectrogram wavelet transform S-transform and soforth Examples of the bilinear TFDs include theWigner-Villedistribution (WVD) and Choi-William distribution (CWD)

Linear TFD

Short-Time Fourier Transform (STFT) To overcome thedisadvantages of time distribution and frequency distributionand satisfy the stationary condition it is common to separatelong haul signals into blocks of narrow fragments [85] Thisought to be sufficiently narrow to be viewed as stationary andtake the FT of every segment Every FT gives the spectralinformation of a different time-slice of the signal givingsimultaneous time and frequency estimation

This was proposed by Gabor who had developed theSTFT the extended version of FT [86]

STFT119909 (119905 119908) = intinfin

minusinfin119909 (120591)119908 (120591 minus 119905) 119890minus2120587119891120591119889120591 (2)

where 119909(120591) is the EMG signal 119908(120591 minus 119905) is the observationwindow and 119905 is the variable that slides the window over thesignal 119909(120591)

The choice of the window function in STFT is criticalin order to get accurate results The shape of the windowcan be either rectangular Gaussian or elliptic depending onthe shape of the signal For sampled STFT using a Gaussianwindow it is often called Gabor transform Even though thewindow should be narrow enough to ensure that the portionof the signal that falls within the window is stationary it mustnot be too narrow since it will lead to bad localization in thefrequency domain For infinitely long window 119908(119905) = 1 willeventually cause STFT to turn into FT providing excellentfrequency localization but no time localization In contrastto infinitely short window 119908(119905) = 120575 will result in the time

6 BioMed Research International

signal (with a phase factor) which will provide excellent timelocalization but no frequency localization

Research byMacIsaac et al usesMNFas fatigue index andhas drawn three important conclusions First the nonstation-arities in EMGdo not seem to affect theMNF values obtainedfrom Fourier transform Secondly the STFT method tomeasure MNF has successfully midpoints out the impacts ofnonstationarities in dynamic compressions Third the STFTis capable of identifying a descending pattern with fatigue indynamic contraction as long as the range of motion remainsconsistent across sequential time interval [87]

Since the STFT is basic in idea and execution and isequipped for distinguishing a pattern in muscle fatigue itis an undeniable choice for applications that require simpleanalysis with acceptable results

Spectrogram The squared magnitude of the STFT is calledspectrogram It can be expressed as

119878119909 (119905 119891) =10038161003816100381610038161003816100381610038161003816intinfin

minusinfin119909 (120591)119908 (120591 minus 119905) 119890minus2120587119891120591

100381610038161003816100381610038161003816100381610038162

119889120591 (3)

where 119909(120591) is the EMG signal 119908(120591 minus 119905) is the observationwindow and 119905 is the variable that slides the window over thesignal 119909(120591)

Spectrogram can be used to obtain the power distributionand energy distribution of the signal along the frequencydirection at a given time In EMG processing the instan-taneous energy is useful to separate muscle activation fromthe baseline which is called the segmentation process Seg-mentation is important in reducing the high computationalcomplexity of TFDs The muscle activation segmentationprocess by using spectrogram has been proposed by Shair etal and resulted in a mean absolute percentage error (MAPE)of just 1404 [88]

For both STFT and spectrogram there is a compromisebetween the time-based and frequency-based perspective ofa signal Both time and frequency are represented in limitedprecision where precision is controlled by the span of thewindow and the size of the window chosen will be the samefor all frequencies

In 2010 Jankovic and Popovic have presented in theirpaper the use of spectrogram in the analysis ofmuscle fatigueThey highlighted that even though the traditional MDFtechnique provides good sensitivity for fatigue indicationhowever there is a significant slope error particularly for lowactivity of coinitiated muscles They proposed a hybrid ofalternative methods (spectrogram and scalogram) that cansuccessfully provide complete information but at the sametime produce less error prediction for low-level coinitiatedmuscles [89] As a recommendation for improvement theyalso proposed further investigation on the robustness of thismethod in dynamic exercise contractions

Later in 2016 Zawawi et al came out with a detailedanalysis of EMG signals by using spectrogram for manuallifting application By taking the average instantaneous RMSvoltage (119881rms(119905)) as fatigue index she concludes that asthe lifting height is increased the average 119881rms(119905) is alsoincreased However as the average 119881rms(119905) decreased thenumber of lifting repetitions is increased [90]

Wavelet Transform (WT) A wavelet is a waveform of effec-tively limited duration that has an average value of zero Someof its properties are short-time localized waves with zerointegral value the possibility of time shifting and flexibilityWavelet analysis produces a time-scale view of the signalTheresult of a continuous wavelet transform (CWT) is waveletcoefficients Multiplying each coefficient with the appropri-ately scaled and shifted wavelet yields the constituent waveletof the original signal Equation (4) shows the formula forCWT

CWT119909 = intinfin

minusinfin119909 (120591) 1radic|119886|Ψ (

120591 minus 119905119886 ) 119889120591 (4)

where 119905 is the translation 119886 is the scale parameter and Ψ isthe mother wavelet

Yochum et al have proposed a new fatigue index basedon the continuous wavelet transform (CWT) named 119868CWTand have compared it to other fatigue indices from literaturein terms of their sensitivity to noiseThe results show that thisnew fatigue index quantifies the EMG signal elongation dur-ing a contraction and thus makes it a suitable fatigue index119868CWT is less subjected to noise and less truncation dependentcompared to other fatigue indices based on the frequency likeMNF and MDF [91]

A comparison has been made by Dantas et al betweenSTFT and CWT in evaluating muscle fatigue in isometricand dynamic contractions The after effects of this studyexhibit that CWT and STFT analysis give comparable fatigueestimates (slope of MDF) in isometric and dynamic contrac-tions However the aftereffects of CWT for both contractionsindicate less variability (higher accuracy) in EMG signalanalysis contrasted with STFT [92]

S-Transform Another TFD technique exists is the S-transform This technique has been widely used in diverseareas of telecommunication power quality geophysics andbiomedicine due to its obvious advantages in processingnonstationary signals [93] In the area of EEG and ECGdespite the use of S-transform in the time-frequency analysisthis technique is also a good alternative for denoising andremoval of artifacts that exist in ECG and EEG signals [94]Even though it has been explored in the areas of EEG andECG there are no available researches conducted in utilizingthis technique for EMG signal

S-transform was introduced in 1996 by StockwellMansinha and Lowe It is a hybrid of two advanced signalprocessing techniques which are STFT and WT [95] Dueto this it inherits the good qualities from both techniquesIt provides good resolution in both time and frequency andallows users to assess any frequency component in the time-frequency domain without the need of using any digital filterEven though S-transform uses a variable window length tomaintain a good time-frequency resolutions for all frequen-cies it still retains the phase information using a Fourierkernel

Expression for S-transform is shown in

ST119909 (119905 119891) = intinfin

minusinfin119909 (120591)

10038161003816100381610038161198911003816100381610038161003816radic2120587119890

minus(120591minus119905)212059122119890minus1198952120587119891120591119889120591 (5)

BioMed Research International 7

where 120591 and119891 denote the time of the spectral localization andFourier frequency respectively

The development of the S-transform is led by the urge toovercome the low resolution of STFT and the absence of thephase information in CWT Since the features had existed inthe S-transform it is sometimes viewed as a variable slidingwindow STFT or a phase-corrected CWT [96]

Although S-transform has better time-frequency resolu-tion than STFT the resolution is still far from perfect andneeds improvement To date there are several improved S-transform introduced by researchers such as the modified S-transform [96] and discrete orthonormal S-transform [94]

Bilinear TFD

Wigner-Ville Distribution (WVD) Wigner first introducedWVD in the area of quantum mechanics in the year 1932and later in 1948 Ville developed and applied the sametransformation to signal processing and spectral analysis[97]

Compared to STFT and WT WVD does not contain awindowing function and thus frees WVD from the smearingeffect due to the windowing function As a result it providesbest possible temporal and frequency resolution in the time-frequency plane It possesses several interesting propertiessuch as energy conservation real-valued marginal prop-erties translation covariance dilation covariance instanta-neous frequency and group delay

However because of its quadratic nature when dealingwith signals with several frequency components WVD suf-fers from the so-called cross components (inference terms)which represent significant defects in this method

These interference terms are troublesome since they mayoverlap with autoterms (signal terms) and in this manner itis hard to visually interpret the WVD image It creates theimpression that these terms must be available or the goodproperties of WVD (localization group delay instantaneousfrequency marginal properties etc) cannot be fulfilledThere is also a trade-off between the amount of interferencesand the number of good properties These are known issueswith the WVD spectrum and there are several ways to com-pensate these cross terms that has been discussed by previousresearchers [98ndash100] Equation (6) shows the formula forWVD

119882119909 (119905 119891) = intinfin

minusinfin119909(119905 + 1205912) 119909

lowast (120591 minus 1205912) 119890minus2119895120587119891120591119889120591 (6)

where 119909(119905 + 1205912)119909lowast(120591 minus 1205912) is the instantaneous autocorre-lation function and lowast shows conjugate operation

In the occasion that time smoothing window 119892(119905) anda frequency smoothing window ℎ(119905) are connected toWVDWVDwill then transform into the smoothed-pseudo-Wigner-Ville distribution (SPWVD) as composed in

SPWVD119909 (119905 119891)= ∬ℎ (119905 minus 120591) 119892 (119891 minus 120576)119882119909 (120591 120576) 119889120591 119889120576

(7)

where119882 is the WVD

A research by Subasi and Kiymik in 2010 has comparedthe performance between STFT SPWVD and CWT in termsof accuracy specificity and sensitivity The results suggestedthat all three methods provide almost similar performanceand are found to be satisfactory despite the fact that there arelittle differences between the results [99]

Choi-William Distribution (CWD) CWD is a member ofCohenrsquos class distribution function and was proposed in theyear 1989 by [101] It is able to avoid one of the main problemsof WVD which is the presence of interference in regionswhere one would expect zero power values Adoption of theexponential kernel in the distribution helps to suppress theeffect of cross term [102] CWD is given by

CWD119909 (119905 119891)

= intinfin

minusinfinintinfin

minusinfin119860119909 (120578 120591)Φ (120578 120591) 119890(1198952120587(120578119905minus120591119891))119889120578 119889120591

(8)

where

119860119909 (120578 120591) = intinfin

minusinfin119909(119905 + 1205912) 119909

lowast (119905 minus 1205912) 119890minus1198952120587119905120578119889119905 (9)

and the kernel function is given by

Φ(120578 120591) = 119890minus120572(120578120591)2 (10)

By studying the influence of muscle contraction towardsthe frequency content of EMG signal usingWVD and CWDAlemu et al observed that cross terms existing in WVD aregreatly reduced by CWD thus resulting in easier interpreta-tion with no loss of definition due to cross terms [103] Thereduction of the cross terms is due to the smoothing param-eters introduced in CWD However this still depends on thesmoothing parameter value selected As observed from theresult itself when the smoothing parameter decreased thereduction of the cross terms is better This however affectsthe time and frequency resolutions by reducing it and leadsto more signal loss As the smoothing parameter approachesinfin it resembles WVDmore [103]

B-Distribution Karthick et al introduced B-distribution in2015 to attenuate the cross terms exist in WVD and CWDThis is realized by introducing a smoothing function referredto as quadratic time-frequency distribution (QTFD) [104]Expression of the QTFD is given as follows [105]

119901 [119899 119896] = 2 sum|119898|lt1198722

119866 [119899119898] lowast 119911 [119899 + 119898] 119911

lowast [119899 minus 119898] 119890minus1198952120587119896119898119872(11)

where 119866[119899119898] is the smoothing kernel and operator lowastrepresents convolution

Kernel of the B-distribution is given by

119866 [119899119898] = ( |2119898|cosh2119899)

120573

lowast (sin 119888119898) (12)

where 120573 is a smoothing parameter

8 BioMed Research International

Three fatigue indices namely instantaneous median fre-quency (IMDF) instantaneousmean frequency (IMNF) andinstantaneous spectral entropy (ISPEn) were used alongsidethe B-distribution and it is found that all three features aredistinct in both fatigue and nonfatigue conditions The studyconcludes that B-distribution may be useful in EMG analysisunder various clinical and normal conditions

Modified B-Distribution A year later the same author cameout with a study using modified B-distribution to monitorthe progression of muscle fatigue Since the performance oftime-frequency distribution is assessed based on its abilityto reduce cross term and provide closely spaced frequencycomponents representation this modified version success-fully removes cross term interference while maintaininghigh time-frequency resolution [106] It is demonstratedthat modified B-distribution based time-frequency distribu-tion performs better in suppressing cross terms with goodtime and frequency resolution for multicomponent signalscompared with other above-mentioned techniques [107]Recently this technique has been widely used to analyzenonstationarities related to EEG signals heart rate signalsand accelerometer data based on fetal movements [108 109]

7 Discussion and Conclusion

The use of EMG processing to assess muscle fatigue duringmanual lifting was reviewed and several fatigue indices wereintroduced Research in the area of EMG signal currentlyevolves around the sensitivity variability and repeatability ofthe fatigue indices as well as the best processing techniqueswith high efficiency and less computational complexityDespite the use of traditional methods such as time distribu-tion and frequency distribution time-frequency distributionis found to be more superior in terms of monitoring since itenables the user to observe the progress of the signal in timeand frequency This is very important as at the point whenmuscle fatigue sets in frequency compression can happenUnlike the conventional frequency distribution techniquetime-frequency analysis demonstrates the time when anychanges in frequencies occur

In recent years researchers had identified that the bilinearTFD could perform better than the linear TFD such asSTFT spectrogram and WT due to the fact that it doesnot suffer from the smearing effects cause by windowingfunction Nonetheless on account of its quadratic naturebilinear TFD suffers the cross term effectsWith the existenceof high resolution Cohen class TFD such as the modified B-distribution introduced in 2016 the cross term effects canbe removed and at the same time maintain the high time-frequency resolution

Other than the processing techniques listed in this paperthere are still several techniques such as Hilbert-Huangtransform which have been used to analyze the EMG signalsThere are also already well established techniques in otherresearch areas and are proven to be effective and have yetto be implemented in analyzing EMG signals such as the S-transform This technique may be suitable for nonstationarysignals like EMG itself due to its good nature including

linearity lossless reversibility multiple resolution good time-frequency resolution and simple algorithm In conclusion itcan be seen that there are many gaps to be filled in this areain either finding new and better fatigue indices or constantlydeveloping new and improved processing techniques

Competing Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

The authors acknowledge and thank the Control and SignalProcessing Group of Universiti Putra Malaysia and Rehabili-tation andAssistive Technology ResearchGroup ofUniversitiTeknikal Malaysia Melaka for their continuous support inthe research This research is fully funded by the Ministry ofHigher Education Malaysia (MOHE)

References

[1] S Gallagher and J R Heberger ldquoExamining the interactionof force and repetition on musculoskeletal disorder risk asystematic literature reviewrdquo Human Factors vol 55 no 1 pp108ndash124 2013

[2] E M Eatough J D Way and C-H Chang ldquoUnderstandingthe link between psychosocial work stressors and work-relatedmusculoskeletal complaintsrdquo Applied Ergonomics vol 43 no 3pp 554ndash563 2012

[3] T RWaters V Putz-Anderson A Garg and L J Fine ldquoRevisedNIOSH equation for the design and evaluation ofmanual liftingtasksrdquo Ergonomics vol 36 no 7 pp 749ndash776 1993

[4] Y Roquelaure C Ha A Leclerc et al ldquoEpidemiologic surveil-lance of upper-extremity musculoskeletal disorders in theworking populationrdquo Arthritis Care and Research vol 55 no5 pp 765ndash778 2006

[5] H Arif M S E Kosnan K Jusoff et al ldquoFinite element analysisof conceptual lumbar spine for different lifting positionrdquoWorldApplied Sciences Journal vol 21 pp 68ndash75 2013

[6] C J Colloca and R N Hinrichs ldquoThe biomechanical and clin-ical significance of the lumbar erector spinae flexion-relaxationphenomenon a review of literaturerdquo Journal of Manipulativeand Physiological Therapeutics vol 28 no 8 pp 623ndash631 2005

[7] N H Kamarudin S A Ahmad M K Hassan R M Yusoffand S Z Dawal ldquoMuscle contraction analysis during liftingtaskrdquo in Proceedings of the 3rd IEEE Conference on BiomedicalEngineering and Sciences (IECBES rsquo14) pp 452ndash457 IEEE KualaLumpur Malaysia December 2014

[8] K R Voge and J B Dingwell ldquoRelative timing of changes inmuscle fatigue and movement coordination during a repetitiveone-hand lifting taskrdquo in Proceedings of the A New Beginningfor Human Health Proceddings of the 25th Annual InternationalConference of the IEEE Engineering in Medicine and BiologySociety (EMBS rsquo03) pp 1807ndash1810 IEEE Cancun MexicoSeptember 2003

[9] TN S T Zawawi A R Abdullah E F Shair I Halim and SMSaleh ldquoEMG signal analysis of fatiguemuscle activity inmanualliftingrdquo Journal of Electrical Systems vol 11 no 3 pp 319ndash3252015

BioMed Research International 9

[10] R A Al-Ashaik M Z Ramadan K S Al-Saleh and T M Kha-laf ldquoEffect of safety shoes type lifting frequency and ambienttemperature on subjectrsquos MAWL and physiological responsesrdquoInternational Journal of Industrial Ergonomics vol 50 pp 43ndash512015

[11] K P Granata and W S Marras ldquoAn EMG-assisted model oftrunk loading during free-dynamic liftingrdquo Journal of Biome-chanics vol 28 no 11 pp 1309ndash1317 1995

[12] S H Roy P Bonato and M Knaflitz ldquoEMG assessment ofback muscle function during cyclical liftingrdquo Journal of Elec-tromyography and Kinesiology vol 8 no 4 pp 233ndash245 1998

[13] R E Seroussi andM H Pope ldquoThe relationship between trunkmuscle electromyography and lifting moments in the sagittaland frontal planesrdquo Journal of Biomechanics vol 20 no 2 pp135ndash146 1987

[14] R B Graham M J Agnew and J M Stevenson ldquoEffectivenessof an on-body lifting aid at reducing low back physical demandsduring an automotive assembly task assessment of EMG res-ponse and user acceptabilityrdquo Applied Ergonomics vol 40 no5 pp 936ndash942 2009

[15] D Gagnon C Lariviere and P Loisel ldquoComparative abilityof EMG optimization and hybrid modelling approaches topredict trunk muscle forces and lumbar spine loading duringdynamic sagittal plane liftingrdquoClinical Biomechanics vol 16 no5 pp 359ndash372 2001

[16] P Dolan and M A Adams ldquoThe relationship between EMGactivity and extensor moment generation in the erector spinaemuscles during bending and lifting activitiesrdquo Journal of Biome-chanics vol 26 no 4-5 pp 513ndash522 1993

[17] P Bonato P Boissy U Della Croce and S H Roy ldquoChanges inthe surface EMG signal and the biomechanics of motion duringa repetitive lifting taskrdquo IEEE Transactions on Neural Systemsand Rehabilitation Engineering vol 10 no 1 pp 38ndash47 2002

[18] J R Potvin ldquoMechanically corrected EMG for the continuousestimation of erector spinae muscle loading during repetitiveliftingrdquo European Journal of Applied Physiology and Occupa-tional Physiology vol 74 no 1-2 pp 119ndash132 1996

[19] I Kingma and J H Van Dieen ldquoLifting over an obstacle effectsof one-handed lifting and hand support on trunk kinematicsand low back loadingrdquo Journal of Biomechanics vol 37 no 2pp 249ndash255 2004

[20] H-J Shin and J-Y Kim ldquoMeasurement of trunk muscle fatigueduring dynamic lifting and lowering as recovery time changesrdquoInternational Journal of Industrial Ergonomics vol 37 no 6 pp545ndash551 2007

[21] J Cholewicki S M McGill and R W Norman ldquoComparisonof muscle forces and joint load from an optimization and EMGassisted lumbar spine model towards development of a hybridapproachrdquo Journal of Biomechanics vol 28 no 3 pp 321ndash3311995

[22] L Straker ldquoEvidence to support using squat semi-squat andstoop techniques to lift low-lying objectsrdquo International Journalof Industrial Ergonomics vol 31 no 3 pp 149ndash160 2003

[23] C K Anderson and D B Chaffin ldquoA biomechanical evaluationof five lifting techniquesrdquo Applied Ergonomics vol 17 no 1 pp2ndash8 1986

[24] R Burgess-Limerick ldquoSquat stoop or something in betweenrdquoInternational Journal of Industrial Ergonomics vol 31 no 3 pp143ndash148 2003

[25] S M Hsiang G E Brogmus and T K Courtney ldquoLow backpain (LBP) and lifting techniquemdasha reviewrdquo International Jour-nal of Industrial Ergonomics vol 19 no 1 pp 59ndash74 1997

[26] A Merlo and I Campanini ldquoTechnical aspects of surface elec-tromyography for cliniciansrdquo The Open Rehabilitation Journalvol 3 no 1 pp 98ndash109 2010

[27] E N Kamavuako J C Rosenvang R Horup W Jensen DFarina and K B Englehart ldquoSurface versus untargeted intra-muscular EMG based classification of simultaneous and dyna-mically changing movementsrdquo IEEE Transactions on NeuralSystems and Rehabilitation Engineering vol 21 no 6 pp 992ndash998 2013

[28] L H Smith and L J Hargrove ldquoComparison of surface andintramuscular EMG pattern recognition for simultaneouswristhand motion classificationrdquo in Proceedings of the 35thAnnual International Conference of the IEEE Engineering inMedicine and Biology Society (EMBC rsquo13) pp 4223ndash4226Osaka Japan July 2013

[29] L J Hargrove K Englehart and B Hudgins ldquoA comparisonof surface and intramuscular myoelectric signal classificationrdquoIEEE Transactions on Biomedical Engineering vol 54 no 5 pp847ndash853 2007

[30] A Fratini M Cesarelli P Bifulco and M Romano ldquoRelevanceof motion artifact in electromyography recordings duringvibration treatmentrdquo Journal of Electromyography and Kinesiol-ogy vol 19 no 4 pp 710ndash718 2009

[31] C J De Luca L Donald Gilmore M Kuznetsov and S HRoy ldquoFiltering the surface EMG signal movement artifact andbaseline noise contaminationrdquo Journal of Biomechanics vol 43no 8 pp 1573ndash1579 2010

[32] H L Butler R Newell C L Hubley-Kozey and J W KozeyldquoThe interpretation of abdominal wall muscle recruitment stra-tegies change when the electrocardiogram (ECG) is removedfrom the electromyogram (EMG)rdquo Journal of Electromyographyand Kinesiology vol 19 no 2 pp e102ndashe113 2009

[33] G Lu J-S Brittain P Holland et al ldquoRemoving ECG noisefrom surface EMG signals using adaptive filteringrdquo Neuro-science Letters vol 462 no 1 pp 14ndash19 2009

[34] N W Willigenburg A Daffertshofer I Kingma and J H vanDieen ldquoRemoving ECG contamination from EMG recordingsa comparison of ICA-based and other filtering proceduresrdquoJournal of Electromyography and Kinesiology vol 22 no 3 pp485ndash493 2012

[35] M I Ibrahimy M R Ahsan and O O Khalifa ldquoDesign andoptimization of Levenberg-Marquardt based neural networkclassifier for EMG signals to identify hand motionsrdquo Measure-ment Science Review vol 13 no 3 pp 142ndash151 2013

[36] S N Sidek and A J Haja Mohideen ldquoMapping of EMG signalto hand grip force at varying wrist anglesrdquo in Proceedings ofthe 2nd IEEE-EMBS Conference on Biomedical Engineering andSciences (IECBES rsquo12) pp 648ndash653 IEEE Langkawi MalaysiaDecember 2012

[37] Y Zhan D Halliday P Jiang X Liu and J Feng ldquoDetectingtime-dependent coherence between non-stationary electro-physiological signalsmdasha combined statistical and time-freq-uency approachrdquo Journal of Neuroscience Methods vol 156 no1-2 pp 322ndash332 2006

[38] E R Ferrara and BWidrow ldquoFetal electrocardiogram enhance-ment by time-sequenced adaptive filteringrdquo IEEE Transactionson Biomedical Engineering vol 29 no 6 pp 458ndash460 1982

[39] R H Shumway ldquoTime-frequency clustering and discriminantanalysisrdquo Statistics amp Probability Letters vol 63 no 3 pp 307ndash314 2003

[40] D Korosec ldquoParametric estimation of the continuous non-stationary spectrum and its dynamics in surface EMG studiesrdquo

10 BioMed Research International

International Journal of Medical Informatics vol 58-59 pp 59ndash69 2000

[41] G Morren S Van Huffel I Helon et al ldquoEffects of non-nutritive sucking on heart rate respiration and oxygenationa model-based signal processing approachrdquo Comparative Bio-chemistry and PhysiologymdashAMolecular and Integrative Physiol-ogy vol 132 no 1 pp 97ndash106 2002

[42] R Latif E Aassif M Laaboubi A Moudden and G MazeldquoDetermination of the cut-off frequency of the anti-symmetriccircumferential waves using the time-varying autoregressiveTVARrdquo in Proceedings of the 3rd International Symposium onCommunications Control and Signal Processing (ISCCSP rsquo08)pp 357ndash361 IEEE March 2008

[43] R Latif E Aassif M Laaboubi and G Maze ldquoDeterminationof the group velocity using the time-varying autoregressiveTVARrdquo in Proceedings of the 9th International Symposium onSignal Processing and Its Applications (ISSPA rsquo07) pp 1ndash4 IEEESharjah United Arab Emirates February 2007

[44] A Al Zaman X Luo M Ferdjallah and A Khamayseh ldquoA newTVARmodeling in cascaded form for nonstationary signalsrdquo inProceedings of the IEEE International Conference on AcousticsSpeech and Signal Processing (ICASSP rsquo06) pp 353ndash356 May2006

[45] A Al Zaman M A Ahad M Ferdjallah and J J Wertsch ldquoAnew approach for muscle fatigue analysis in young adults atdifferent MVC levelsrdquo in Proceedings of the IEEE International48th Midwest Symposium on Circuits and Systems (MWSCASrsquo05) pp 499ndash502 August 2005

[46] Z G Zhang H T Liu S C Chan K D K Luk and Y HuldquoTime-dependent power spectral density estimation of surfaceelectromyography during isometric muscle contraction meth-ods and comparisonsrdquo Journal of Electromyography and Kinesi-ology vol 20 no 1 pp 89ndash101 2010

[47] AAl ZamanM Ferdjallah andAKhamayseh ldquoMuscle fatigueanalysis for healthy adults using TVAR model with instanta-neous frequency estimationrdquo in Proceedings of the 38th South-eastern Symposium on SystemTheory pp 44ndash47 March 2006

[48] K Englehart B Hudgins P A Parker and M StevensonldquoClassification of the myoelectric signal using time-frequencybased representationsrdquoMedical Engineering and Physics vol 21no 6-7 pp 431ndash438 1999

[49] N Selvaraj J Lee and K H Chon ldquoTime-varying methods forcharac-terizing nonstationary dynamics of physiological sys-temsrdquo Methods of Information in Medicine vol 49 no 5 pp435ndash442 2010

[50] D Tkach H Huang and T A Kuiken ldquoStudy of stability oftime-domain features for electromyographic pattern recogni-tionrdquo Journal of NeuroEngineering and Rehabilitation vol 7article 21 13 pages 2010

[51] S Du and M Vuskovic ldquoTemporal vs spectral approach tofeature extraction from prehensile EMG signalsrdquo in Proceedingsof the IEEE International Conference on Information Reuseand Integration (IRI rsquo04) pp 344ndash350 Las Vegas Nev USANovember 2004

[52] R M Enoka and J Duchateau ldquoMuscle fatigue what why andhow it influences muscle functionrdquo Journal of Physiology vol586 no 1 pp 11ndash23 2008

[53] W J Williams and J Jeong ldquoKernel design for reduced interfer-ence distributionsrdquo IEEE Transactions on Signal Processing vol40 no 2 pp 402ndash412 1992

[54] M B I Reaz M S Hussain and F Mohd-Yasin ldquoTechniquesof EMG signal analysis detection processing classification and

applicationsrdquo Biological Procedures Online vol 8 no 1 pp 11ndash35 2006

[55] R A Malinzak S M Colby D T Kirkendall B Yu and W EGarrett ldquoA comparison of knee joint motion patterns betweenmen and women in selected athletic tasksrdquo Clinical Biomechan-ics vol 16 no 5 pp 438ndash445 2001

[56] T I Arabadzhiev V G Dimitrov N A Dimitrova and G VDimitrov ldquoInterpretation of EMG integral or RMS and esti-mates of lsquoneuromuscular efficiencyrsquo can bemisleading in fatigu-ing contractionrdquo Journal of Electromyography and Kinesiologyvol 20 no 2 pp 223ndash232 2010

[57] C Suetta P Aagaard A Rosted et al ldquoTraining-inducedchanges in muscle CSA muscle strength EMG and rate offorce development in elderly subjects after long-term unilateraldisuserdquo Journal of Applied Physiology vol 97 no 5 pp 1954ndash1961 2004

[58] M A Oskoei andHHu ldquoSupport vectormachine-based classi-fication scheme for myoelectric control applied to upper limbrdquoIEEE Transactions on Biomedical Engineering vol 55 no 8 pp1956ndash1965 2008

[59] J J Villarejo A Frizera T F Bastos and J F Sarmiento ldquoPatternrecognition of hand movements with low density sEMG forprosthesis control purposesrdquo in Proceedings of the IEEE 13thInternational Conference onRehabilitation Robotics (ICORR rsquo13)Seattle Wash USA June 2013

[60] A Phinyomark C Limsakul and P Phukpattaranont ldquoA novelfeature extraction for robust EMG pattern recognitionrdquo Journalof Computing vol 1 no 1 pp 71ndash80 2009

[61] M Zardoshti-Kermani B C Wheeler K Badie and R MHashemi ldquoEMG Feature Evaluation for Movement Control ofUpper Extremity Prosthesesrdquo IEEE Transactions on Rehabilita-tion Engineering vol 3 no 4 pp 324ndash333 1995

[62] D Tkach H Huang and T A Kuiken ldquoStudy of stability oftime-domain features for electromyographic pattern recogni-tionrdquo Journal of NeuroEngineering and Rehabilitation vol 7 no1 article no 21 2010

[63] K Kiguchi S Kariya K Watanabe K Izumi and T FukudaldquoAn exoskeletal robot for human elbowmotion supportmdashsensorfusion adaptation and controlrdquo IEEE Transactions on SystemsMan and Cybernetics Part B Cybernetics vol 31 no 3 pp 353ndash361 2001

[64] F Al Omari J Hui C Mei and G Liu ldquoPattern recognition ofeight hand motions using feature extraction of forearm EMGsignalrdquo Proceedings of the National Academy of Sciences IndiaSection AmdashPhysical Sciences vol 84 no 3 pp 473ndash480 2014

[65] A R Siddiqi S N Sidek andAKhorshidtalab ldquoSignal process-ing of EMG signal for continuous thumb-angle estimationrdquo inProceedings of the 41st Annual Conference of the IEEE IndustrialElectronics Society (IECON rsquo15) pp 374ndash379 Yokohama JapanNovember 2015

[66] A Kilbom G M Hagg and C Kall ldquoOne-handed loadcarryingmdashcardiovascular muscular and subjective indices ofendurance and fatiguerdquo European Journal of Applied Physiologyand Occupational Physiology vol 65 no 1 pp 52ndash58 1992

[67] F AlOmari and G Liu ldquoAnalysis of extracted forearm sEMGsignal using LDA QDA K-NN classification algorithmsrdquoOpenAutomation and Control Systems Journal vol 6 no 1 pp 108ndash116 2014

[68] A Phinyomark G Chujit P Phukpattaranont C Limsakuland H Hu ldquoA preliminary study assessing time-domain EMGfeatures of classifying exercises in preventing falls in the elderlyrdquoin Proceedings of the 9th International Conference on Electri-cal EngineeringElectronics Computer Telecommunications and

BioMed Research International 11

Information Technology (ECTI-CON rsquo12) pp 1ndash4 IEEE Phetch-aburi Thailand May 2012

[69] D R Rogers and D T MacIsaac ldquoEMG-based muscle fatigueassessment during dynamic contractions using principal com-ponent analysisrdquo Journal of Electromyography and Kinesiologyvol 21 no 5 pp 811ndash818 2011

[70] K Buchenrieder ldquoDimensionality reduction for the controlof powered upper limb prosthesesrdquo in Proceedings of the 14thAnnual IEEE International Conference and Workshops on theEngineering of Computer-Based Systems (ECBS rsquo07) pp 327ndash333IEEE Tucson Ariz USA March 2007

[71] G Venugopal M Navaneethakrishna and S RamakrishnanldquoExtraction and analysis of multiple time window featuresassociated with muscle fatigue conditions using sEMG signalsrdquoExpert Systems with Applications vol 41 no 6 pp 2652ndash26592014

[72] A Phinyomark C Limsakul and P Phukpattaranont ldquoEMGfeature extraction for tolerance of white Gaussian noiserdquo inProceedings of the International Workshop and Symposium onScience and Technology (I-SEEC rsquo08) pp 178ndash183 December2008

[73] M S Al-Quraishi A J Ishak S A Ahmad and M K HasanldquoImpact of feature extraction techniques on classification accu-racy for EMG based ankle joint movementsrdquo in Proceedings ofthe 10th Asian Control Conference (ASCC rsquo15) pp 1ndash5 SabahMalaysia June 2015

[74] G-C Chang W-J Kang L Jer-Junn et al ldquoReal-time imple-mentation of electromyogram pattern recognition as a controlcommand of man-machine interfacerdquoMedical Engineering andPhysics vol 18 no 7 pp 529ndash537 1996

[75] A Georgakis L K Stergioulas and G Giakas ldquoFatigue analysisof the surface EMG signal in isometric constant force con-tractions using the averaged instantaneous frequencyrdquo IEEETransactions on Biomedical Engineering vol 50 no 2 pp 262ndash265 2003

[76] R Merletti and L R Lo Conte ldquoSurface EMG signal processingduring isometric contractionsrdquo Journal of Electromyography andKinesiology vol 7 no 4 pp 241ndash250 1997

[77] C J De Luca ldquoUse of the surface EMG signal for performanceevaluation of backmusclesrdquoMuscle and Nerve vol 16 no 2 pp210ndash216 1993

[78] F Khanam and M Ahmad ldquoFrequency based EMG powerspectrum analysis of Salat associated muscle contractionrdquo inProceedings of the 1st International Conference on Electricaland Electronic Engineering (ICEEE rsquo15) Rajshahi BangladeshNovember 2015

[79] M M Altaf B M Elbagoury F Alraddady and M RoushdyldquoExtended case-based behavior control for multi-humanoidrobotsrdquo International Journal of Humanoid Robotics vol 13 no2 2016

[80] A Phinyomark A Nuidod P Phukpattaranont and C Lim-sakul ldquoFeature extraction and reduction of wavelet transformcoefficients for EMG pattern classificationrdquo Elektronika ir Elek-trotechnika vol 122 no 6 pp 27ndash32 2012

[81] M Gonzalez-Izal A Malanda I Navarro-Amezqueta et alldquoEMG spectral indices and muscle power fatigue duringdynamic contractionsrdquo Journal of Electromyography and Kine-siology vol 20 no 2 pp 233ndash240 2010

[82] G V Dimitrov T I Arabadzhiev K N Mileva J L Bowtell NCrichton andNADimitrova ldquoMuscle fatigue during dynamiccontractions assessed by new spectral indicesrdquo Medicine andScience in Sports and Exercise vol 38 no 11 pp 1971ndash1979 2006

[83] MVitor-Costa H Bortolotti T V Camata et al ldquoEMG spectralanalysis of incremental exercise in cyclists and non-cyclistsusing fourier and wavelet transformsrdquo Revista Brasileira deCineantropometria e Desempenho Humano vol 14 no 6 pp660ndash670 2012

[84] M Wacker and H Witte ldquoTime-frequency techniques in bio-medical signal analysis a tutorial review of similarities anddifferencesrdquoMethods of Information in Medicine vol 52 no 4pp 279ndash296 2013

[85] S Karlsson J Yu and M Akay ldquoTime-frequency analysis ofmyoelectric signals during dynamic contractions a compara-tive studyrdquo IEEE Transactions on Biomedical Engineering vol47 no 2 pp 228ndash238 2000

[86] A L Martinez-Herrera L M Ledesma-Carrillo M Lopez-Ramirez S Salazar-Colores E Cabal-Yepez and A Garcia-Perez ldquoGabor and theWigner-Ville transforms for broken rotorbars detection in induction motorsrdquo in Proceedings of the 24thInternational Conference on Electronics Communications andComputers (CONIELECOMP rsquo14) pp 83ndash87 IEEE CholulaMexico February 2014

[87] D MacIsaac P A Parker and R N Scott ldquoThe short-timeFourier transform and muscle fatigue assessment in dynamiccontractionsrdquo Journal of Electromyography and Kinesiology vol11 no 6 pp 439ndash449 2001

[88] E F Shair A R Abdullah T N S Tengku Zawawi S AAhmad and SMohamad Saleh ldquoAuto-segmentation analysis ofEMG signal for lifting muscle contraction activitiesrdquo Journal ofTelecommunication Electronic and Computer Engineering vol8 no 7 pp 17ndash22 2016

[89] MM Jankovic andD B Popovic ldquoAnEMGsystem for studyingmotor control strategies and fatiguerdquo in Proceedings of the 10thSymposium on Neural Network Applications in Electrical Engi-neering (NEUREL-rsquo10) pp 15ndash18 Belgrade Serbia September2010

[90] T N S T Zawawi A R Abdullah I Halim E F Shair andS M Salleh ldquoApplication of spectrogram in analysing elec-tromyography (EMG) signals of manual liftingrdquo ARPN Journalof Engineering andApplied Sciences vol 11 no 6 pp 3603ndash36092016

[91] M Yochum T Bakir R Lepers and S Binczak ldquoEstimationof muscular fatigue under electromyostimulation using CWTrdquoIEEE Transactions on Biomedical Engineering vol 59 no 12 pp3372ndash3378 2012

[92] J L Dantas T V Camata M A O C Brunetto A C MoraesT Abrao and L R Altimari ldquoFourier and Wavelet spectralanalysis of EMG signals in isometric and dynamic maximaleffort exerciserdquo in Proceedings of the 32nd Annual InternationalConference of the IEEEEngineering inMedicine andBiology Soci-ety (EMBC rsquo10) pp 5979ndash5982 IEEE Buenos Aires ArgentinaSeptember 2010

[93] L Ai J Nan and J Shen ldquoApplication of S-transform to drivingfatigue in EEG analysisrdquo in Proceedings of the IEEE InternationalConference on Intelligent Computing and Intelligent Systems(ICIS rsquo10) pp 814ndash817 IEEE Guilin China October 2010

[94] R Upadhyay P K Padhy and P K Kankar ldquoEEG artifactremoval and noise suppression by discrete orthonormal S-transform denoisingrdquo Computers and Electrical Engineeringvol 53 pp 125ndash142 2016

[95] C Dora and P K Biswal ldquoRobust ECG artifact removalfrom EEG using continuous wavelet transformation and linearregressionrdquo in 2016 International Conference on Signal Process-ing and Communications (SPCOM) pp 1ndash5 Bangalore IndiaJune 2016

12 BioMed Research International

[96] S Assous and B Boashash ldquoEvaluation of themodified S-trans-form for time-frequency synchrony analysis and source locali-sationrdquo Eurasip Journal on Advances in Signal Processing vol2012 article 49 2012

[97] A L Ricamato R G Absher M T Moffroid and J P Tra-nowski ldquoA time-frequency approach to evaluate electromyo-graphic recordingsrdquo in Proceedings of the 5th Annual IEEESymposium on Computer-Based Medical Systems pp 520ndash527IEEE Durham NC USA 1992

[98] M Khezri and M Jahed ldquoAn inventive quadratic time-freq-uency scheme based on Wigner-Ville distribution for classifi-cation of sEMG signalsrdquo in Proceedings of the 6th InternationalSpecial Topic Conference on Information TechnologyApplicationsin Biomedicine pp 261ndash264 IEEE Tokyo Japan November2007

[99] A Subasi and M K Kiymik ldquoMuscle fatigue detection in EMGusing time-frequency methods ICA and neural networksrdquoJournal of Medical Systems vol 34 no 4 pp 777ndash785 2010

[100] W Martin and P Flandrin ldquoWigner-Ville Spectral Analysisof Nonstationary Processesrdquo IEEE Transactions on AcousticsSpeech and Signal Processing vol 33 no 6 pp 1461ndash1470 1985

[101] H-I Choi and W J Williams ldquoImproved time-frequency rep-resentation of multicomponent signals using exponential ker-nelsrdquo IEEE Transactions on Acoustics Speech and Signal Pro-cessing vol 37 no 6 pp 862ndash871 1989

[102] G R Pereira L F de Oliveira and J Nadal ldquoReducing crossterms effects in the Choi-Williams transform of mioelectricsignalsrdquo Computer Methods and Programs in Biomedicine vol111 no 3 pp 685ndash692 2013

[103] M Alemu D K Kumar and A Bradley ldquoTime-frequency anal-ysis of SEMGmdashwith special consideration to the interelectrodespacingrdquo IEEE Transactions on Neural Systems and Rehabilita-tion Engineering vol 11 no 4 pp 341ndash345 2003

[104] P A Karthick G Venugopal and S Ramakrishnan ldquoAnalysis ofsurface EMG signals under fatigue and non-fatigue conditionsusing B-distribution based quadratic time frequency distribu-tionrdquo Journal of Mechanics in Medicine and Biology vol 15 no2 Article ID 1540028 2015

[105] V Sucic B Barkat and B Boashash ldquoPerformance evaluationof the B-distributionrdquo in Proceedings of the 5th InternationalSymposium on Signal Processing and Its Applications (ISSPA rsquo99)pp 267ndash270 Queensland Australia August 1999

[106] PAKarthick and S Ramakrishnan ldquoSurface electromyographybasedmuscle fatigue progression analysis usingmodified B dis-tribution time-frequency featuresrdquo Biomedical Signal Processingand Control vol 26 pp 42ndash51 2016

[107] B Boashash and V Sucic ldquoResolution measure criteria for theobjective assessment of the performance of quadratic time-frequency distributionsrdquo IEEE Transactions on Signal Process-ing vol 51 no 5 pp 1253ndash1263 2003

[108] S Dong G Azemi and B Boashash ldquoImproved characteriza-tion of HRV signals based on instantaneous frequency featuresestimated from quadratic time-frequency distributions withdata-adapted kernelsrdquoBiomedical Signal Processing andControlvol 10 no 1 pp 153ndash165 2014

[109] B Boashash and T Ben-Jabeur ldquoDesign of a high-resolutionseparable-kernel quadratic TFD for improving newborn healthoutcomes using fetal movement detectionrdquo in Proceedings ofthe 11th International Conference on Information Science SignalProcessing and their Applications (ISSPA rsquo12) pp 354ndash359Quebec Canada July 2012

Submit your manuscripts athttpswwwhindawicom

Stem CellsInternational

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MEDIATORSINFLAMMATION

of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Behavioural Neurology

EndocrinologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Disease Markers

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

BioMed Research International

OncologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Oxidative Medicine and Cellular Longevity

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

PPAR Research

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Immunology ResearchHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

ObesityJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Computational and Mathematical Methods in Medicine

OphthalmologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Diabetes ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Research and TreatmentAIDS

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Gastroenterology Research and Practice

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Parkinsonrsquos Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttpwwwhindawicom

Page 3: EMG Processing Based Measures of Fatigue Assessment during …downloads.hindawi.com/journals/bmri/2017/3937254.pdf · 2019-07-30 · BioMedResearchInternational 3 5. Electromyography

BioMed Research International 3

5 Electromyography Preprocessing

When dealing with EMG signal noise removal is a veryimportant factor to be taken into consideration This is dueto the sensitive nature of EMG signal itself that different typesof noises or artifacts unavoidably contaminate itThese noisesare caused by various sources originates from either the skin-electrode interface hardware that intensifies the signal orother external sourcesTherefore before proceeding with theanalysis and classification of EMG signal a preprocessing isusually introduced to remove these noises Some of the noisesthat may affect the signals are inherent noise in electronicsequipment ambient noise motion artifact ECG artifactsinherent instability of signal and crosstalk However thisreview only covers three types of noises motion artifact ECGnoise and device noise

51 Motion Artifact Cutting edge technology is considerablyinvulnerable to some of the EMG noises but not to themotion artifact as it has frequency spectra that are affectingthe low frequency part of EMG signal This motion artifactoriginates at the electrode-skin interface where there existmuscle movements underneath the skin and force impulsethat goes through the muscle and the skin underlying theelectrodes causing movements [30] These movements willresult in a time-varying voltage across the two electrodes

To remove the low frequency noise caused by the motionartifact a study has been conducted to ascertain a reasonablevalue for the high-pass corner frequency filter The result ofthe study recommends the use of a Butterworth filter with a20Hz corner frequency and a 12 dBoct slope [31]

52 ECG Noise When EMG signal is measure close to theheart ECG bursts may pollute the EMG recording whichcan affect the analysis and result in misinterpretations [32]This is a natural relic that is often unavoidable It can bediminished by proper skin preparation and adjusted positionof the ground electrode To date other than the conventionalECG removal procedures such as the use of high past filteringseveral ECG noise removal techniques have been introducedthat can clean these ECG bursts but still maintaining theregular EMG characteristics

In 2009 Lu et al came out with a study to evaluate theperformance of recursive-least-square adaptive filter in theremoval of ECG interference from EMG signal [33] Theresult of the study shows that due to fast convergence ofthe recursive-least-square algorithm the filter is reliable toeffectively remove ECG noise

Later in 2012 independent component analysis (ICA)wasproposed to remove the ECG contamination The filteringeffects are measured in terms of root mean square relativeerrors and linear envelopes correlation of uncontaminatedand contaminated EMGThis technique excellently producesgood results when EMG and ECG are statistically indepen-dent [34]

53 Device Noise Another type of noise that usually affectsEMG signal is the noise generated inside the EMG device

itself that is commonly referred to as inherent noise inelectronics equipment [35] This noise cannot be completelyeliminated as all electronic equipment generates noise [36]However the severity of the noise can be reduced using high-quality electronic components

6 Electromyography Processing

Biosignal processing is a critical part in biomedical engineer-ing in order to classify the frequency content of a signalThesesignals include EMG electrocardiography (ECG) and elec-troencephalography (EEG)Most of these biosignals are non-stationary signals due to its time-varying characteristics [37ndash39] The problem of this nonstationary signal is the processof assuming it to be stationary over short-time intervalswhere stationary analysis techniques are used However thisassumption is not always suitable and further methods fornonstationary processes are needed

Past researchers have developed several techniques toprocess the nonstationary signal specifically the EMG signaland these include parametric and nonparametric approaches[40 41] The differences between these two approaches aremainly due to the parameters involved Since most of the sig-nals encounters are unknown signals without known param-eters the development of nonparametric approach whichinvolves a variety of transforms is actively researched com-pared to the parametric approach that requires the modelingof the nonstationary process Figure 1 provides an overviewof the EMG signal processing methods that are reviewed inthe manuscript

61 Parametric Approach In the field of EMG processingone type of parametric approach that has been explored isbased on the time-varying linear predictive models Thisincludes time-varying autoregressive (TVAR) model whereits parameters vary with time The time-varying spectrum inTVAR is estimated from the time-varying model parametersand the nonstationary signalrsquos instantaneous frequency (IF)can be extracted [42] IF represents the spectral peak locationof the EMG signal as it varies with time [43] Even thoughthe poles and zeros of the estimated model can be estimateddirectly from the EMG signal it is not guaranteed that thetime-varying poles remain inside the unit circle on the 119911-plane thus making TVAR model temporarily unstable [44]The constraints can be easily imposed by factorizing thedenominator of TVAR in direct form and formulating it incascaded form [45] Computation of the mean frequency(MNF) is taken as the time average of the IF estimation basedon the cascaded model in each interval [46]

Research conducted by Al Zaman et al has comparedTVAR with short-time Fourier transform (STFT) and con-cludes that even though both TVAR and STFThave producedsimilar decreasing muscle fatigue patterns from the MNFTVAR has achieved a better performance with the slope andinterception of the linear regression line is closer [47]

Another parametric approach that has been used is thetime-varying autoregressive moving average (TVARMA)Studies have been made to compare the performance of

4 BioMed Research International

Electromyography processing

Parametric approach

Time distribution Frequency distribution

Linear

transform (STFT)

Spectrogram

Wavelet transform (WT)

(TVAR) average (TVARMA)

Bilinear

(WVD)

(CWD)

Time-varying autoregressive Time-varying autoregressive moving

Time-frequency distribution

Nonparametric approach

B-distributionWigner-Ville distribution

Choi-William distributionS-transform

Short-time Fourier

Figure 1 Overall view of the electromyography signal processing methods that are reviewed in this manuscript

TVAR and TVARMA in analyzing EMG signals and it isconcluded that AR models are more favorable than ARMAbecause of the following reasons (a) estimation of ARmodelsis moderately basic since they are represented by an arrange-ment of direct mathematical statements (b) AR modelsare predominant in displaying the low recurrence sEMGsegment and (c) fluctuation of the AR model estimation onshorter fragments is lower than for ARMAmodels [40]

62 Nonparametric Approach Another EMG processingmethod is the nonparametric approaches which includeSTFT spectrogram andwavelet transformThese approachesappear to be more favorable to researchers since the fatiguesindices can be obtained directly from the EMG signal withoutknowing its parametersThe analysis of the EMG signal basedon the nonparametric approaches can be further dividedinto three types of distribution techniques time domainfrequency domain and time-frequency domain [48ndash50]

621 Time Distribution (TD) Extraction of time domainfatigue indices is very simple and easy and involves lowcomputational complexity compared to the other two tech-niques since it does not involve any transformation [51] Itcan be directly obtained from the time representation of theraw EMG signal just by doing some simple mathematicalstatisticsThe idea of fatigue is approximately connected withthe amount and rate of change of some variables that reflectmuscle alterations amid sustained contractions [52]

Listed in Table 2 are several fatigue indices in the timedomain that have been explored by previous researchers

Even though the analysis based on time domain is alreadyestablished and arewidely used since the last decade howeverit is considered less accurate since their calculations are basedon EMG signal amplitude values where there aremuch inter-ference acquired through the recording especially for fea-tures that are extracted from energy property [53] Anothermajor disadvantage of this time domain features that affectsthe accuracy comes from the nonstationary property of theEMG signal where the statistical properties changes overtime but time domain features assume the data as stationarysignal [54]

Nowadays researches involving time distribution onlyfocus on finding new fatigue indices with higher robustnesswhile most of other researchers nowadays opt for time-frequency analysis which is considered to be more accurate

622 FrequencyDistribution (FD) Inmany studies of fatiguemuscle frequency domain features are usually used to extractinformation from a signal To date there are more than20 fatigue indices in the frequency domain where two ofthe widely used parameters are the mean frequency (MNF)and median frequency (MDF) [75] Other fatigue indices arelisted in Table 3 Several techniques have been used to extractinformation fromEMG signals such as fast Fourier transform(FFT) power spectral density (PSD) and parametric meth-ods (ie AR model)

The EMG signal should undergo Fourier transform inorder to represent the signal in frequency domain The

BioMed Research International 5

Table 2 Fatigue indices in time domain

Reference Fatigue indicesMalinzak et al [55] Integrated EMG (IEMG)Arabadzhiev et al [56] Root mean square (RMS)Suetta et al [57] Mean absolute value (MAV)

Oskoei and Hu [58] Modified mean absolute value type 1(MAV1)

Villarejo et al [59] Modified mean absolute value type2 (MAV2)

Phinyomark et al [60] Simple integral square (SSI)Zardoshti-Kermani et al [61] Variance of EMG (VAR)Tkach et al [62] v-order (V)Zardoshti-Kermani et al [61] Log detector (LOG)Kiguchi et al [63] Waveform length (WL)Al Omari et al [64] Average amplitude change (AAC)

Siddiqi et al [65] Difference absolute standarddeviation value (DASDV)

Kilbom et al [66] Zero crossing (ZC)AlOmari and Liu [67] Myopulse percentage rate (MYOP)Phinyomark et al [68] Willison amplitude (WAMP)Rogers et al [69] Slope sign change (SSC)

Buchenrieder [70] Mean absolute value slope(MAVSLP)

Venugopal et al [71] Multiple hamming windows(MHW)

Du and Vuskovic [51] Multiple trapezoidal windows(MTW)

Phinyomark et al [72] Histogram of EMG (HIST)Al-Quraishi et al [73] Autoregressive coefficient (AR)Chang et al [74] Cepstral coefficient (CC)

definition of Fourier transform in the form of power spec-trum is shown in

119878119909 (119891) = 111987910038161003816100381610038161003816100381610038161003816intinfin

minusinfin119909 (119905) sdot 119890minus2120587119891119905119889119905

100381610038161003816100381610038161003816100381610038162

(1)

where 119909(119905) is the signal in time domainHowever the fluctuations of the EMG frequency compo-

nent due to the adjustments in the muscle force length andcontraction speed throughout time have caused challenges inthe usage of FFT and other traditional processing methodsThe fundamental confinement of a FFT is that it cannot givesimultaneous time and frequency localization [83]Thereforeinvestigation of the EMG signal in dynamic contraction(ie repetitive lifting) utilizing these methods may not besuccessful since it requires the signal to be stationary

623 Time-Frequency Distribution (TFD) Time-frequencyrepresentation (TFR) of a signal maps a one-dimensionalsignal of time into a two-dimensional of time and frequencyIn analyzing modifying and synthesizing nonstationarysignals TFR arewidely used since both representation of timeand frequency are taken into account thus leading to higheraccuracy [84]

Table 3 Fatigue indices in frequency domain

Reference Fatigue indicesMerletti and Lo Conte [76] Mean frequency (MNF)De Luca [77] Median frequency (MDF)Khanam and Ahmad [78] Peak frequency (PKF)Khanam and Ahmad [78] Mean power (MNP)Khanam and Ahmad [78] Total power (TTP)

Altaf et al [79] The 1st 2nd and 3rd spectralmoments (SM1 SM2 and SM3)

Phinyomark et al [80] Power spectrum ratio (PSR)

Gonzalez-Izal et al [81] Instantaneous frequency variance(119865var)

Georgakis et al [75] Averaged instantaneous frequency(AIF)

Dimitrov et al [82] Dimitrovrsquos spectral fatigue index(FInsm5)

In general TFDs consist of linear TFDs and bilinear(quadratic) TFDs The most basic form of TFD technique isthe short-time Fourier transform (STFT) STFT is one of thelinear TFDs that have beenused to analyze EMGsignals alongwith spectrogram wavelet transform S-transform and soforth Examples of the bilinear TFDs include theWigner-Villedistribution (WVD) and Choi-William distribution (CWD)

Linear TFD

Short-Time Fourier Transform (STFT) To overcome thedisadvantages of time distribution and frequency distributionand satisfy the stationary condition it is common to separatelong haul signals into blocks of narrow fragments [85] Thisought to be sufficiently narrow to be viewed as stationary andtake the FT of every segment Every FT gives the spectralinformation of a different time-slice of the signal givingsimultaneous time and frequency estimation

This was proposed by Gabor who had developed theSTFT the extended version of FT [86]

STFT119909 (119905 119908) = intinfin

minusinfin119909 (120591)119908 (120591 minus 119905) 119890minus2120587119891120591119889120591 (2)

where 119909(120591) is the EMG signal 119908(120591 minus 119905) is the observationwindow and 119905 is the variable that slides the window over thesignal 119909(120591)

The choice of the window function in STFT is criticalin order to get accurate results The shape of the windowcan be either rectangular Gaussian or elliptic depending onthe shape of the signal For sampled STFT using a Gaussianwindow it is often called Gabor transform Even though thewindow should be narrow enough to ensure that the portionof the signal that falls within the window is stationary it mustnot be too narrow since it will lead to bad localization in thefrequency domain For infinitely long window 119908(119905) = 1 willeventually cause STFT to turn into FT providing excellentfrequency localization but no time localization In contrastto infinitely short window 119908(119905) = 120575 will result in the time

6 BioMed Research International

signal (with a phase factor) which will provide excellent timelocalization but no frequency localization

Research byMacIsaac et al usesMNFas fatigue index andhas drawn three important conclusions First the nonstation-arities in EMGdo not seem to affect theMNF values obtainedfrom Fourier transform Secondly the STFT method tomeasure MNF has successfully midpoints out the impacts ofnonstationarities in dynamic compressions Third the STFTis capable of identifying a descending pattern with fatigue indynamic contraction as long as the range of motion remainsconsistent across sequential time interval [87]

Since the STFT is basic in idea and execution and isequipped for distinguishing a pattern in muscle fatigue itis an undeniable choice for applications that require simpleanalysis with acceptable results

Spectrogram The squared magnitude of the STFT is calledspectrogram It can be expressed as

119878119909 (119905 119891) =10038161003816100381610038161003816100381610038161003816intinfin

minusinfin119909 (120591)119908 (120591 minus 119905) 119890minus2120587119891120591

100381610038161003816100381610038161003816100381610038162

119889120591 (3)

where 119909(120591) is the EMG signal 119908(120591 minus 119905) is the observationwindow and 119905 is the variable that slides the window over thesignal 119909(120591)

Spectrogram can be used to obtain the power distributionand energy distribution of the signal along the frequencydirection at a given time In EMG processing the instan-taneous energy is useful to separate muscle activation fromthe baseline which is called the segmentation process Seg-mentation is important in reducing the high computationalcomplexity of TFDs The muscle activation segmentationprocess by using spectrogram has been proposed by Shair etal and resulted in a mean absolute percentage error (MAPE)of just 1404 [88]

For both STFT and spectrogram there is a compromisebetween the time-based and frequency-based perspective ofa signal Both time and frequency are represented in limitedprecision where precision is controlled by the span of thewindow and the size of the window chosen will be the samefor all frequencies

In 2010 Jankovic and Popovic have presented in theirpaper the use of spectrogram in the analysis ofmuscle fatigueThey highlighted that even though the traditional MDFtechnique provides good sensitivity for fatigue indicationhowever there is a significant slope error particularly for lowactivity of coinitiated muscles They proposed a hybrid ofalternative methods (spectrogram and scalogram) that cansuccessfully provide complete information but at the sametime produce less error prediction for low-level coinitiatedmuscles [89] As a recommendation for improvement theyalso proposed further investigation on the robustness of thismethod in dynamic exercise contractions

Later in 2016 Zawawi et al came out with a detailedanalysis of EMG signals by using spectrogram for manuallifting application By taking the average instantaneous RMSvoltage (119881rms(119905)) as fatigue index she concludes that asthe lifting height is increased the average 119881rms(119905) is alsoincreased However as the average 119881rms(119905) decreased thenumber of lifting repetitions is increased [90]

Wavelet Transform (WT) A wavelet is a waveform of effec-tively limited duration that has an average value of zero Someof its properties are short-time localized waves with zerointegral value the possibility of time shifting and flexibilityWavelet analysis produces a time-scale view of the signalTheresult of a continuous wavelet transform (CWT) is waveletcoefficients Multiplying each coefficient with the appropri-ately scaled and shifted wavelet yields the constituent waveletof the original signal Equation (4) shows the formula forCWT

CWT119909 = intinfin

minusinfin119909 (120591) 1radic|119886|Ψ (

120591 minus 119905119886 ) 119889120591 (4)

where 119905 is the translation 119886 is the scale parameter and Ψ isthe mother wavelet

Yochum et al have proposed a new fatigue index basedon the continuous wavelet transform (CWT) named 119868CWTand have compared it to other fatigue indices from literaturein terms of their sensitivity to noiseThe results show that thisnew fatigue index quantifies the EMG signal elongation dur-ing a contraction and thus makes it a suitable fatigue index119868CWT is less subjected to noise and less truncation dependentcompared to other fatigue indices based on the frequency likeMNF and MDF [91]

A comparison has been made by Dantas et al betweenSTFT and CWT in evaluating muscle fatigue in isometricand dynamic contractions The after effects of this studyexhibit that CWT and STFT analysis give comparable fatigueestimates (slope of MDF) in isometric and dynamic contrac-tions However the aftereffects of CWT for both contractionsindicate less variability (higher accuracy) in EMG signalanalysis contrasted with STFT [92]

S-Transform Another TFD technique exists is the S-transform This technique has been widely used in diverseareas of telecommunication power quality geophysics andbiomedicine due to its obvious advantages in processingnonstationary signals [93] In the area of EEG and ECGdespite the use of S-transform in the time-frequency analysisthis technique is also a good alternative for denoising andremoval of artifacts that exist in ECG and EEG signals [94]Even though it has been explored in the areas of EEG andECG there are no available researches conducted in utilizingthis technique for EMG signal

S-transform was introduced in 1996 by StockwellMansinha and Lowe It is a hybrid of two advanced signalprocessing techniques which are STFT and WT [95] Dueto this it inherits the good qualities from both techniquesIt provides good resolution in both time and frequency andallows users to assess any frequency component in the time-frequency domain without the need of using any digital filterEven though S-transform uses a variable window length tomaintain a good time-frequency resolutions for all frequen-cies it still retains the phase information using a Fourierkernel

Expression for S-transform is shown in

ST119909 (119905 119891) = intinfin

minusinfin119909 (120591)

10038161003816100381610038161198911003816100381610038161003816radic2120587119890

minus(120591minus119905)212059122119890minus1198952120587119891120591119889120591 (5)

BioMed Research International 7

where 120591 and119891 denote the time of the spectral localization andFourier frequency respectively

The development of the S-transform is led by the urge toovercome the low resolution of STFT and the absence of thephase information in CWT Since the features had existed inthe S-transform it is sometimes viewed as a variable slidingwindow STFT or a phase-corrected CWT [96]

Although S-transform has better time-frequency resolu-tion than STFT the resolution is still far from perfect andneeds improvement To date there are several improved S-transform introduced by researchers such as the modified S-transform [96] and discrete orthonormal S-transform [94]

Bilinear TFD

Wigner-Ville Distribution (WVD) Wigner first introducedWVD in the area of quantum mechanics in the year 1932and later in 1948 Ville developed and applied the sametransformation to signal processing and spectral analysis[97]

Compared to STFT and WT WVD does not contain awindowing function and thus frees WVD from the smearingeffect due to the windowing function As a result it providesbest possible temporal and frequency resolution in the time-frequency plane It possesses several interesting propertiessuch as energy conservation real-valued marginal prop-erties translation covariance dilation covariance instanta-neous frequency and group delay

However because of its quadratic nature when dealingwith signals with several frequency components WVD suf-fers from the so-called cross components (inference terms)which represent significant defects in this method

These interference terms are troublesome since they mayoverlap with autoterms (signal terms) and in this manner itis hard to visually interpret the WVD image It creates theimpression that these terms must be available or the goodproperties of WVD (localization group delay instantaneousfrequency marginal properties etc) cannot be fulfilledThere is also a trade-off between the amount of interferencesand the number of good properties These are known issueswith the WVD spectrum and there are several ways to com-pensate these cross terms that has been discussed by previousresearchers [98ndash100] Equation (6) shows the formula forWVD

119882119909 (119905 119891) = intinfin

minusinfin119909(119905 + 1205912) 119909

lowast (120591 minus 1205912) 119890minus2119895120587119891120591119889120591 (6)

where 119909(119905 + 1205912)119909lowast(120591 minus 1205912) is the instantaneous autocorre-lation function and lowast shows conjugate operation

In the occasion that time smoothing window 119892(119905) anda frequency smoothing window ℎ(119905) are connected toWVDWVDwill then transform into the smoothed-pseudo-Wigner-Ville distribution (SPWVD) as composed in

SPWVD119909 (119905 119891)= ∬ℎ (119905 minus 120591) 119892 (119891 minus 120576)119882119909 (120591 120576) 119889120591 119889120576

(7)

where119882 is the WVD

A research by Subasi and Kiymik in 2010 has comparedthe performance between STFT SPWVD and CWT in termsof accuracy specificity and sensitivity The results suggestedthat all three methods provide almost similar performanceand are found to be satisfactory despite the fact that there arelittle differences between the results [99]

Choi-William Distribution (CWD) CWD is a member ofCohenrsquos class distribution function and was proposed in theyear 1989 by [101] It is able to avoid one of the main problemsof WVD which is the presence of interference in regionswhere one would expect zero power values Adoption of theexponential kernel in the distribution helps to suppress theeffect of cross term [102] CWD is given by

CWD119909 (119905 119891)

= intinfin

minusinfinintinfin

minusinfin119860119909 (120578 120591)Φ (120578 120591) 119890(1198952120587(120578119905minus120591119891))119889120578 119889120591

(8)

where

119860119909 (120578 120591) = intinfin

minusinfin119909(119905 + 1205912) 119909

lowast (119905 minus 1205912) 119890minus1198952120587119905120578119889119905 (9)

and the kernel function is given by

Φ(120578 120591) = 119890minus120572(120578120591)2 (10)

By studying the influence of muscle contraction towardsthe frequency content of EMG signal usingWVD and CWDAlemu et al observed that cross terms existing in WVD aregreatly reduced by CWD thus resulting in easier interpreta-tion with no loss of definition due to cross terms [103] Thereduction of the cross terms is due to the smoothing param-eters introduced in CWD However this still depends on thesmoothing parameter value selected As observed from theresult itself when the smoothing parameter decreased thereduction of the cross terms is better This however affectsthe time and frequency resolutions by reducing it and leadsto more signal loss As the smoothing parameter approachesinfin it resembles WVDmore [103]

B-Distribution Karthick et al introduced B-distribution in2015 to attenuate the cross terms exist in WVD and CWDThis is realized by introducing a smoothing function referredto as quadratic time-frequency distribution (QTFD) [104]Expression of the QTFD is given as follows [105]

119901 [119899 119896] = 2 sum|119898|lt1198722

119866 [119899119898] lowast 119911 [119899 + 119898] 119911

lowast [119899 minus 119898] 119890minus1198952120587119896119898119872(11)

where 119866[119899119898] is the smoothing kernel and operator lowastrepresents convolution

Kernel of the B-distribution is given by

119866 [119899119898] = ( |2119898|cosh2119899)

120573

lowast (sin 119888119898) (12)

where 120573 is a smoothing parameter

8 BioMed Research International

Three fatigue indices namely instantaneous median fre-quency (IMDF) instantaneousmean frequency (IMNF) andinstantaneous spectral entropy (ISPEn) were used alongsidethe B-distribution and it is found that all three features aredistinct in both fatigue and nonfatigue conditions The studyconcludes that B-distribution may be useful in EMG analysisunder various clinical and normal conditions

Modified B-Distribution A year later the same author cameout with a study using modified B-distribution to monitorthe progression of muscle fatigue Since the performance oftime-frequency distribution is assessed based on its abilityto reduce cross term and provide closely spaced frequencycomponents representation this modified version success-fully removes cross term interference while maintaininghigh time-frequency resolution [106] It is demonstratedthat modified B-distribution based time-frequency distribu-tion performs better in suppressing cross terms with goodtime and frequency resolution for multicomponent signalscompared with other above-mentioned techniques [107]Recently this technique has been widely used to analyzenonstationarities related to EEG signals heart rate signalsand accelerometer data based on fetal movements [108 109]

7 Discussion and Conclusion

The use of EMG processing to assess muscle fatigue duringmanual lifting was reviewed and several fatigue indices wereintroduced Research in the area of EMG signal currentlyevolves around the sensitivity variability and repeatability ofthe fatigue indices as well as the best processing techniqueswith high efficiency and less computational complexityDespite the use of traditional methods such as time distribu-tion and frequency distribution time-frequency distributionis found to be more superior in terms of monitoring since itenables the user to observe the progress of the signal in timeand frequency This is very important as at the point whenmuscle fatigue sets in frequency compression can happenUnlike the conventional frequency distribution techniquetime-frequency analysis demonstrates the time when anychanges in frequencies occur

In recent years researchers had identified that the bilinearTFD could perform better than the linear TFD such asSTFT spectrogram and WT due to the fact that it doesnot suffer from the smearing effects cause by windowingfunction Nonetheless on account of its quadratic naturebilinear TFD suffers the cross term effectsWith the existenceof high resolution Cohen class TFD such as the modified B-distribution introduced in 2016 the cross term effects canbe removed and at the same time maintain the high time-frequency resolution

Other than the processing techniques listed in this paperthere are still several techniques such as Hilbert-Huangtransform which have been used to analyze the EMG signalsThere are also already well established techniques in otherresearch areas and are proven to be effective and have yetto be implemented in analyzing EMG signals such as the S-transform This technique may be suitable for nonstationarysignals like EMG itself due to its good nature including

linearity lossless reversibility multiple resolution good time-frequency resolution and simple algorithm In conclusion itcan be seen that there are many gaps to be filled in this areain either finding new and better fatigue indices or constantlydeveloping new and improved processing techniques

Competing Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

The authors acknowledge and thank the Control and SignalProcessing Group of Universiti Putra Malaysia and Rehabili-tation andAssistive Technology ResearchGroup ofUniversitiTeknikal Malaysia Melaka for their continuous support inthe research This research is fully funded by the Ministry ofHigher Education Malaysia (MOHE)

References

[1] S Gallagher and J R Heberger ldquoExamining the interactionof force and repetition on musculoskeletal disorder risk asystematic literature reviewrdquo Human Factors vol 55 no 1 pp108ndash124 2013

[2] E M Eatough J D Way and C-H Chang ldquoUnderstandingthe link between psychosocial work stressors and work-relatedmusculoskeletal complaintsrdquo Applied Ergonomics vol 43 no 3pp 554ndash563 2012

[3] T RWaters V Putz-Anderson A Garg and L J Fine ldquoRevisedNIOSH equation for the design and evaluation ofmanual liftingtasksrdquo Ergonomics vol 36 no 7 pp 749ndash776 1993

[4] Y Roquelaure C Ha A Leclerc et al ldquoEpidemiologic surveil-lance of upper-extremity musculoskeletal disorders in theworking populationrdquo Arthritis Care and Research vol 55 no5 pp 765ndash778 2006

[5] H Arif M S E Kosnan K Jusoff et al ldquoFinite element analysisof conceptual lumbar spine for different lifting positionrdquoWorldApplied Sciences Journal vol 21 pp 68ndash75 2013

[6] C J Colloca and R N Hinrichs ldquoThe biomechanical and clin-ical significance of the lumbar erector spinae flexion-relaxationphenomenon a review of literaturerdquo Journal of Manipulativeand Physiological Therapeutics vol 28 no 8 pp 623ndash631 2005

[7] N H Kamarudin S A Ahmad M K Hassan R M Yusoffand S Z Dawal ldquoMuscle contraction analysis during liftingtaskrdquo in Proceedings of the 3rd IEEE Conference on BiomedicalEngineering and Sciences (IECBES rsquo14) pp 452ndash457 IEEE KualaLumpur Malaysia December 2014

[8] K R Voge and J B Dingwell ldquoRelative timing of changes inmuscle fatigue and movement coordination during a repetitiveone-hand lifting taskrdquo in Proceedings of the A New Beginningfor Human Health Proceddings of the 25th Annual InternationalConference of the IEEE Engineering in Medicine and BiologySociety (EMBS rsquo03) pp 1807ndash1810 IEEE Cancun MexicoSeptember 2003

[9] TN S T Zawawi A R Abdullah E F Shair I Halim and SMSaleh ldquoEMG signal analysis of fatiguemuscle activity inmanualliftingrdquo Journal of Electrical Systems vol 11 no 3 pp 319ndash3252015

BioMed Research International 9

[10] R A Al-Ashaik M Z Ramadan K S Al-Saleh and T M Kha-laf ldquoEffect of safety shoes type lifting frequency and ambienttemperature on subjectrsquos MAWL and physiological responsesrdquoInternational Journal of Industrial Ergonomics vol 50 pp 43ndash512015

[11] K P Granata and W S Marras ldquoAn EMG-assisted model oftrunk loading during free-dynamic liftingrdquo Journal of Biome-chanics vol 28 no 11 pp 1309ndash1317 1995

[12] S H Roy P Bonato and M Knaflitz ldquoEMG assessment ofback muscle function during cyclical liftingrdquo Journal of Elec-tromyography and Kinesiology vol 8 no 4 pp 233ndash245 1998

[13] R E Seroussi andM H Pope ldquoThe relationship between trunkmuscle electromyography and lifting moments in the sagittaland frontal planesrdquo Journal of Biomechanics vol 20 no 2 pp135ndash146 1987

[14] R B Graham M J Agnew and J M Stevenson ldquoEffectivenessof an on-body lifting aid at reducing low back physical demandsduring an automotive assembly task assessment of EMG res-ponse and user acceptabilityrdquo Applied Ergonomics vol 40 no5 pp 936ndash942 2009

[15] D Gagnon C Lariviere and P Loisel ldquoComparative abilityof EMG optimization and hybrid modelling approaches topredict trunk muscle forces and lumbar spine loading duringdynamic sagittal plane liftingrdquoClinical Biomechanics vol 16 no5 pp 359ndash372 2001

[16] P Dolan and M A Adams ldquoThe relationship between EMGactivity and extensor moment generation in the erector spinaemuscles during bending and lifting activitiesrdquo Journal of Biome-chanics vol 26 no 4-5 pp 513ndash522 1993

[17] P Bonato P Boissy U Della Croce and S H Roy ldquoChanges inthe surface EMG signal and the biomechanics of motion duringa repetitive lifting taskrdquo IEEE Transactions on Neural Systemsand Rehabilitation Engineering vol 10 no 1 pp 38ndash47 2002

[18] J R Potvin ldquoMechanically corrected EMG for the continuousestimation of erector spinae muscle loading during repetitiveliftingrdquo European Journal of Applied Physiology and Occupa-tional Physiology vol 74 no 1-2 pp 119ndash132 1996

[19] I Kingma and J H Van Dieen ldquoLifting over an obstacle effectsof one-handed lifting and hand support on trunk kinematicsand low back loadingrdquo Journal of Biomechanics vol 37 no 2pp 249ndash255 2004

[20] H-J Shin and J-Y Kim ldquoMeasurement of trunk muscle fatigueduring dynamic lifting and lowering as recovery time changesrdquoInternational Journal of Industrial Ergonomics vol 37 no 6 pp545ndash551 2007

[21] J Cholewicki S M McGill and R W Norman ldquoComparisonof muscle forces and joint load from an optimization and EMGassisted lumbar spine model towards development of a hybridapproachrdquo Journal of Biomechanics vol 28 no 3 pp 321ndash3311995

[22] L Straker ldquoEvidence to support using squat semi-squat andstoop techniques to lift low-lying objectsrdquo International Journalof Industrial Ergonomics vol 31 no 3 pp 149ndash160 2003

[23] C K Anderson and D B Chaffin ldquoA biomechanical evaluationof five lifting techniquesrdquo Applied Ergonomics vol 17 no 1 pp2ndash8 1986

[24] R Burgess-Limerick ldquoSquat stoop or something in betweenrdquoInternational Journal of Industrial Ergonomics vol 31 no 3 pp143ndash148 2003

[25] S M Hsiang G E Brogmus and T K Courtney ldquoLow backpain (LBP) and lifting techniquemdasha reviewrdquo International Jour-nal of Industrial Ergonomics vol 19 no 1 pp 59ndash74 1997

[26] A Merlo and I Campanini ldquoTechnical aspects of surface elec-tromyography for cliniciansrdquo The Open Rehabilitation Journalvol 3 no 1 pp 98ndash109 2010

[27] E N Kamavuako J C Rosenvang R Horup W Jensen DFarina and K B Englehart ldquoSurface versus untargeted intra-muscular EMG based classification of simultaneous and dyna-mically changing movementsrdquo IEEE Transactions on NeuralSystems and Rehabilitation Engineering vol 21 no 6 pp 992ndash998 2013

[28] L H Smith and L J Hargrove ldquoComparison of surface andintramuscular EMG pattern recognition for simultaneouswristhand motion classificationrdquo in Proceedings of the 35thAnnual International Conference of the IEEE Engineering inMedicine and Biology Society (EMBC rsquo13) pp 4223ndash4226Osaka Japan July 2013

[29] L J Hargrove K Englehart and B Hudgins ldquoA comparisonof surface and intramuscular myoelectric signal classificationrdquoIEEE Transactions on Biomedical Engineering vol 54 no 5 pp847ndash853 2007

[30] A Fratini M Cesarelli P Bifulco and M Romano ldquoRelevanceof motion artifact in electromyography recordings duringvibration treatmentrdquo Journal of Electromyography and Kinesiol-ogy vol 19 no 4 pp 710ndash718 2009

[31] C J De Luca L Donald Gilmore M Kuznetsov and S HRoy ldquoFiltering the surface EMG signal movement artifact andbaseline noise contaminationrdquo Journal of Biomechanics vol 43no 8 pp 1573ndash1579 2010

[32] H L Butler R Newell C L Hubley-Kozey and J W KozeyldquoThe interpretation of abdominal wall muscle recruitment stra-tegies change when the electrocardiogram (ECG) is removedfrom the electromyogram (EMG)rdquo Journal of Electromyographyand Kinesiology vol 19 no 2 pp e102ndashe113 2009

[33] G Lu J-S Brittain P Holland et al ldquoRemoving ECG noisefrom surface EMG signals using adaptive filteringrdquo Neuro-science Letters vol 462 no 1 pp 14ndash19 2009

[34] N W Willigenburg A Daffertshofer I Kingma and J H vanDieen ldquoRemoving ECG contamination from EMG recordingsa comparison of ICA-based and other filtering proceduresrdquoJournal of Electromyography and Kinesiology vol 22 no 3 pp485ndash493 2012

[35] M I Ibrahimy M R Ahsan and O O Khalifa ldquoDesign andoptimization of Levenberg-Marquardt based neural networkclassifier for EMG signals to identify hand motionsrdquo Measure-ment Science Review vol 13 no 3 pp 142ndash151 2013

[36] S N Sidek and A J Haja Mohideen ldquoMapping of EMG signalto hand grip force at varying wrist anglesrdquo in Proceedings ofthe 2nd IEEE-EMBS Conference on Biomedical Engineering andSciences (IECBES rsquo12) pp 648ndash653 IEEE Langkawi MalaysiaDecember 2012

[37] Y Zhan D Halliday P Jiang X Liu and J Feng ldquoDetectingtime-dependent coherence between non-stationary electro-physiological signalsmdasha combined statistical and time-freq-uency approachrdquo Journal of Neuroscience Methods vol 156 no1-2 pp 322ndash332 2006

[38] E R Ferrara and BWidrow ldquoFetal electrocardiogram enhance-ment by time-sequenced adaptive filteringrdquo IEEE Transactionson Biomedical Engineering vol 29 no 6 pp 458ndash460 1982

[39] R H Shumway ldquoTime-frequency clustering and discriminantanalysisrdquo Statistics amp Probability Letters vol 63 no 3 pp 307ndash314 2003

[40] D Korosec ldquoParametric estimation of the continuous non-stationary spectrum and its dynamics in surface EMG studiesrdquo

10 BioMed Research International

International Journal of Medical Informatics vol 58-59 pp 59ndash69 2000

[41] G Morren S Van Huffel I Helon et al ldquoEffects of non-nutritive sucking on heart rate respiration and oxygenationa model-based signal processing approachrdquo Comparative Bio-chemistry and PhysiologymdashAMolecular and Integrative Physiol-ogy vol 132 no 1 pp 97ndash106 2002

[42] R Latif E Aassif M Laaboubi A Moudden and G MazeldquoDetermination of the cut-off frequency of the anti-symmetriccircumferential waves using the time-varying autoregressiveTVARrdquo in Proceedings of the 3rd International Symposium onCommunications Control and Signal Processing (ISCCSP rsquo08)pp 357ndash361 IEEE March 2008

[43] R Latif E Aassif M Laaboubi and G Maze ldquoDeterminationof the group velocity using the time-varying autoregressiveTVARrdquo in Proceedings of the 9th International Symposium onSignal Processing and Its Applications (ISSPA rsquo07) pp 1ndash4 IEEESharjah United Arab Emirates February 2007

[44] A Al Zaman X Luo M Ferdjallah and A Khamayseh ldquoA newTVARmodeling in cascaded form for nonstationary signalsrdquo inProceedings of the IEEE International Conference on AcousticsSpeech and Signal Processing (ICASSP rsquo06) pp 353ndash356 May2006

[45] A Al Zaman M A Ahad M Ferdjallah and J J Wertsch ldquoAnew approach for muscle fatigue analysis in young adults atdifferent MVC levelsrdquo in Proceedings of the IEEE International48th Midwest Symposium on Circuits and Systems (MWSCASrsquo05) pp 499ndash502 August 2005

[46] Z G Zhang H T Liu S C Chan K D K Luk and Y HuldquoTime-dependent power spectral density estimation of surfaceelectromyography during isometric muscle contraction meth-ods and comparisonsrdquo Journal of Electromyography and Kinesi-ology vol 20 no 1 pp 89ndash101 2010

[47] AAl ZamanM Ferdjallah andAKhamayseh ldquoMuscle fatigueanalysis for healthy adults using TVAR model with instanta-neous frequency estimationrdquo in Proceedings of the 38th South-eastern Symposium on SystemTheory pp 44ndash47 March 2006

[48] K Englehart B Hudgins P A Parker and M StevensonldquoClassification of the myoelectric signal using time-frequencybased representationsrdquoMedical Engineering and Physics vol 21no 6-7 pp 431ndash438 1999

[49] N Selvaraj J Lee and K H Chon ldquoTime-varying methods forcharac-terizing nonstationary dynamics of physiological sys-temsrdquo Methods of Information in Medicine vol 49 no 5 pp435ndash442 2010

[50] D Tkach H Huang and T A Kuiken ldquoStudy of stability oftime-domain features for electromyographic pattern recogni-tionrdquo Journal of NeuroEngineering and Rehabilitation vol 7article 21 13 pages 2010

[51] S Du and M Vuskovic ldquoTemporal vs spectral approach tofeature extraction from prehensile EMG signalsrdquo in Proceedingsof the IEEE International Conference on Information Reuseand Integration (IRI rsquo04) pp 344ndash350 Las Vegas Nev USANovember 2004

[52] R M Enoka and J Duchateau ldquoMuscle fatigue what why andhow it influences muscle functionrdquo Journal of Physiology vol586 no 1 pp 11ndash23 2008

[53] W J Williams and J Jeong ldquoKernel design for reduced interfer-ence distributionsrdquo IEEE Transactions on Signal Processing vol40 no 2 pp 402ndash412 1992

[54] M B I Reaz M S Hussain and F Mohd-Yasin ldquoTechniquesof EMG signal analysis detection processing classification and

applicationsrdquo Biological Procedures Online vol 8 no 1 pp 11ndash35 2006

[55] R A Malinzak S M Colby D T Kirkendall B Yu and W EGarrett ldquoA comparison of knee joint motion patterns betweenmen and women in selected athletic tasksrdquo Clinical Biomechan-ics vol 16 no 5 pp 438ndash445 2001

[56] T I Arabadzhiev V G Dimitrov N A Dimitrova and G VDimitrov ldquoInterpretation of EMG integral or RMS and esti-mates of lsquoneuromuscular efficiencyrsquo can bemisleading in fatigu-ing contractionrdquo Journal of Electromyography and Kinesiologyvol 20 no 2 pp 223ndash232 2010

[57] C Suetta P Aagaard A Rosted et al ldquoTraining-inducedchanges in muscle CSA muscle strength EMG and rate offorce development in elderly subjects after long-term unilateraldisuserdquo Journal of Applied Physiology vol 97 no 5 pp 1954ndash1961 2004

[58] M A Oskoei andHHu ldquoSupport vectormachine-based classi-fication scheme for myoelectric control applied to upper limbrdquoIEEE Transactions on Biomedical Engineering vol 55 no 8 pp1956ndash1965 2008

[59] J J Villarejo A Frizera T F Bastos and J F Sarmiento ldquoPatternrecognition of hand movements with low density sEMG forprosthesis control purposesrdquo in Proceedings of the IEEE 13thInternational Conference onRehabilitation Robotics (ICORR rsquo13)Seattle Wash USA June 2013

[60] A Phinyomark C Limsakul and P Phukpattaranont ldquoA novelfeature extraction for robust EMG pattern recognitionrdquo Journalof Computing vol 1 no 1 pp 71ndash80 2009

[61] M Zardoshti-Kermani B C Wheeler K Badie and R MHashemi ldquoEMG Feature Evaluation for Movement Control ofUpper Extremity Prosthesesrdquo IEEE Transactions on Rehabilita-tion Engineering vol 3 no 4 pp 324ndash333 1995

[62] D Tkach H Huang and T A Kuiken ldquoStudy of stability oftime-domain features for electromyographic pattern recogni-tionrdquo Journal of NeuroEngineering and Rehabilitation vol 7 no1 article no 21 2010

[63] K Kiguchi S Kariya K Watanabe K Izumi and T FukudaldquoAn exoskeletal robot for human elbowmotion supportmdashsensorfusion adaptation and controlrdquo IEEE Transactions on SystemsMan and Cybernetics Part B Cybernetics vol 31 no 3 pp 353ndash361 2001

[64] F Al Omari J Hui C Mei and G Liu ldquoPattern recognition ofeight hand motions using feature extraction of forearm EMGsignalrdquo Proceedings of the National Academy of Sciences IndiaSection AmdashPhysical Sciences vol 84 no 3 pp 473ndash480 2014

[65] A R Siddiqi S N Sidek andAKhorshidtalab ldquoSignal process-ing of EMG signal for continuous thumb-angle estimationrdquo inProceedings of the 41st Annual Conference of the IEEE IndustrialElectronics Society (IECON rsquo15) pp 374ndash379 Yokohama JapanNovember 2015

[66] A Kilbom G M Hagg and C Kall ldquoOne-handed loadcarryingmdashcardiovascular muscular and subjective indices ofendurance and fatiguerdquo European Journal of Applied Physiologyand Occupational Physiology vol 65 no 1 pp 52ndash58 1992

[67] F AlOmari and G Liu ldquoAnalysis of extracted forearm sEMGsignal using LDA QDA K-NN classification algorithmsrdquoOpenAutomation and Control Systems Journal vol 6 no 1 pp 108ndash116 2014

[68] A Phinyomark G Chujit P Phukpattaranont C Limsakuland H Hu ldquoA preliminary study assessing time-domain EMGfeatures of classifying exercises in preventing falls in the elderlyrdquoin Proceedings of the 9th International Conference on Electri-cal EngineeringElectronics Computer Telecommunications and

BioMed Research International 11

Information Technology (ECTI-CON rsquo12) pp 1ndash4 IEEE Phetch-aburi Thailand May 2012

[69] D R Rogers and D T MacIsaac ldquoEMG-based muscle fatigueassessment during dynamic contractions using principal com-ponent analysisrdquo Journal of Electromyography and Kinesiologyvol 21 no 5 pp 811ndash818 2011

[70] K Buchenrieder ldquoDimensionality reduction for the controlof powered upper limb prosthesesrdquo in Proceedings of the 14thAnnual IEEE International Conference and Workshops on theEngineering of Computer-Based Systems (ECBS rsquo07) pp 327ndash333IEEE Tucson Ariz USA March 2007

[71] G Venugopal M Navaneethakrishna and S RamakrishnanldquoExtraction and analysis of multiple time window featuresassociated with muscle fatigue conditions using sEMG signalsrdquoExpert Systems with Applications vol 41 no 6 pp 2652ndash26592014

[72] A Phinyomark C Limsakul and P Phukpattaranont ldquoEMGfeature extraction for tolerance of white Gaussian noiserdquo inProceedings of the International Workshop and Symposium onScience and Technology (I-SEEC rsquo08) pp 178ndash183 December2008

[73] M S Al-Quraishi A J Ishak S A Ahmad and M K HasanldquoImpact of feature extraction techniques on classification accu-racy for EMG based ankle joint movementsrdquo in Proceedings ofthe 10th Asian Control Conference (ASCC rsquo15) pp 1ndash5 SabahMalaysia June 2015

[74] G-C Chang W-J Kang L Jer-Junn et al ldquoReal-time imple-mentation of electromyogram pattern recognition as a controlcommand of man-machine interfacerdquoMedical Engineering andPhysics vol 18 no 7 pp 529ndash537 1996

[75] A Georgakis L K Stergioulas and G Giakas ldquoFatigue analysisof the surface EMG signal in isometric constant force con-tractions using the averaged instantaneous frequencyrdquo IEEETransactions on Biomedical Engineering vol 50 no 2 pp 262ndash265 2003

[76] R Merletti and L R Lo Conte ldquoSurface EMG signal processingduring isometric contractionsrdquo Journal of Electromyography andKinesiology vol 7 no 4 pp 241ndash250 1997

[77] C J De Luca ldquoUse of the surface EMG signal for performanceevaluation of backmusclesrdquoMuscle and Nerve vol 16 no 2 pp210ndash216 1993

[78] F Khanam and M Ahmad ldquoFrequency based EMG powerspectrum analysis of Salat associated muscle contractionrdquo inProceedings of the 1st International Conference on Electricaland Electronic Engineering (ICEEE rsquo15) Rajshahi BangladeshNovember 2015

[79] M M Altaf B M Elbagoury F Alraddady and M RoushdyldquoExtended case-based behavior control for multi-humanoidrobotsrdquo International Journal of Humanoid Robotics vol 13 no2 2016

[80] A Phinyomark A Nuidod P Phukpattaranont and C Lim-sakul ldquoFeature extraction and reduction of wavelet transformcoefficients for EMG pattern classificationrdquo Elektronika ir Elek-trotechnika vol 122 no 6 pp 27ndash32 2012

[81] M Gonzalez-Izal A Malanda I Navarro-Amezqueta et alldquoEMG spectral indices and muscle power fatigue duringdynamic contractionsrdquo Journal of Electromyography and Kine-siology vol 20 no 2 pp 233ndash240 2010

[82] G V Dimitrov T I Arabadzhiev K N Mileva J L Bowtell NCrichton andNADimitrova ldquoMuscle fatigue during dynamiccontractions assessed by new spectral indicesrdquo Medicine andScience in Sports and Exercise vol 38 no 11 pp 1971ndash1979 2006

[83] MVitor-Costa H Bortolotti T V Camata et al ldquoEMG spectralanalysis of incremental exercise in cyclists and non-cyclistsusing fourier and wavelet transformsrdquo Revista Brasileira deCineantropometria e Desempenho Humano vol 14 no 6 pp660ndash670 2012

[84] M Wacker and H Witte ldquoTime-frequency techniques in bio-medical signal analysis a tutorial review of similarities anddifferencesrdquoMethods of Information in Medicine vol 52 no 4pp 279ndash296 2013

[85] S Karlsson J Yu and M Akay ldquoTime-frequency analysis ofmyoelectric signals during dynamic contractions a compara-tive studyrdquo IEEE Transactions on Biomedical Engineering vol47 no 2 pp 228ndash238 2000

[86] A L Martinez-Herrera L M Ledesma-Carrillo M Lopez-Ramirez S Salazar-Colores E Cabal-Yepez and A Garcia-Perez ldquoGabor and theWigner-Ville transforms for broken rotorbars detection in induction motorsrdquo in Proceedings of the 24thInternational Conference on Electronics Communications andComputers (CONIELECOMP rsquo14) pp 83ndash87 IEEE CholulaMexico February 2014

[87] D MacIsaac P A Parker and R N Scott ldquoThe short-timeFourier transform and muscle fatigue assessment in dynamiccontractionsrdquo Journal of Electromyography and Kinesiology vol11 no 6 pp 439ndash449 2001

[88] E F Shair A R Abdullah T N S Tengku Zawawi S AAhmad and SMohamad Saleh ldquoAuto-segmentation analysis ofEMG signal for lifting muscle contraction activitiesrdquo Journal ofTelecommunication Electronic and Computer Engineering vol8 no 7 pp 17ndash22 2016

[89] MM Jankovic andD B Popovic ldquoAnEMGsystem for studyingmotor control strategies and fatiguerdquo in Proceedings of the 10thSymposium on Neural Network Applications in Electrical Engi-neering (NEUREL-rsquo10) pp 15ndash18 Belgrade Serbia September2010

[90] T N S T Zawawi A R Abdullah I Halim E F Shair andS M Salleh ldquoApplication of spectrogram in analysing elec-tromyography (EMG) signals of manual liftingrdquo ARPN Journalof Engineering andApplied Sciences vol 11 no 6 pp 3603ndash36092016

[91] M Yochum T Bakir R Lepers and S Binczak ldquoEstimationof muscular fatigue under electromyostimulation using CWTrdquoIEEE Transactions on Biomedical Engineering vol 59 no 12 pp3372ndash3378 2012

[92] J L Dantas T V Camata M A O C Brunetto A C MoraesT Abrao and L R Altimari ldquoFourier and Wavelet spectralanalysis of EMG signals in isometric and dynamic maximaleffort exerciserdquo in Proceedings of the 32nd Annual InternationalConference of the IEEEEngineering inMedicine andBiology Soci-ety (EMBC rsquo10) pp 5979ndash5982 IEEE Buenos Aires ArgentinaSeptember 2010

[93] L Ai J Nan and J Shen ldquoApplication of S-transform to drivingfatigue in EEG analysisrdquo in Proceedings of the IEEE InternationalConference on Intelligent Computing and Intelligent Systems(ICIS rsquo10) pp 814ndash817 IEEE Guilin China October 2010

[94] R Upadhyay P K Padhy and P K Kankar ldquoEEG artifactremoval and noise suppression by discrete orthonormal S-transform denoisingrdquo Computers and Electrical Engineeringvol 53 pp 125ndash142 2016

[95] C Dora and P K Biswal ldquoRobust ECG artifact removalfrom EEG using continuous wavelet transformation and linearregressionrdquo in 2016 International Conference on Signal Process-ing and Communications (SPCOM) pp 1ndash5 Bangalore IndiaJune 2016

12 BioMed Research International

[96] S Assous and B Boashash ldquoEvaluation of themodified S-trans-form for time-frequency synchrony analysis and source locali-sationrdquo Eurasip Journal on Advances in Signal Processing vol2012 article 49 2012

[97] A L Ricamato R G Absher M T Moffroid and J P Tra-nowski ldquoA time-frequency approach to evaluate electromyo-graphic recordingsrdquo in Proceedings of the 5th Annual IEEESymposium on Computer-Based Medical Systems pp 520ndash527IEEE Durham NC USA 1992

[98] M Khezri and M Jahed ldquoAn inventive quadratic time-freq-uency scheme based on Wigner-Ville distribution for classifi-cation of sEMG signalsrdquo in Proceedings of the 6th InternationalSpecial Topic Conference on Information TechnologyApplicationsin Biomedicine pp 261ndash264 IEEE Tokyo Japan November2007

[99] A Subasi and M K Kiymik ldquoMuscle fatigue detection in EMGusing time-frequency methods ICA and neural networksrdquoJournal of Medical Systems vol 34 no 4 pp 777ndash785 2010

[100] W Martin and P Flandrin ldquoWigner-Ville Spectral Analysisof Nonstationary Processesrdquo IEEE Transactions on AcousticsSpeech and Signal Processing vol 33 no 6 pp 1461ndash1470 1985

[101] H-I Choi and W J Williams ldquoImproved time-frequency rep-resentation of multicomponent signals using exponential ker-nelsrdquo IEEE Transactions on Acoustics Speech and Signal Pro-cessing vol 37 no 6 pp 862ndash871 1989

[102] G R Pereira L F de Oliveira and J Nadal ldquoReducing crossterms effects in the Choi-Williams transform of mioelectricsignalsrdquo Computer Methods and Programs in Biomedicine vol111 no 3 pp 685ndash692 2013

[103] M Alemu D K Kumar and A Bradley ldquoTime-frequency anal-ysis of SEMGmdashwith special consideration to the interelectrodespacingrdquo IEEE Transactions on Neural Systems and Rehabilita-tion Engineering vol 11 no 4 pp 341ndash345 2003

[104] P A Karthick G Venugopal and S Ramakrishnan ldquoAnalysis ofsurface EMG signals under fatigue and non-fatigue conditionsusing B-distribution based quadratic time frequency distribu-tionrdquo Journal of Mechanics in Medicine and Biology vol 15 no2 Article ID 1540028 2015

[105] V Sucic B Barkat and B Boashash ldquoPerformance evaluationof the B-distributionrdquo in Proceedings of the 5th InternationalSymposium on Signal Processing and Its Applications (ISSPA rsquo99)pp 267ndash270 Queensland Australia August 1999

[106] PAKarthick and S Ramakrishnan ldquoSurface electromyographybasedmuscle fatigue progression analysis usingmodified B dis-tribution time-frequency featuresrdquo Biomedical Signal Processingand Control vol 26 pp 42ndash51 2016

[107] B Boashash and V Sucic ldquoResolution measure criteria for theobjective assessment of the performance of quadratic time-frequency distributionsrdquo IEEE Transactions on Signal Process-ing vol 51 no 5 pp 1253ndash1263 2003

[108] S Dong G Azemi and B Boashash ldquoImproved characteriza-tion of HRV signals based on instantaneous frequency featuresestimated from quadratic time-frequency distributions withdata-adapted kernelsrdquoBiomedical Signal Processing andControlvol 10 no 1 pp 153ndash165 2014

[109] B Boashash and T Ben-Jabeur ldquoDesign of a high-resolutionseparable-kernel quadratic TFD for improving newborn healthoutcomes using fetal movement detectionrdquo in Proceedings ofthe 11th International Conference on Information Science SignalProcessing and their Applications (ISSPA rsquo12) pp 354ndash359Quebec Canada July 2012

Submit your manuscripts athttpswwwhindawicom

Stem CellsInternational

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MEDIATORSINFLAMMATION

of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Behavioural Neurology

EndocrinologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Disease Markers

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

BioMed Research International

OncologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Oxidative Medicine and Cellular Longevity

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

PPAR Research

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Immunology ResearchHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

ObesityJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Computational and Mathematical Methods in Medicine

OphthalmologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Diabetes ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Research and TreatmentAIDS

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Gastroenterology Research and Practice

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Parkinsonrsquos Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttpwwwhindawicom

Page 4: EMG Processing Based Measures of Fatigue Assessment during …downloads.hindawi.com/journals/bmri/2017/3937254.pdf · 2019-07-30 · BioMedResearchInternational 3 5. Electromyography

4 BioMed Research International

Electromyography processing

Parametric approach

Time distribution Frequency distribution

Linear

transform (STFT)

Spectrogram

Wavelet transform (WT)

(TVAR) average (TVARMA)

Bilinear

(WVD)

(CWD)

Time-varying autoregressive Time-varying autoregressive moving

Time-frequency distribution

Nonparametric approach

B-distributionWigner-Ville distribution

Choi-William distributionS-transform

Short-time Fourier

Figure 1 Overall view of the electromyography signal processing methods that are reviewed in this manuscript

TVAR and TVARMA in analyzing EMG signals and it isconcluded that AR models are more favorable than ARMAbecause of the following reasons (a) estimation of ARmodelsis moderately basic since they are represented by an arrange-ment of direct mathematical statements (b) AR modelsare predominant in displaying the low recurrence sEMGsegment and (c) fluctuation of the AR model estimation onshorter fragments is lower than for ARMAmodels [40]

62 Nonparametric Approach Another EMG processingmethod is the nonparametric approaches which includeSTFT spectrogram andwavelet transformThese approachesappear to be more favorable to researchers since the fatiguesindices can be obtained directly from the EMG signal withoutknowing its parametersThe analysis of the EMG signal basedon the nonparametric approaches can be further dividedinto three types of distribution techniques time domainfrequency domain and time-frequency domain [48ndash50]

621 Time Distribution (TD) Extraction of time domainfatigue indices is very simple and easy and involves lowcomputational complexity compared to the other two tech-niques since it does not involve any transformation [51] Itcan be directly obtained from the time representation of theraw EMG signal just by doing some simple mathematicalstatisticsThe idea of fatigue is approximately connected withthe amount and rate of change of some variables that reflectmuscle alterations amid sustained contractions [52]

Listed in Table 2 are several fatigue indices in the timedomain that have been explored by previous researchers

Even though the analysis based on time domain is alreadyestablished and arewidely used since the last decade howeverit is considered less accurate since their calculations are basedon EMG signal amplitude values where there aremuch inter-ference acquired through the recording especially for fea-tures that are extracted from energy property [53] Anothermajor disadvantage of this time domain features that affectsthe accuracy comes from the nonstationary property of theEMG signal where the statistical properties changes overtime but time domain features assume the data as stationarysignal [54]

Nowadays researches involving time distribution onlyfocus on finding new fatigue indices with higher robustnesswhile most of other researchers nowadays opt for time-frequency analysis which is considered to be more accurate

622 FrequencyDistribution (FD) Inmany studies of fatiguemuscle frequency domain features are usually used to extractinformation from a signal To date there are more than20 fatigue indices in the frequency domain where two ofthe widely used parameters are the mean frequency (MNF)and median frequency (MDF) [75] Other fatigue indices arelisted in Table 3 Several techniques have been used to extractinformation fromEMG signals such as fast Fourier transform(FFT) power spectral density (PSD) and parametric meth-ods (ie AR model)

The EMG signal should undergo Fourier transform inorder to represent the signal in frequency domain The

BioMed Research International 5

Table 2 Fatigue indices in time domain

Reference Fatigue indicesMalinzak et al [55] Integrated EMG (IEMG)Arabadzhiev et al [56] Root mean square (RMS)Suetta et al [57] Mean absolute value (MAV)

Oskoei and Hu [58] Modified mean absolute value type 1(MAV1)

Villarejo et al [59] Modified mean absolute value type2 (MAV2)

Phinyomark et al [60] Simple integral square (SSI)Zardoshti-Kermani et al [61] Variance of EMG (VAR)Tkach et al [62] v-order (V)Zardoshti-Kermani et al [61] Log detector (LOG)Kiguchi et al [63] Waveform length (WL)Al Omari et al [64] Average amplitude change (AAC)

Siddiqi et al [65] Difference absolute standarddeviation value (DASDV)

Kilbom et al [66] Zero crossing (ZC)AlOmari and Liu [67] Myopulse percentage rate (MYOP)Phinyomark et al [68] Willison amplitude (WAMP)Rogers et al [69] Slope sign change (SSC)

Buchenrieder [70] Mean absolute value slope(MAVSLP)

Venugopal et al [71] Multiple hamming windows(MHW)

Du and Vuskovic [51] Multiple trapezoidal windows(MTW)

Phinyomark et al [72] Histogram of EMG (HIST)Al-Quraishi et al [73] Autoregressive coefficient (AR)Chang et al [74] Cepstral coefficient (CC)

definition of Fourier transform in the form of power spec-trum is shown in

119878119909 (119891) = 111987910038161003816100381610038161003816100381610038161003816intinfin

minusinfin119909 (119905) sdot 119890minus2120587119891119905119889119905

100381610038161003816100381610038161003816100381610038162

(1)

where 119909(119905) is the signal in time domainHowever the fluctuations of the EMG frequency compo-

nent due to the adjustments in the muscle force length andcontraction speed throughout time have caused challenges inthe usage of FFT and other traditional processing methodsThe fundamental confinement of a FFT is that it cannot givesimultaneous time and frequency localization [83]Thereforeinvestigation of the EMG signal in dynamic contraction(ie repetitive lifting) utilizing these methods may not besuccessful since it requires the signal to be stationary

623 Time-Frequency Distribution (TFD) Time-frequencyrepresentation (TFR) of a signal maps a one-dimensionalsignal of time into a two-dimensional of time and frequencyIn analyzing modifying and synthesizing nonstationarysignals TFR arewidely used since both representation of timeand frequency are taken into account thus leading to higheraccuracy [84]

Table 3 Fatigue indices in frequency domain

Reference Fatigue indicesMerletti and Lo Conte [76] Mean frequency (MNF)De Luca [77] Median frequency (MDF)Khanam and Ahmad [78] Peak frequency (PKF)Khanam and Ahmad [78] Mean power (MNP)Khanam and Ahmad [78] Total power (TTP)

Altaf et al [79] The 1st 2nd and 3rd spectralmoments (SM1 SM2 and SM3)

Phinyomark et al [80] Power spectrum ratio (PSR)

Gonzalez-Izal et al [81] Instantaneous frequency variance(119865var)

Georgakis et al [75] Averaged instantaneous frequency(AIF)

Dimitrov et al [82] Dimitrovrsquos spectral fatigue index(FInsm5)

In general TFDs consist of linear TFDs and bilinear(quadratic) TFDs The most basic form of TFD technique isthe short-time Fourier transform (STFT) STFT is one of thelinear TFDs that have beenused to analyze EMGsignals alongwith spectrogram wavelet transform S-transform and soforth Examples of the bilinear TFDs include theWigner-Villedistribution (WVD) and Choi-William distribution (CWD)

Linear TFD

Short-Time Fourier Transform (STFT) To overcome thedisadvantages of time distribution and frequency distributionand satisfy the stationary condition it is common to separatelong haul signals into blocks of narrow fragments [85] Thisought to be sufficiently narrow to be viewed as stationary andtake the FT of every segment Every FT gives the spectralinformation of a different time-slice of the signal givingsimultaneous time and frequency estimation

This was proposed by Gabor who had developed theSTFT the extended version of FT [86]

STFT119909 (119905 119908) = intinfin

minusinfin119909 (120591)119908 (120591 minus 119905) 119890minus2120587119891120591119889120591 (2)

where 119909(120591) is the EMG signal 119908(120591 minus 119905) is the observationwindow and 119905 is the variable that slides the window over thesignal 119909(120591)

The choice of the window function in STFT is criticalin order to get accurate results The shape of the windowcan be either rectangular Gaussian or elliptic depending onthe shape of the signal For sampled STFT using a Gaussianwindow it is often called Gabor transform Even though thewindow should be narrow enough to ensure that the portionof the signal that falls within the window is stationary it mustnot be too narrow since it will lead to bad localization in thefrequency domain For infinitely long window 119908(119905) = 1 willeventually cause STFT to turn into FT providing excellentfrequency localization but no time localization In contrastto infinitely short window 119908(119905) = 120575 will result in the time

6 BioMed Research International

signal (with a phase factor) which will provide excellent timelocalization but no frequency localization

Research byMacIsaac et al usesMNFas fatigue index andhas drawn three important conclusions First the nonstation-arities in EMGdo not seem to affect theMNF values obtainedfrom Fourier transform Secondly the STFT method tomeasure MNF has successfully midpoints out the impacts ofnonstationarities in dynamic compressions Third the STFTis capable of identifying a descending pattern with fatigue indynamic contraction as long as the range of motion remainsconsistent across sequential time interval [87]

Since the STFT is basic in idea and execution and isequipped for distinguishing a pattern in muscle fatigue itis an undeniable choice for applications that require simpleanalysis with acceptable results

Spectrogram The squared magnitude of the STFT is calledspectrogram It can be expressed as

119878119909 (119905 119891) =10038161003816100381610038161003816100381610038161003816intinfin

minusinfin119909 (120591)119908 (120591 minus 119905) 119890minus2120587119891120591

100381610038161003816100381610038161003816100381610038162

119889120591 (3)

where 119909(120591) is the EMG signal 119908(120591 minus 119905) is the observationwindow and 119905 is the variable that slides the window over thesignal 119909(120591)

Spectrogram can be used to obtain the power distributionand energy distribution of the signal along the frequencydirection at a given time In EMG processing the instan-taneous energy is useful to separate muscle activation fromthe baseline which is called the segmentation process Seg-mentation is important in reducing the high computationalcomplexity of TFDs The muscle activation segmentationprocess by using spectrogram has been proposed by Shair etal and resulted in a mean absolute percentage error (MAPE)of just 1404 [88]

For both STFT and spectrogram there is a compromisebetween the time-based and frequency-based perspective ofa signal Both time and frequency are represented in limitedprecision where precision is controlled by the span of thewindow and the size of the window chosen will be the samefor all frequencies

In 2010 Jankovic and Popovic have presented in theirpaper the use of spectrogram in the analysis ofmuscle fatigueThey highlighted that even though the traditional MDFtechnique provides good sensitivity for fatigue indicationhowever there is a significant slope error particularly for lowactivity of coinitiated muscles They proposed a hybrid ofalternative methods (spectrogram and scalogram) that cansuccessfully provide complete information but at the sametime produce less error prediction for low-level coinitiatedmuscles [89] As a recommendation for improvement theyalso proposed further investigation on the robustness of thismethod in dynamic exercise contractions

Later in 2016 Zawawi et al came out with a detailedanalysis of EMG signals by using spectrogram for manuallifting application By taking the average instantaneous RMSvoltage (119881rms(119905)) as fatigue index she concludes that asthe lifting height is increased the average 119881rms(119905) is alsoincreased However as the average 119881rms(119905) decreased thenumber of lifting repetitions is increased [90]

Wavelet Transform (WT) A wavelet is a waveform of effec-tively limited duration that has an average value of zero Someof its properties are short-time localized waves with zerointegral value the possibility of time shifting and flexibilityWavelet analysis produces a time-scale view of the signalTheresult of a continuous wavelet transform (CWT) is waveletcoefficients Multiplying each coefficient with the appropri-ately scaled and shifted wavelet yields the constituent waveletof the original signal Equation (4) shows the formula forCWT

CWT119909 = intinfin

minusinfin119909 (120591) 1radic|119886|Ψ (

120591 minus 119905119886 ) 119889120591 (4)

where 119905 is the translation 119886 is the scale parameter and Ψ isthe mother wavelet

Yochum et al have proposed a new fatigue index basedon the continuous wavelet transform (CWT) named 119868CWTand have compared it to other fatigue indices from literaturein terms of their sensitivity to noiseThe results show that thisnew fatigue index quantifies the EMG signal elongation dur-ing a contraction and thus makes it a suitable fatigue index119868CWT is less subjected to noise and less truncation dependentcompared to other fatigue indices based on the frequency likeMNF and MDF [91]

A comparison has been made by Dantas et al betweenSTFT and CWT in evaluating muscle fatigue in isometricand dynamic contractions The after effects of this studyexhibit that CWT and STFT analysis give comparable fatigueestimates (slope of MDF) in isometric and dynamic contrac-tions However the aftereffects of CWT for both contractionsindicate less variability (higher accuracy) in EMG signalanalysis contrasted with STFT [92]

S-Transform Another TFD technique exists is the S-transform This technique has been widely used in diverseareas of telecommunication power quality geophysics andbiomedicine due to its obvious advantages in processingnonstationary signals [93] In the area of EEG and ECGdespite the use of S-transform in the time-frequency analysisthis technique is also a good alternative for denoising andremoval of artifacts that exist in ECG and EEG signals [94]Even though it has been explored in the areas of EEG andECG there are no available researches conducted in utilizingthis technique for EMG signal

S-transform was introduced in 1996 by StockwellMansinha and Lowe It is a hybrid of two advanced signalprocessing techniques which are STFT and WT [95] Dueto this it inherits the good qualities from both techniquesIt provides good resolution in both time and frequency andallows users to assess any frequency component in the time-frequency domain without the need of using any digital filterEven though S-transform uses a variable window length tomaintain a good time-frequency resolutions for all frequen-cies it still retains the phase information using a Fourierkernel

Expression for S-transform is shown in

ST119909 (119905 119891) = intinfin

minusinfin119909 (120591)

10038161003816100381610038161198911003816100381610038161003816radic2120587119890

minus(120591minus119905)212059122119890minus1198952120587119891120591119889120591 (5)

BioMed Research International 7

where 120591 and119891 denote the time of the spectral localization andFourier frequency respectively

The development of the S-transform is led by the urge toovercome the low resolution of STFT and the absence of thephase information in CWT Since the features had existed inthe S-transform it is sometimes viewed as a variable slidingwindow STFT or a phase-corrected CWT [96]

Although S-transform has better time-frequency resolu-tion than STFT the resolution is still far from perfect andneeds improvement To date there are several improved S-transform introduced by researchers such as the modified S-transform [96] and discrete orthonormal S-transform [94]

Bilinear TFD

Wigner-Ville Distribution (WVD) Wigner first introducedWVD in the area of quantum mechanics in the year 1932and later in 1948 Ville developed and applied the sametransformation to signal processing and spectral analysis[97]

Compared to STFT and WT WVD does not contain awindowing function and thus frees WVD from the smearingeffect due to the windowing function As a result it providesbest possible temporal and frequency resolution in the time-frequency plane It possesses several interesting propertiessuch as energy conservation real-valued marginal prop-erties translation covariance dilation covariance instanta-neous frequency and group delay

However because of its quadratic nature when dealingwith signals with several frequency components WVD suf-fers from the so-called cross components (inference terms)which represent significant defects in this method

These interference terms are troublesome since they mayoverlap with autoterms (signal terms) and in this manner itis hard to visually interpret the WVD image It creates theimpression that these terms must be available or the goodproperties of WVD (localization group delay instantaneousfrequency marginal properties etc) cannot be fulfilledThere is also a trade-off between the amount of interferencesand the number of good properties These are known issueswith the WVD spectrum and there are several ways to com-pensate these cross terms that has been discussed by previousresearchers [98ndash100] Equation (6) shows the formula forWVD

119882119909 (119905 119891) = intinfin

minusinfin119909(119905 + 1205912) 119909

lowast (120591 minus 1205912) 119890minus2119895120587119891120591119889120591 (6)

where 119909(119905 + 1205912)119909lowast(120591 minus 1205912) is the instantaneous autocorre-lation function and lowast shows conjugate operation

In the occasion that time smoothing window 119892(119905) anda frequency smoothing window ℎ(119905) are connected toWVDWVDwill then transform into the smoothed-pseudo-Wigner-Ville distribution (SPWVD) as composed in

SPWVD119909 (119905 119891)= ∬ℎ (119905 minus 120591) 119892 (119891 minus 120576)119882119909 (120591 120576) 119889120591 119889120576

(7)

where119882 is the WVD

A research by Subasi and Kiymik in 2010 has comparedthe performance between STFT SPWVD and CWT in termsof accuracy specificity and sensitivity The results suggestedthat all three methods provide almost similar performanceand are found to be satisfactory despite the fact that there arelittle differences between the results [99]

Choi-William Distribution (CWD) CWD is a member ofCohenrsquos class distribution function and was proposed in theyear 1989 by [101] It is able to avoid one of the main problemsof WVD which is the presence of interference in regionswhere one would expect zero power values Adoption of theexponential kernel in the distribution helps to suppress theeffect of cross term [102] CWD is given by

CWD119909 (119905 119891)

= intinfin

minusinfinintinfin

minusinfin119860119909 (120578 120591)Φ (120578 120591) 119890(1198952120587(120578119905minus120591119891))119889120578 119889120591

(8)

where

119860119909 (120578 120591) = intinfin

minusinfin119909(119905 + 1205912) 119909

lowast (119905 minus 1205912) 119890minus1198952120587119905120578119889119905 (9)

and the kernel function is given by

Φ(120578 120591) = 119890minus120572(120578120591)2 (10)

By studying the influence of muscle contraction towardsthe frequency content of EMG signal usingWVD and CWDAlemu et al observed that cross terms existing in WVD aregreatly reduced by CWD thus resulting in easier interpreta-tion with no loss of definition due to cross terms [103] Thereduction of the cross terms is due to the smoothing param-eters introduced in CWD However this still depends on thesmoothing parameter value selected As observed from theresult itself when the smoothing parameter decreased thereduction of the cross terms is better This however affectsthe time and frequency resolutions by reducing it and leadsto more signal loss As the smoothing parameter approachesinfin it resembles WVDmore [103]

B-Distribution Karthick et al introduced B-distribution in2015 to attenuate the cross terms exist in WVD and CWDThis is realized by introducing a smoothing function referredto as quadratic time-frequency distribution (QTFD) [104]Expression of the QTFD is given as follows [105]

119901 [119899 119896] = 2 sum|119898|lt1198722

119866 [119899119898] lowast 119911 [119899 + 119898] 119911

lowast [119899 minus 119898] 119890minus1198952120587119896119898119872(11)

where 119866[119899119898] is the smoothing kernel and operator lowastrepresents convolution

Kernel of the B-distribution is given by

119866 [119899119898] = ( |2119898|cosh2119899)

120573

lowast (sin 119888119898) (12)

where 120573 is a smoothing parameter

8 BioMed Research International

Three fatigue indices namely instantaneous median fre-quency (IMDF) instantaneousmean frequency (IMNF) andinstantaneous spectral entropy (ISPEn) were used alongsidethe B-distribution and it is found that all three features aredistinct in both fatigue and nonfatigue conditions The studyconcludes that B-distribution may be useful in EMG analysisunder various clinical and normal conditions

Modified B-Distribution A year later the same author cameout with a study using modified B-distribution to monitorthe progression of muscle fatigue Since the performance oftime-frequency distribution is assessed based on its abilityto reduce cross term and provide closely spaced frequencycomponents representation this modified version success-fully removes cross term interference while maintaininghigh time-frequency resolution [106] It is demonstratedthat modified B-distribution based time-frequency distribu-tion performs better in suppressing cross terms with goodtime and frequency resolution for multicomponent signalscompared with other above-mentioned techniques [107]Recently this technique has been widely used to analyzenonstationarities related to EEG signals heart rate signalsand accelerometer data based on fetal movements [108 109]

7 Discussion and Conclusion

The use of EMG processing to assess muscle fatigue duringmanual lifting was reviewed and several fatigue indices wereintroduced Research in the area of EMG signal currentlyevolves around the sensitivity variability and repeatability ofthe fatigue indices as well as the best processing techniqueswith high efficiency and less computational complexityDespite the use of traditional methods such as time distribu-tion and frequency distribution time-frequency distributionis found to be more superior in terms of monitoring since itenables the user to observe the progress of the signal in timeand frequency This is very important as at the point whenmuscle fatigue sets in frequency compression can happenUnlike the conventional frequency distribution techniquetime-frequency analysis demonstrates the time when anychanges in frequencies occur

In recent years researchers had identified that the bilinearTFD could perform better than the linear TFD such asSTFT spectrogram and WT due to the fact that it doesnot suffer from the smearing effects cause by windowingfunction Nonetheless on account of its quadratic naturebilinear TFD suffers the cross term effectsWith the existenceof high resolution Cohen class TFD such as the modified B-distribution introduced in 2016 the cross term effects canbe removed and at the same time maintain the high time-frequency resolution

Other than the processing techniques listed in this paperthere are still several techniques such as Hilbert-Huangtransform which have been used to analyze the EMG signalsThere are also already well established techniques in otherresearch areas and are proven to be effective and have yetto be implemented in analyzing EMG signals such as the S-transform This technique may be suitable for nonstationarysignals like EMG itself due to its good nature including

linearity lossless reversibility multiple resolution good time-frequency resolution and simple algorithm In conclusion itcan be seen that there are many gaps to be filled in this areain either finding new and better fatigue indices or constantlydeveloping new and improved processing techniques

Competing Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

The authors acknowledge and thank the Control and SignalProcessing Group of Universiti Putra Malaysia and Rehabili-tation andAssistive Technology ResearchGroup ofUniversitiTeknikal Malaysia Melaka for their continuous support inthe research This research is fully funded by the Ministry ofHigher Education Malaysia (MOHE)

References

[1] S Gallagher and J R Heberger ldquoExamining the interactionof force and repetition on musculoskeletal disorder risk asystematic literature reviewrdquo Human Factors vol 55 no 1 pp108ndash124 2013

[2] E M Eatough J D Way and C-H Chang ldquoUnderstandingthe link between psychosocial work stressors and work-relatedmusculoskeletal complaintsrdquo Applied Ergonomics vol 43 no 3pp 554ndash563 2012

[3] T RWaters V Putz-Anderson A Garg and L J Fine ldquoRevisedNIOSH equation for the design and evaluation ofmanual liftingtasksrdquo Ergonomics vol 36 no 7 pp 749ndash776 1993

[4] Y Roquelaure C Ha A Leclerc et al ldquoEpidemiologic surveil-lance of upper-extremity musculoskeletal disorders in theworking populationrdquo Arthritis Care and Research vol 55 no5 pp 765ndash778 2006

[5] H Arif M S E Kosnan K Jusoff et al ldquoFinite element analysisof conceptual lumbar spine for different lifting positionrdquoWorldApplied Sciences Journal vol 21 pp 68ndash75 2013

[6] C J Colloca and R N Hinrichs ldquoThe biomechanical and clin-ical significance of the lumbar erector spinae flexion-relaxationphenomenon a review of literaturerdquo Journal of Manipulativeand Physiological Therapeutics vol 28 no 8 pp 623ndash631 2005

[7] N H Kamarudin S A Ahmad M K Hassan R M Yusoffand S Z Dawal ldquoMuscle contraction analysis during liftingtaskrdquo in Proceedings of the 3rd IEEE Conference on BiomedicalEngineering and Sciences (IECBES rsquo14) pp 452ndash457 IEEE KualaLumpur Malaysia December 2014

[8] K R Voge and J B Dingwell ldquoRelative timing of changes inmuscle fatigue and movement coordination during a repetitiveone-hand lifting taskrdquo in Proceedings of the A New Beginningfor Human Health Proceddings of the 25th Annual InternationalConference of the IEEE Engineering in Medicine and BiologySociety (EMBS rsquo03) pp 1807ndash1810 IEEE Cancun MexicoSeptember 2003

[9] TN S T Zawawi A R Abdullah E F Shair I Halim and SMSaleh ldquoEMG signal analysis of fatiguemuscle activity inmanualliftingrdquo Journal of Electrical Systems vol 11 no 3 pp 319ndash3252015

BioMed Research International 9

[10] R A Al-Ashaik M Z Ramadan K S Al-Saleh and T M Kha-laf ldquoEffect of safety shoes type lifting frequency and ambienttemperature on subjectrsquos MAWL and physiological responsesrdquoInternational Journal of Industrial Ergonomics vol 50 pp 43ndash512015

[11] K P Granata and W S Marras ldquoAn EMG-assisted model oftrunk loading during free-dynamic liftingrdquo Journal of Biome-chanics vol 28 no 11 pp 1309ndash1317 1995

[12] S H Roy P Bonato and M Knaflitz ldquoEMG assessment ofback muscle function during cyclical liftingrdquo Journal of Elec-tromyography and Kinesiology vol 8 no 4 pp 233ndash245 1998

[13] R E Seroussi andM H Pope ldquoThe relationship between trunkmuscle electromyography and lifting moments in the sagittaland frontal planesrdquo Journal of Biomechanics vol 20 no 2 pp135ndash146 1987

[14] R B Graham M J Agnew and J M Stevenson ldquoEffectivenessof an on-body lifting aid at reducing low back physical demandsduring an automotive assembly task assessment of EMG res-ponse and user acceptabilityrdquo Applied Ergonomics vol 40 no5 pp 936ndash942 2009

[15] D Gagnon C Lariviere and P Loisel ldquoComparative abilityof EMG optimization and hybrid modelling approaches topredict trunk muscle forces and lumbar spine loading duringdynamic sagittal plane liftingrdquoClinical Biomechanics vol 16 no5 pp 359ndash372 2001

[16] P Dolan and M A Adams ldquoThe relationship between EMGactivity and extensor moment generation in the erector spinaemuscles during bending and lifting activitiesrdquo Journal of Biome-chanics vol 26 no 4-5 pp 513ndash522 1993

[17] P Bonato P Boissy U Della Croce and S H Roy ldquoChanges inthe surface EMG signal and the biomechanics of motion duringa repetitive lifting taskrdquo IEEE Transactions on Neural Systemsand Rehabilitation Engineering vol 10 no 1 pp 38ndash47 2002

[18] J R Potvin ldquoMechanically corrected EMG for the continuousestimation of erector spinae muscle loading during repetitiveliftingrdquo European Journal of Applied Physiology and Occupa-tional Physiology vol 74 no 1-2 pp 119ndash132 1996

[19] I Kingma and J H Van Dieen ldquoLifting over an obstacle effectsof one-handed lifting and hand support on trunk kinematicsand low back loadingrdquo Journal of Biomechanics vol 37 no 2pp 249ndash255 2004

[20] H-J Shin and J-Y Kim ldquoMeasurement of trunk muscle fatigueduring dynamic lifting and lowering as recovery time changesrdquoInternational Journal of Industrial Ergonomics vol 37 no 6 pp545ndash551 2007

[21] J Cholewicki S M McGill and R W Norman ldquoComparisonof muscle forces and joint load from an optimization and EMGassisted lumbar spine model towards development of a hybridapproachrdquo Journal of Biomechanics vol 28 no 3 pp 321ndash3311995

[22] L Straker ldquoEvidence to support using squat semi-squat andstoop techniques to lift low-lying objectsrdquo International Journalof Industrial Ergonomics vol 31 no 3 pp 149ndash160 2003

[23] C K Anderson and D B Chaffin ldquoA biomechanical evaluationof five lifting techniquesrdquo Applied Ergonomics vol 17 no 1 pp2ndash8 1986

[24] R Burgess-Limerick ldquoSquat stoop or something in betweenrdquoInternational Journal of Industrial Ergonomics vol 31 no 3 pp143ndash148 2003

[25] S M Hsiang G E Brogmus and T K Courtney ldquoLow backpain (LBP) and lifting techniquemdasha reviewrdquo International Jour-nal of Industrial Ergonomics vol 19 no 1 pp 59ndash74 1997

[26] A Merlo and I Campanini ldquoTechnical aspects of surface elec-tromyography for cliniciansrdquo The Open Rehabilitation Journalvol 3 no 1 pp 98ndash109 2010

[27] E N Kamavuako J C Rosenvang R Horup W Jensen DFarina and K B Englehart ldquoSurface versus untargeted intra-muscular EMG based classification of simultaneous and dyna-mically changing movementsrdquo IEEE Transactions on NeuralSystems and Rehabilitation Engineering vol 21 no 6 pp 992ndash998 2013

[28] L H Smith and L J Hargrove ldquoComparison of surface andintramuscular EMG pattern recognition for simultaneouswristhand motion classificationrdquo in Proceedings of the 35thAnnual International Conference of the IEEE Engineering inMedicine and Biology Society (EMBC rsquo13) pp 4223ndash4226Osaka Japan July 2013

[29] L J Hargrove K Englehart and B Hudgins ldquoA comparisonof surface and intramuscular myoelectric signal classificationrdquoIEEE Transactions on Biomedical Engineering vol 54 no 5 pp847ndash853 2007

[30] A Fratini M Cesarelli P Bifulco and M Romano ldquoRelevanceof motion artifact in electromyography recordings duringvibration treatmentrdquo Journal of Electromyography and Kinesiol-ogy vol 19 no 4 pp 710ndash718 2009

[31] C J De Luca L Donald Gilmore M Kuznetsov and S HRoy ldquoFiltering the surface EMG signal movement artifact andbaseline noise contaminationrdquo Journal of Biomechanics vol 43no 8 pp 1573ndash1579 2010

[32] H L Butler R Newell C L Hubley-Kozey and J W KozeyldquoThe interpretation of abdominal wall muscle recruitment stra-tegies change when the electrocardiogram (ECG) is removedfrom the electromyogram (EMG)rdquo Journal of Electromyographyand Kinesiology vol 19 no 2 pp e102ndashe113 2009

[33] G Lu J-S Brittain P Holland et al ldquoRemoving ECG noisefrom surface EMG signals using adaptive filteringrdquo Neuro-science Letters vol 462 no 1 pp 14ndash19 2009

[34] N W Willigenburg A Daffertshofer I Kingma and J H vanDieen ldquoRemoving ECG contamination from EMG recordingsa comparison of ICA-based and other filtering proceduresrdquoJournal of Electromyography and Kinesiology vol 22 no 3 pp485ndash493 2012

[35] M I Ibrahimy M R Ahsan and O O Khalifa ldquoDesign andoptimization of Levenberg-Marquardt based neural networkclassifier for EMG signals to identify hand motionsrdquo Measure-ment Science Review vol 13 no 3 pp 142ndash151 2013

[36] S N Sidek and A J Haja Mohideen ldquoMapping of EMG signalto hand grip force at varying wrist anglesrdquo in Proceedings ofthe 2nd IEEE-EMBS Conference on Biomedical Engineering andSciences (IECBES rsquo12) pp 648ndash653 IEEE Langkawi MalaysiaDecember 2012

[37] Y Zhan D Halliday P Jiang X Liu and J Feng ldquoDetectingtime-dependent coherence between non-stationary electro-physiological signalsmdasha combined statistical and time-freq-uency approachrdquo Journal of Neuroscience Methods vol 156 no1-2 pp 322ndash332 2006

[38] E R Ferrara and BWidrow ldquoFetal electrocardiogram enhance-ment by time-sequenced adaptive filteringrdquo IEEE Transactionson Biomedical Engineering vol 29 no 6 pp 458ndash460 1982

[39] R H Shumway ldquoTime-frequency clustering and discriminantanalysisrdquo Statistics amp Probability Letters vol 63 no 3 pp 307ndash314 2003

[40] D Korosec ldquoParametric estimation of the continuous non-stationary spectrum and its dynamics in surface EMG studiesrdquo

10 BioMed Research International

International Journal of Medical Informatics vol 58-59 pp 59ndash69 2000

[41] G Morren S Van Huffel I Helon et al ldquoEffects of non-nutritive sucking on heart rate respiration and oxygenationa model-based signal processing approachrdquo Comparative Bio-chemistry and PhysiologymdashAMolecular and Integrative Physiol-ogy vol 132 no 1 pp 97ndash106 2002

[42] R Latif E Aassif M Laaboubi A Moudden and G MazeldquoDetermination of the cut-off frequency of the anti-symmetriccircumferential waves using the time-varying autoregressiveTVARrdquo in Proceedings of the 3rd International Symposium onCommunications Control and Signal Processing (ISCCSP rsquo08)pp 357ndash361 IEEE March 2008

[43] R Latif E Aassif M Laaboubi and G Maze ldquoDeterminationof the group velocity using the time-varying autoregressiveTVARrdquo in Proceedings of the 9th International Symposium onSignal Processing and Its Applications (ISSPA rsquo07) pp 1ndash4 IEEESharjah United Arab Emirates February 2007

[44] A Al Zaman X Luo M Ferdjallah and A Khamayseh ldquoA newTVARmodeling in cascaded form for nonstationary signalsrdquo inProceedings of the IEEE International Conference on AcousticsSpeech and Signal Processing (ICASSP rsquo06) pp 353ndash356 May2006

[45] A Al Zaman M A Ahad M Ferdjallah and J J Wertsch ldquoAnew approach for muscle fatigue analysis in young adults atdifferent MVC levelsrdquo in Proceedings of the IEEE International48th Midwest Symposium on Circuits and Systems (MWSCASrsquo05) pp 499ndash502 August 2005

[46] Z G Zhang H T Liu S C Chan K D K Luk and Y HuldquoTime-dependent power spectral density estimation of surfaceelectromyography during isometric muscle contraction meth-ods and comparisonsrdquo Journal of Electromyography and Kinesi-ology vol 20 no 1 pp 89ndash101 2010

[47] AAl ZamanM Ferdjallah andAKhamayseh ldquoMuscle fatigueanalysis for healthy adults using TVAR model with instanta-neous frequency estimationrdquo in Proceedings of the 38th South-eastern Symposium on SystemTheory pp 44ndash47 March 2006

[48] K Englehart B Hudgins P A Parker and M StevensonldquoClassification of the myoelectric signal using time-frequencybased representationsrdquoMedical Engineering and Physics vol 21no 6-7 pp 431ndash438 1999

[49] N Selvaraj J Lee and K H Chon ldquoTime-varying methods forcharac-terizing nonstationary dynamics of physiological sys-temsrdquo Methods of Information in Medicine vol 49 no 5 pp435ndash442 2010

[50] D Tkach H Huang and T A Kuiken ldquoStudy of stability oftime-domain features for electromyographic pattern recogni-tionrdquo Journal of NeuroEngineering and Rehabilitation vol 7article 21 13 pages 2010

[51] S Du and M Vuskovic ldquoTemporal vs spectral approach tofeature extraction from prehensile EMG signalsrdquo in Proceedingsof the IEEE International Conference on Information Reuseand Integration (IRI rsquo04) pp 344ndash350 Las Vegas Nev USANovember 2004

[52] R M Enoka and J Duchateau ldquoMuscle fatigue what why andhow it influences muscle functionrdquo Journal of Physiology vol586 no 1 pp 11ndash23 2008

[53] W J Williams and J Jeong ldquoKernel design for reduced interfer-ence distributionsrdquo IEEE Transactions on Signal Processing vol40 no 2 pp 402ndash412 1992

[54] M B I Reaz M S Hussain and F Mohd-Yasin ldquoTechniquesof EMG signal analysis detection processing classification and

applicationsrdquo Biological Procedures Online vol 8 no 1 pp 11ndash35 2006

[55] R A Malinzak S M Colby D T Kirkendall B Yu and W EGarrett ldquoA comparison of knee joint motion patterns betweenmen and women in selected athletic tasksrdquo Clinical Biomechan-ics vol 16 no 5 pp 438ndash445 2001

[56] T I Arabadzhiev V G Dimitrov N A Dimitrova and G VDimitrov ldquoInterpretation of EMG integral or RMS and esti-mates of lsquoneuromuscular efficiencyrsquo can bemisleading in fatigu-ing contractionrdquo Journal of Electromyography and Kinesiologyvol 20 no 2 pp 223ndash232 2010

[57] C Suetta P Aagaard A Rosted et al ldquoTraining-inducedchanges in muscle CSA muscle strength EMG and rate offorce development in elderly subjects after long-term unilateraldisuserdquo Journal of Applied Physiology vol 97 no 5 pp 1954ndash1961 2004

[58] M A Oskoei andHHu ldquoSupport vectormachine-based classi-fication scheme for myoelectric control applied to upper limbrdquoIEEE Transactions on Biomedical Engineering vol 55 no 8 pp1956ndash1965 2008

[59] J J Villarejo A Frizera T F Bastos and J F Sarmiento ldquoPatternrecognition of hand movements with low density sEMG forprosthesis control purposesrdquo in Proceedings of the IEEE 13thInternational Conference onRehabilitation Robotics (ICORR rsquo13)Seattle Wash USA June 2013

[60] A Phinyomark C Limsakul and P Phukpattaranont ldquoA novelfeature extraction for robust EMG pattern recognitionrdquo Journalof Computing vol 1 no 1 pp 71ndash80 2009

[61] M Zardoshti-Kermani B C Wheeler K Badie and R MHashemi ldquoEMG Feature Evaluation for Movement Control ofUpper Extremity Prosthesesrdquo IEEE Transactions on Rehabilita-tion Engineering vol 3 no 4 pp 324ndash333 1995

[62] D Tkach H Huang and T A Kuiken ldquoStudy of stability oftime-domain features for electromyographic pattern recogni-tionrdquo Journal of NeuroEngineering and Rehabilitation vol 7 no1 article no 21 2010

[63] K Kiguchi S Kariya K Watanabe K Izumi and T FukudaldquoAn exoskeletal robot for human elbowmotion supportmdashsensorfusion adaptation and controlrdquo IEEE Transactions on SystemsMan and Cybernetics Part B Cybernetics vol 31 no 3 pp 353ndash361 2001

[64] F Al Omari J Hui C Mei and G Liu ldquoPattern recognition ofeight hand motions using feature extraction of forearm EMGsignalrdquo Proceedings of the National Academy of Sciences IndiaSection AmdashPhysical Sciences vol 84 no 3 pp 473ndash480 2014

[65] A R Siddiqi S N Sidek andAKhorshidtalab ldquoSignal process-ing of EMG signal for continuous thumb-angle estimationrdquo inProceedings of the 41st Annual Conference of the IEEE IndustrialElectronics Society (IECON rsquo15) pp 374ndash379 Yokohama JapanNovember 2015

[66] A Kilbom G M Hagg and C Kall ldquoOne-handed loadcarryingmdashcardiovascular muscular and subjective indices ofendurance and fatiguerdquo European Journal of Applied Physiologyand Occupational Physiology vol 65 no 1 pp 52ndash58 1992

[67] F AlOmari and G Liu ldquoAnalysis of extracted forearm sEMGsignal using LDA QDA K-NN classification algorithmsrdquoOpenAutomation and Control Systems Journal vol 6 no 1 pp 108ndash116 2014

[68] A Phinyomark G Chujit P Phukpattaranont C Limsakuland H Hu ldquoA preliminary study assessing time-domain EMGfeatures of classifying exercises in preventing falls in the elderlyrdquoin Proceedings of the 9th International Conference on Electri-cal EngineeringElectronics Computer Telecommunications and

BioMed Research International 11

Information Technology (ECTI-CON rsquo12) pp 1ndash4 IEEE Phetch-aburi Thailand May 2012

[69] D R Rogers and D T MacIsaac ldquoEMG-based muscle fatigueassessment during dynamic contractions using principal com-ponent analysisrdquo Journal of Electromyography and Kinesiologyvol 21 no 5 pp 811ndash818 2011

[70] K Buchenrieder ldquoDimensionality reduction for the controlof powered upper limb prosthesesrdquo in Proceedings of the 14thAnnual IEEE International Conference and Workshops on theEngineering of Computer-Based Systems (ECBS rsquo07) pp 327ndash333IEEE Tucson Ariz USA March 2007

[71] G Venugopal M Navaneethakrishna and S RamakrishnanldquoExtraction and analysis of multiple time window featuresassociated with muscle fatigue conditions using sEMG signalsrdquoExpert Systems with Applications vol 41 no 6 pp 2652ndash26592014

[72] A Phinyomark C Limsakul and P Phukpattaranont ldquoEMGfeature extraction for tolerance of white Gaussian noiserdquo inProceedings of the International Workshop and Symposium onScience and Technology (I-SEEC rsquo08) pp 178ndash183 December2008

[73] M S Al-Quraishi A J Ishak S A Ahmad and M K HasanldquoImpact of feature extraction techniques on classification accu-racy for EMG based ankle joint movementsrdquo in Proceedings ofthe 10th Asian Control Conference (ASCC rsquo15) pp 1ndash5 SabahMalaysia June 2015

[74] G-C Chang W-J Kang L Jer-Junn et al ldquoReal-time imple-mentation of electromyogram pattern recognition as a controlcommand of man-machine interfacerdquoMedical Engineering andPhysics vol 18 no 7 pp 529ndash537 1996

[75] A Georgakis L K Stergioulas and G Giakas ldquoFatigue analysisof the surface EMG signal in isometric constant force con-tractions using the averaged instantaneous frequencyrdquo IEEETransactions on Biomedical Engineering vol 50 no 2 pp 262ndash265 2003

[76] R Merletti and L R Lo Conte ldquoSurface EMG signal processingduring isometric contractionsrdquo Journal of Electromyography andKinesiology vol 7 no 4 pp 241ndash250 1997

[77] C J De Luca ldquoUse of the surface EMG signal for performanceevaluation of backmusclesrdquoMuscle and Nerve vol 16 no 2 pp210ndash216 1993

[78] F Khanam and M Ahmad ldquoFrequency based EMG powerspectrum analysis of Salat associated muscle contractionrdquo inProceedings of the 1st International Conference on Electricaland Electronic Engineering (ICEEE rsquo15) Rajshahi BangladeshNovember 2015

[79] M M Altaf B M Elbagoury F Alraddady and M RoushdyldquoExtended case-based behavior control for multi-humanoidrobotsrdquo International Journal of Humanoid Robotics vol 13 no2 2016

[80] A Phinyomark A Nuidod P Phukpattaranont and C Lim-sakul ldquoFeature extraction and reduction of wavelet transformcoefficients for EMG pattern classificationrdquo Elektronika ir Elek-trotechnika vol 122 no 6 pp 27ndash32 2012

[81] M Gonzalez-Izal A Malanda I Navarro-Amezqueta et alldquoEMG spectral indices and muscle power fatigue duringdynamic contractionsrdquo Journal of Electromyography and Kine-siology vol 20 no 2 pp 233ndash240 2010

[82] G V Dimitrov T I Arabadzhiev K N Mileva J L Bowtell NCrichton andNADimitrova ldquoMuscle fatigue during dynamiccontractions assessed by new spectral indicesrdquo Medicine andScience in Sports and Exercise vol 38 no 11 pp 1971ndash1979 2006

[83] MVitor-Costa H Bortolotti T V Camata et al ldquoEMG spectralanalysis of incremental exercise in cyclists and non-cyclistsusing fourier and wavelet transformsrdquo Revista Brasileira deCineantropometria e Desempenho Humano vol 14 no 6 pp660ndash670 2012

[84] M Wacker and H Witte ldquoTime-frequency techniques in bio-medical signal analysis a tutorial review of similarities anddifferencesrdquoMethods of Information in Medicine vol 52 no 4pp 279ndash296 2013

[85] S Karlsson J Yu and M Akay ldquoTime-frequency analysis ofmyoelectric signals during dynamic contractions a compara-tive studyrdquo IEEE Transactions on Biomedical Engineering vol47 no 2 pp 228ndash238 2000

[86] A L Martinez-Herrera L M Ledesma-Carrillo M Lopez-Ramirez S Salazar-Colores E Cabal-Yepez and A Garcia-Perez ldquoGabor and theWigner-Ville transforms for broken rotorbars detection in induction motorsrdquo in Proceedings of the 24thInternational Conference on Electronics Communications andComputers (CONIELECOMP rsquo14) pp 83ndash87 IEEE CholulaMexico February 2014

[87] D MacIsaac P A Parker and R N Scott ldquoThe short-timeFourier transform and muscle fatigue assessment in dynamiccontractionsrdquo Journal of Electromyography and Kinesiology vol11 no 6 pp 439ndash449 2001

[88] E F Shair A R Abdullah T N S Tengku Zawawi S AAhmad and SMohamad Saleh ldquoAuto-segmentation analysis ofEMG signal for lifting muscle contraction activitiesrdquo Journal ofTelecommunication Electronic and Computer Engineering vol8 no 7 pp 17ndash22 2016

[89] MM Jankovic andD B Popovic ldquoAnEMGsystem for studyingmotor control strategies and fatiguerdquo in Proceedings of the 10thSymposium on Neural Network Applications in Electrical Engi-neering (NEUREL-rsquo10) pp 15ndash18 Belgrade Serbia September2010

[90] T N S T Zawawi A R Abdullah I Halim E F Shair andS M Salleh ldquoApplication of spectrogram in analysing elec-tromyography (EMG) signals of manual liftingrdquo ARPN Journalof Engineering andApplied Sciences vol 11 no 6 pp 3603ndash36092016

[91] M Yochum T Bakir R Lepers and S Binczak ldquoEstimationof muscular fatigue under electromyostimulation using CWTrdquoIEEE Transactions on Biomedical Engineering vol 59 no 12 pp3372ndash3378 2012

[92] J L Dantas T V Camata M A O C Brunetto A C MoraesT Abrao and L R Altimari ldquoFourier and Wavelet spectralanalysis of EMG signals in isometric and dynamic maximaleffort exerciserdquo in Proceedings of the 32nd Annual InternationalConference of the IEEEEngineering inMedicine andBiology Soci-ety (EMBC rsquo10) pp 5979ndash5982 IEEE Buenos Aires ArgentinaSeptember 2010

[93] L Ai J Nan and J Shen ldquoApplication of S-transform to drivingfatigue in EEG analysisrdquo in Proceedings of the IEEE InternationalConference on Intelligent Computing and Intelligent Systems(ICIS rsquo10) pp 814ndash817 IEEE Guilin China October 2010

[94] R Upadhyay P K Padhy and P K Kankar ldquoEEG artifactremoval and noise suppression by discrete orthonormal S-transform denoisingrdquo Computers and Electrical Engineeringvol 53 pp 125ndash142 2016

[95] C Dora and P K Biswal ldquoRobust ECG artifact removalfrom EEG using continuous wavelet transformation and linearregressionrdquo in 2016 International Conference on Signal Process-ing and Communications (SPCOM) pp 1ndash5 Bangalore IndiaJune 2016

12 BioMed Research International

[96] S Assous and B Boashash ldquoEvaluation of themodified S-trans-form for time-frequency synchrony analysis and source locali-sationrdquo Eurasip Journal on Advances in Signal Processing vol2012 article 49 2012

[97] A L Ricamato R G Absher M T Moffroid and J P Tra-nowski ldquoA time-frequency approach to evaluate electromyo-graphic recordingsrdquo in Proceedings of the 5th Annual IEEESymposium on Computer-Based Medical Systems pp 520ndash527IEEE Durham NC USA 1992

[98] M Khezri and M Jahed ldquoAn inventive quadratic time-freq-uency scheme based on Wigner-Ville distribution for classifi-cation of sEMG signalsrdquo in Proceedings of the 6th InternationalSpecial Topic Conference on Information TechnologyApplicationsin Biomedicine pp 261ndash264 IEEE Tokyo Japan November2007

[99] A Subasi and M K Kiymik ldquoMuscle fatigue detection in EMGusing time-frequency methods ICA and neural networksrdquoJournal of Medical Systems vol 34 no 4 pp 777ndash785 2010

[100] W Martin and P Flandrin ldquoWigner-Ville Spectral Analysisof Nonstationary Processesrdquo IEEE Transactions on AcousticsSpeech and Signal Processing vol 33 no 6 pp 1461ndash1470 1985

[101] H-I Choi and W J Williams ldquoImproved time-frequency rep-resentation of multicomponent signals using exponential ker-nelsrdquo IEEE Transactions on Acoustics Speech and Signal Pro-cessing vol 37 no 6 pp 862ndash871 1989

[102] G R Pereira L F de Oliveira and J Nadal ldquoReducing crossterms effects in the Choi-Williams transform of mioelectricsignalsrdquo Computer Methods and Programs in Biomedicine vol111 no 3 pp 685ndash692 2013

[103] M Alemu D K Kumar and A Bradley ldquoTime-frequency anal-ysis of SEMGmdashwith special consideration to the interelectrodespacingrdquo IEEE Transactions on Neural Systems and Rehabilita-tion Engineering vol 11 no 4 pp 341ndash345 2003

[104] P A Karthick G Venugopal and S Ramakrishnan ldquoAnalysis ofsurface EMG signals under fatigue and non-fatigue conditionsusing B-distribution based quadratic time frequency distribu-tionrdquo Journal of Mechanics in Medicine and Biology vol 15 no2 Article ID 1540028 2015

[105] V Sucic B Barkat and B Boashash ldquoPerformance evaluationof the B-distributionrdquo in Proceedings of the 5th InternationalSymposium on Signal Processing and Its Applications (ISSPA rsquo99)pp 267ndash270 Queensland Australia August 1999

[106] PAKarthick and S Ramakrishnan ldquoSurface electromyographybasedmuscle fatigue progression analysis usingmodified B dis-tribution time-frequency featuresrdquo Biomedical Signal Processingand Control vol 26 pp 42ndash51 2016

[107] B Boashash and V Sucic ldquoResolution measure criteria for theobjective assessment of the performance of quadratic time-frequency distributionsrdquo IEEE Transactions on Signal Process-ing vol 51 no 5 pp 1253ndash1263 2003

[108] S Dong G Azemi and B Boashash ldquoImproved characteriza-tion of HRV signals based on instantaneous frequency featuresestimated from quadratic time-frequency distributions withdata-adapted kernelsrdquoBiomedical Signal Processing andControlvol 10 no 1 pp 153ndash165 2014

[109] B Boashash and T Ben-Jabeur ldquoDesign of a high-resolutionseparable-kernel quadratic TFD for improving newborn healthoutcomes using fetal movement detectionrdquo in Proceedings ofthe 11th International Conference on Information Science SignalProcessing and their Applications (ISSPA rsquo12) pp 354ndash359Quebec Canada July 2012

Submit your manuscripts athttpswwwhindawicom

Stem CellsInternational

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MEDIATORSINFLAMMATION

of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Behavioural Neurology

EndocrinologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Disease Markers

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

BioMed Research International

OncologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Oxidative Medicine and Cellular Longevity

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

PPAR Research

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Immunology ResearchHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

ObesityJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Computational and Mathematical Methods in Medicine

OphthalmologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Diabetes ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Research and TreatmentAIDS

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Gastroenterology Research and Practice

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Parkinsonrsquos Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttpwwwhindawicom

Page 5: EMG Processing Based Measures of Fatigue Assessment during …downloads.hindawi.com/journals/bmri/2017/3937254.pdf · 2019-07-30 · BioMedResearchInternational 3 5. Electromyography

BioMed Research International 5

Table 2 Fatigue indices in time domain

Reference Fatigue indicesMalinzak et al [55] Integrated EMG (IEMG)Arabadzhiev et al [56] Root mean square (RMS)Suetta et al [57] Mean absolute value (MAV)

Oskoei and Hu [58] Modified mean absolute value type 1(MAV1)

Villarejo et al [59] Modified mean absolute value type2 (MAV2)

Phinyomark et al [60] Simple integral square (SSI)Zardoshti-Kermani et al [61] Variance of EMG (VAR)Tkach et al [62] v-order (V)Zardoshti-Kermani et al [61] Log detector (LOG)Kiguchi et al [63] Waveform length (WL)Al Omari et al [64] Average amplitude change (AAC)

Siddiqi et al [65] Difference absolute standarddeviation value (DASDV)

Kilbom et al [66] Zero crossing (ZC)AlOmari and Liu [67] Myopulse percentage rate (MYOP)Phinyomark et al [68] Willison amplitude (WAMP)Rogers et al [69] Slope sign change (SSC)

Buchenrieder [70] Mean absolute value slope(MAVSLP)

Venugopal et al [71] Multiple hamming windows(MHW)

Du and Vuskovic [51] Multiple trapezoidal windows(MTW)

Phinyomark et al [72] Histogram of EMG (HIST)Al-Quraishi et al [73] Autoregressive coefficient (AR)Chang et al [74] Cepstral coefficient (CC)

definition of Fourier transform in the form of power spec-trum is shown in

119878119909 (119891) = 111987910038161003816100381610038161003816100381610038161003816intinfin

minusinfin119909 (119905) sdot 119890minus2120587119891119905119889119905

100381610038161003816100381610038161003816100381610038162

(1)

where 119909(119905) is the signal in time domainHowever the fluctuations of the EMG frequency compo-

nent due to the adjustments in the muscle force length andcontraction speed throughout time have caused challenges inthe usage of FFT and other traditional processing methodsThe fundamental confinement of a FFT is that it cannot givesimultaneous time and frequency localization [83]Thereforeinvestigation of the EMG signal in dynamic contraction(ie repetitive lifting) utilizing these methods may not besuccessful since it requires the signal to be stationary

623 Time-Frequency Distribution (TFD) Time-frequencyrepresentation (TFR) of a signal maps a one-dimensionalsignal of time into a two-dimensional of time and frequencyIn analyzing modifying and synthesizing nonstationarysignals TFR arewidely used since both representation of timeand frequency are taken into account thus leading to higheraccuracy [84]

Table 3 Fatigue indices in frequency domain

Reference Fatigue indicesMerletti and Lo Conte [76] Mean frequency (MNF)De Luca [77] Median frequency (MDF)Khanam and Ahmad [78] Peak frequency (PKF)Khanam and Ahmad [78] Mean power (MNP)Khanam and Ahmad [78] Total power (TTP)

Altaf et al [79] The 1st 2nd and 3rd spectralmoments (SM1 SM2 and SM3)

Phinyomark et al [80] Power spectrum ratio (PSR)

Gonzalez-Izal et al [81] Instantaneous frequency variance(119865var)

Georgakis et al [75] Averaged instantaneous frequency(AIF)

Dimitrov et al [82] Dimitrovrsquos spectral fatigue index(FInsm5)

In general TFDs consist of linear TFDs and bilinear(quadratic) TFDs The most basic form of TFD technique isthe short-time Fourier transform (STFT) STFT is one of thelinear TFDs that have beenused to analyze EMGsignals alongwith spectrogram wavelet transform S-transform and soforth Examples of the bilinear TFDs include theWigner-Villedistribution (WVD) and Choi-William distribution (CWD)

Linear TFD

Short-Time Fourier Transform (STFT) To overcome thedisadvantages of time distribution and frequency distributionand satisfy the stationary condition it is common to separatelong haul signals into blocks of narrow fragments [85] Thisought to be sufficiently narrow to be viewed as stationary andtake the FT of every segment Every FT gives the spectralinformation of a different time-slice of the signal givingsimultaneous time and frequency estimation

This was proposed by Gabor who had developed theSTFT the extended version of FT [86]

STFT119909 (119905 119908) = intinfin

minusinfin119909 (120591)119908 (120591 minus 119905) 119890minus2120587119891120591119889120591 (2)

where 119909(120591) is the EMG signal 119908(120591 minus 119905) is the observationwindow and 119905 is the variable that slides the window over thesignal 119909(120591)

The choice of the window function in STFT is criticalin order to get accurate results The shape of the windowcan be either rectangular Gaussian or elliptic depending onthe shape of the signal For sampled STFT using a Gaussianwindow it is often called Gabor transform Even though thewindow should be narrow enough to ensure that the portionof the signal that falls within the window is stationary it mustnot be too narrow since it will lead to bad localization in thefrequency domain For infinitely long window 119908(119905) = 1 willeventually cause STFT to turn into FT providing excellentfrequency localization but no time localization In contrastto infinitely short window 119908(119905) = 120575 will result in the time

6 BioMed Research International

signal (with a phase factor) which will provide excellent timelocalization but no frequency localization

Research byMacIsaac et al usesMNFas fatigue index andhas drawn three important conclusions First the nonstation-arities in EMGdo not seem to affect theMNF values obtainedfrom Fourier transform Secondly the STFT method tomeasure MNF has successfully midpoints out the impacts ofnonstationarities in dynamic compressions Third the STFTis capable of identifying a descending pattern with fatigue indynamic contraction as long as the range of motion remainsconsistent across sequential time interval [87]

Since the STFT is basic in idea and execution and isequipped for distinguishing a pattern in muscle fatigue itis an undeniable choice for applications that require simpleanalysis with acceptable results

Spectrogram The squared magnitude of the STFT is calledspectrogram It can be expressed as

119878119909 (119905 119891) =10038161003816100381610038161003816100381610038161003816intinfin

minusinfin119909 (120591)119908 (120591 minus 119905) 119890minus2120587119891120591

100381610038161003816100381610038161003816100381610038162

119889120591 (3)

where 119909(120591) is the EMG signal 119908(120591 minus 119905) is the observationwindow and 119905 is the variable that slides the window over thesignal 119909(120591)

Spectrogram can be used to obtain the power distributionand energy distribution of the signal along the frequencydirection at a given time In EMG processing the instan-taneous energy is useful to separate muscle activation fromthe baseline which is called the segmentation process Seg-mentation is important in reducing the high computationalcomplexity of TFDs The muscle activation segmentationprocess by using spectrogram has been proposed by Shair etal and resulted in a mean absolute percentage error (MAPE)of just 1404 [88]

For both STFT and spectrogram there is a compromisebetween the time-based and frequency-based perspective ofa signal Both time and frequency are represented in limitedprecision where precision is controlled by the span of thewindow and the size of the window chosen will be the samefor all frequencies

In 2010 Jankovic and Popovic have presented in theirpaper the use of spectrogram in the analysis ofmuscle fatigueThey highlighted that even though the traditional MDFtechnique provides good sensitivity for fatigue indicationhowever there is a significant slope error particularly for lowactivity of coinitiated muscles They proposed a hybrid ofalternative methods (spectrogram and scalogram) that cansuccessfully provide complete information but at the sametime produce less error prediction for low-level coinitiatedmuscles [89] As a recommendation for improvement theyalso proposed further investigation on the robustness of thismethod in dynamic exercise contractions

Later in 2016 Zawawi et al came out with a detailedanalysis of EMG signals by using spectrogram for manuallifting application By taking the average instantaneous RMSvoltage (119881rms(119905)) as fatigue index she concludes that asthe lifting height is increased the average 119881rms(119905) is alsoincreased However as the average 119881rms(119905) decreased thenumber of lifting repetitions is increased [90]

Wavelet Transform (WT) A wavelet is a waveform of effec-tively limited duration that has an average value of zero Someof its properties are short-time localized waves with zerointegral value the possibility of time shifting and flexibilityWavelet analysis produces a time-scale view of the signalTheresult of a continuous wavelet transform (CWT) is waveletcoefficients Multiplying each coefficient with the appropri-ately scaled and shifted wavelet yields the constituent waveletof the original signal Equation (4) shows the formula forCWT

CWT119909 = intinfin

minusinfin119909 (120591) 1radic|119886|Ψ (

120591 minus 119905119886 ) 119889120591 (4)

where 119905 is the translation 119886 is the scale parameter and Ψ isthe mother wavelet

Yochum et al have proposed a new fatigue index basedon the continuous wavelet transform (CWT) named 119868CWTand have compared it to other fatigue indices from literaturein terms of their sensitivity to noiseThe results show that thisnew fatigue index quantifies the EMG signal elongation dur-ing a contraction and thus makes it a suitable fatigue index119868CWT is less subjected to noise and less truncation dependentcompared to other fatigue indices based on the frequency likeMNF and MDF [91]

A comparison has been made by Dantas et al betweenSTFT and CWT in evaluating muscle fatigue in isometricand dynamic contractions The after effects of this studyexhibit that CWT and STFT analysis give comparable fatigueestimates (slope of MDF) in isometric and dynamic contrac-tions However the aftereffects of CWT for both contractionsindicate less variability (higher accuracy) in EMG signalanalysis contrasted with STFT [92]

S-Transform Another TFD technique exists is the S-transform This technique has been widely used in diverseareas of telecommunication power quality geophysics andbiomedicine due to its obvious advantages in processingnonstationary signals [93] In the area of EEG and ECGdespite the use of S-transform in the time-frequency analysisthis technique is also a good alternative for denoising andremoval of artifacts that exist in ECG and EEG signals [94]Even though it has been explored in the areas of EEG andECG there are no available researches conducted in utilizingthis technique for EMG signal

S-transform was introduced in 1996 by StockwellMansinha and Lowe It is a hybrid of two advanced signalprocessing techniques which are STFT and WT [95] Dueto this it inherits the good qualities from both techniquesIt provides good resolution in both time and frequency andallows users to assess any frequency component in the time-frequency domain without the need of using any digital filterEven though S-transform uses a variable window length tomaintain a good time-frequency resolutions for all frequen-cies it still retains the phase information using a Fourierkernel

Expression for S-transform is shown in

ST119909 (119905 119891) = intinfin

minusinfin119909 (120591)

10038161003816100381610038161198911003816100381610038161003816radic2120587119890

minus(120591minus119905)212059122119890minus1198952120587119891120591119889120591 (5)

BioMed Research International 7

where 120591 and119891 denote the time of the spectral localization andFourier frequency respectively

The development of the S-transform is led by the urge toovercome the low resolution of STFT and the absence of thephase information in CWT Since the features had existed inthe S-transform it is sometimes viewed as a variable slidingwindow STFT or a phase-corrected CWT [96]

Although S-transform has better time-frequency resolu-tion than STFT the resolution is still far from perfect andneeds improvement To date there are several improved S-transform introduced by researchers such as the modified S-transform [96] and discrete orthonormal S-transform [94]

Bilinear TFD

Wigner-Ville Distribution (WVD) Wigner first introducedWVD in the area of quantum mechanics in the year 1932and later in 1948 Ville developed and applied the sametransformation to signal processing and spectral analysis[97]

Compared to STFT and WT WVD does not contain awindowing function and thus frees WVD from the smearingeffect due to the windowing function As a result it providesbest possible temporal and frequency resolution in the time-frequency plane It possesses several interesting propertiessuch as energy conservation real-valued marginal prop-erties translation covariance dilation covariance instanta-neous frequency and group delay

However because of its quadratic nature when dealingwith signals with several frequency components WVD suf-fers from the so-called cross components (inference terms)which represent significant defects in this method

These interference terms are troublesome since they mayoverlap with autoterms (signal terms) and in this manner itis hard to visually interpret the WVD image It creates theimpression that these terms must be available or the goodproperties of WVD (localization group delay instantaneousfrequency marginal properties etc) cannot be fulfilledThere is also a trade-off between the amount of interferencesand the number of good properties These are known issueswith the WVD spectrum and there are several ways to com-pensate these cross terms that has been discussed by previousresearchers [98ndash100] Equation (6) shows the formula forWVD

119882119909 (119905 119891) = intinfin

minusinfin119909(119905 + 1205912) 119909

lowast (120591 minus 1205912) 119890minus2119895120587119891120591119889120591 (6)

where 119909(119905 + 1205912)119909lowast(120591 minus 1205912) is the instantaneous autocorre-lation function and lowast shows conjugate operation

In the occasion that time smoothing window 119892(119905) anda frequency smoothing window ℎ(119905) are connected toWVDWVDwill then transform into the smoothed-pseudo-Wigner-Ville distribution (SPWVD) as composed in

SPWVD119909 (119905 119891)= ∬ℎ (119905 minus 120591) 119892 (119891 minus 120576)119882119909 (120591 120576) 119889120591 119889120576

(7)

where119882 is the WVD

A research by Subasi and Kiymik in 2010 has comparedthe performance between STFT SPWVD and CWT in termsof accuracy specificity and sensitivity The results suggestedthat all three methods provide almost similar performanceand are found to be satisfactory despite the fact that there arelittle differences between the results [99]

Choi-William Distribution (CWD) CWD is a member ofCohenrsquos class distribution function and was proposed in theyear 1989 by [101] It is able to avoid one of the main problemsof WVD which is the presence of interference in regionswhere one would expect zero power values Adoption of theexponential kernel in the distribution helps to suppress theeffect of cross term [102] CWD is given by

CWD119909 (119905 119891)

= intinfin

minusinfinintinfin

minusinfin119860119909 (120578 120591)Φ (120578 120591) 119890(1198952120587(120578119905minus120591119891))119889120578 119889120591

(8)

where

119860119909 (120578 120591) = intinfin

minusinfin119909(119905 + 1205912) 119909

lowast (119905 minus 1205912) 119890minus1198952120587119905120578119889119905 (9)

and the kernel function is given by

Φ(120578 120591) = 119890minus120572(120578120591)2 (10)

By studying the influence of muscle contraction towardsthe frequency content of EMG signal usingWVD and CWDAlemu et al observed that cross terms existing in WVD aregreatly reduced by CWD thus resulting in easier interpreta-tion with no loss of definition due to cross terms [103] Thereduction of the cross terms is due to the smoothing param-eters introduced in CWD However this still depends on thesmoothing parameter value selected As observed from theresult itself when the smoothing parameter decreased thereduction of the cross terms is better This however affectsthe time and frequency resolutions by reducing it and leadsto more signal loss As the smoothing parameter approachesinfin it resembles WVDmore [103]

B-Distribution Karthick et al introduced B-distribution in2015 to attenuate the cross terms exist in WVD and CWDThis is realized by introducing a smoothing function referredto as quadratic time-frequency distribution (QTFD) [104]Expression of the QTFD is given as follows [105]

119901 [119899 119896] = 2 sum|119898|lt1198722

119866 [119899119898] lowast 119911 [119899 + 119898] 119911

lowast [119899 minus 119898] 119890minus1198952120587119896119898119872(11)

where 119866[119899119898] is the smoothing kernel and operator lowastrepresents convolution

Kernel of the B-distribution is given by

119866 [119899119898] = ( |2119898|cosh2119899)

120573

lowast (sin 119888119898) (12)

where 120573 is a smoothing parameter

8 BioMed Research International

Three fatigue indices namely instantaneous median fre-quency (IMDF) instantaneousmean frequency (IMNF) andinstantaneous spectral entropy (ISPEn) were used alongsidethe B-distribution and it is found that all three features aredistinct in both fatigue and nonfatigue conditions The studyconcludes that B-distribution may be useful in EMG analysisunder various clinical and normal conditions

Modified B-Distribution A year later the same author cameout with a study using modified B-distribution to monitorthe progression of muscle fatigue Since the performance oftime-frequency distribution is assessed based on its abilityto reduce cross term and provide closely spaced frequencycomponents representation this modified version success-fully removes cross term interference while maintaininghigh time-frequency resolution [106] It is demonstratedthat modified B-distribution based time-frequency distribu-tion performs better in suppressing cross terms with goodtime and frequency resolution for multicomponent signalscompared with other above-mentioned techniques [107]Recently this technique has been widely used to analyzenonstationarities related to EEG signals heart rate signalsand accelerometer data based on fetal movements [108 109]

7 Discussion and Conclusion

The use of EMG processing to assess muscle fatigue duringmanual lifting was reviewed and several fatigue indices wereintroduced Research in the area of EMG signal currentlyevolves around the sensitivity variability and repeatability ofthe fatigue indices as well as the best processing techniqueswith high efficiency and less computational complexityDespite the use of traditional methods such as time distribu-tion and frequency distribution time-frequency distributionis found to be more superior in terms of monitoring since itenables the user to observe the progress of the signal in timeand frequency This is very important as at the point whenmuscle fatigue sets in frequency compression can happenUnlike the conventional frequency distribution techniquetime-frequency analysis demonstrates the time when anychanges in frequencies occur

In recent years researchers had identified that the bilinearTFD could perform better than the linear TFD such asSTFT spectrogram and WT due to the fact that it doesnot suffer from the smearing effects cause by windowingfunction Nonetheless on account of its quadratic naturebilinear TFD suffers the cross term effectsWith the existenceof high resolution Cohen class TFD such as the modified B-distribution introduced in 2016 the cross term effects canbe removed and at the same time maintain the high time-frequency resolution

Other than the processing techniques listed in this paperthere are still several techniques such as Hilbert-Huangtransform which have been used to analyze the EMG signalsThere are also already well established techniques in otherresearch areas and are proven to be effective and have yetto be implemented in analyzing EMG signals such as the S-transform This technique may be suitable for nonstationarysignals like EMG itself due to its good nature including

linearity lossless reversibility multiple resolution good time-frequency resolution and simple algorithm In conclusion itcan be seen that there are many gaps to be filled in this areain either finding new and better fatigue indices or constantlydeveloping new and improved processing techniques

Competing Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

The authors acknowledge and thank the Control and SignalProcessing Group of Universiti Putra Malaysia and Rehabili-tation andAssistive Technology ResearchGroup ofUniversitiTeknikal Malaysia Melaka for their continuous support inthe research This research is fully funded by the Ministry ofHigher Education Malaysia (MOHE)

References

[1] S Gallagher and J R Heberger ldquoExamining the interactionof force and repetition on musculoskeletal disorder risk asystematic literature reviewrdquo Human Factors vol 55 no 1 pp108ndash124 2013

[2] E M Eatough J D Way and C-H Chang ldquoUnderstandingthe link between psychosocial work stressors and work-relatedmusculoskeletal complaintsrdquo Applied Ergonomics vol 43 no 3pp 554ndash563 2012

[3] T RWaters V Putz-Anderson A Garg and L J Fine ldquoRevisedNIOSH equation for the design and evaluation ofmanual liftingtasksrdquo Ergonomics vol 36 no 7 pp 749ndash776 1993

[4] Y Roquelaure C Ha A Leclerc et al ldquoEpidemiologic surveil-lance of upper-extremity musculoskeletal disorders in theworking populationrdquo Arthritis Care and Research vol 55 no5 pp 765ndash778 2006

[5] H Arif M S E Kosnan K Jusoff et al ldquoFinite element analysisof conceptual lumbar spine for different lifting positionrdquoWorldApplied Sciences Journal vol 21 pp 68ndash75 2013

[6] C J Colloca and R N Hinrichs ldquoThe biomechanical and clin-ical significance of the lumbar erector spinae flexion-relaxationphenomenon a review of literaturerdquo Journal of Manipulativeand Physiological Therapeutics vol 28 no 8 pp 623ndash631 2005

[7] N H Kamarudin S A Ahmad M K Hassan R M Yusoffand S Z Dawal ldquoMuscle contraction analysis during liftingtaskrdquo in Proceedings of the 3rd IEEE Conference on BiomedicalEngineering and Sciences (IECBES rsquo14) pp 452ndash457 IEEE KualaLumpur Malaysia December 2014

[8] K R Voge and J B Dingwell ldquoRelative timing of changes inmuscle fatigue and movement coordination during a repetitiveone-hand lifting taskrdquo in Proceedings of the A New Beginningfor Human Health Proceddings of the 25th Annual InternationalConference of the IEEE Engineering in Medicine and BiologySociety (EMBS rsquo03) pp 1807ndash1810 IEEE Cancun MexicoSeptember 2003

[9] TN S T Zawawi A R Abdullah E F Shair I Halim and SMSaleh ldquoEMG signal analysis of fatiguemuscle activity inmanualliftingrdquo Journal of Electrical Systems vol 11 no 3 pp 319ndash3252015

BioMed Research International 9

[10] R A Al-Ashaik M Z Ramadan K S Al-Saleh and T M Kha-laf ldquoEffect of safety shoes type lifting frequency and ambienttemperature on subjectrsquos MAWL and physiological responsesrdquoInternational Journal of Industrial Ergonomics vol 50 pp 43ndash512015

[11] K P Granata and W S Marras ldquoAn EMG-assisted model oftrunk loading during free-dynamic liftingrdquo Journal of Biome-chanics vol 28 no 11 pp 1309ndash1317 1995

[12] S H Roy P Bonato and M Knaflitz ldquoEMG assessment ofback muscle function during cyclical liftingrdquo Journal of Elec-tromyography and Kinesiology vol 8 no 4 pp 233ndash245 1998

[13] R E Seroussi andM H Pope ldquoThe relationship between trunkmuscle electromyography and lifting moments in the sagittaland frontal planesrdquo Journal of Biomechanics vol 20 no 2 pp135ndash146 1987

[14] R B Graham M J Agnew and J M Stevenson ldquoEffectivenessof an on-body lifting aid at reducing low back physical demandsduring an automotive assembly task assessment of EMG res-ponse and user acceptabilityrdquo Applied Ergonomics vol 40 no5 pp 936ndash942 2009

[15] D Gagnon C Lariviere and P Loisel ldquoComparative abilityof EMG optimization and hybrid modelling approaches topredict trunk muscle forces and lumbar spine loading duringdynamic sagittal plane liftingrdquoClinical Biomechanics vol 16 no5 pp 359ndash372 2001

[16] P Dolan and M A Adams ldquoThe relationship between EMGactivity and extensor moment generation in the erector spinaemuscles during bending and lifting activitiesrdquo Journal of Biome-chanics vol 26 no 4-5 pp 513ndash522 1993

[17] P Bonato P Boissy U Della Croce and S H Roy ldquoChanges inthe surface EMG signal and the biomechanics of motion duringa repetitive lifting taskrdquo IEEE Transactions on Neural Systemsand Rehabilitation Engineering vol 10 no 1 pp 38ndash47 2002

[18] J R Potvin ldquoMechanically corrected EMG for the continuousestimation of erector spinae muscle loading during repetitiveliftingrdquo European Journal of Applied Physiology and Occupa-tional Physiology vol 74 no 1-2 pp 119ndash132 1996

[19] I Kingma and J H Van Dieen ldquoLifting over an obstacle effectsof one-handed lifting and hand support on trunk kinematicsand low back loadingrdquo Journal of Biomechanics vol 37 no 2pp 249ndash255 2004

[20] H-J Shin and J-Y Kim ldquoMeasurement of trunk muscle fatigueduring dynamic lifting and lowering as recovery time changesrdquoInternational Journal of Industrial Ergonomics vol 37 no 6 pp545ndash551 2007

[21] J Cholewicki S M McGill and R W Norman ldquoComparisonof muscle forces and joint load from an optimization and EMGassisted lumbar spine model towards development of a hybridapproachrdquo Journal of Biomechanics vol 28 no 3 pp 321ndash3311995

[22] L Straker ldquoEvidence to support using squat semi-squat andstoop techniques to lift low-lying objectsrdquo International Journalof Industrial Ergonomics vol 31 no 3 pp 149ndash160 2003

[23] C K Anderson and D B Chaffin ldquoA biomechanical evaluationof five lifting techniquesrdquo Applied Ergonomics vol 17 no 1 pp2ndash8 1986

[24] R Burgess-Limerick ldquoSquat stoop or something in betweenrdquoInternational Journal of Industrial Ergonomics vol 31 no 3 pp143ndash148 2003

[25] S M Hsiang G E Brogmus and T K Courtney ldquoLow backpain (LBP) and lifting techniquemdasha reviewrdquo International Jour-nal of Industrial Ergonomics vol 19 no 1 pp 59ndash74 1997

[26] A Merlo and I Campanini ldquoTechnical aspects of surface elec-tromyography for cliniciansrdquo The Open Rehabilitation Journalvol 3 no 1 pp 98ndash109 2010

[27] E N Kamavuako J C Rosenvang R Horup W Jensen DFarina and K B Englehart ldquoSurface versus untargeted intra-muscular EMG based classification of simultaneous and dyna-mically changing movementsrdquo IEEE Transactions on NeuralSystems and Rehabilitation Engineering vol 21 no 6 pp 992ndash998 2013

[28] L H Smith and L J Hargrove ldquoComparison of surface andintramuscular EMG pattern recognition for simultaneouswristhand motion classificationrdquo in Proceedings of the 35thAnnual International Conference of the IEEE Engineering inMedicine and Biology Society (EMBC rsquo13) pp 4223ndash4226Osaka Japan July 2013

[29] L J Hargrove K Englehart and B Hudgins ldquoA comparisonof surface and intramuscular myoelectric signal classificationrdquoIEEE Transactions on Biomedical Engineering vol 54 no 5 pp847ndash853 2007

[30] A Fratini M Cesarelli P Bifulco and M Romano ldquoRelevanceof motion artifact in electromyography recordings duringvibration treatmentrdquo Journal of Electromyography and Kinesiol-ogy vol 19 no 4 pp 710ndash718 2009

[31] C J De Luca L Donald Gilmore M Kuznetsov and S HRoy ldquoFiltering the surface EMG signal movement artifact andbaseline noise contaminationrdquo Journal of Biomechanics vol 43no 8 pp 1573ndash1579 2010

[32] H L Butler R Newell C L Hubley-Kozey and J W KozeyldquoThe interpretation of abdominal wall muscle recruitment stra-tegies change when the electrocardiogram (ECG) is removedfrom the electromyogram (EMG)rdquo Journal of Electromyographyand Kinesiology vol 19 no 2 pp e102ndashe113 2009

[33] G Lu J-S Brittain P Holland et al ldquoRemoving ECG noisefrom surface EMG signals using adaptive filteringrdquo Neuro-science Letters vol 462 no 1 pp 14ndash19 2009

[34] N W Willigenburg A Daffertshofer I Kingma and J H vanDieen ldquoRemoving ECG contamination from EMG recordingsa comparison of ICA-based and other filtering proceduresrdquoJournal of Electromyography and Kinesiology vol 22 no 3 pp485ndash493 2012

[35] M I Ibrahimy M R Ahsan and O O Khalifa ldquoDesign andoptimization of Levenberg-Marquardt based neural networkclassifier for EMG signals to identify hand motionsrdquo Measure-ment Science Review vol 13 no 3 pp 142ndash151 2013

[36] S N Sidek and A J Haja Mohideen ldquoMapping of EMG signalto hand grip force at varying wrist anglesrdquo in Proceedings ofthe 2nd IEEE-EMBS Conference on Biomedical Engineering andSciences (IECBES rsquo12) pp 648ndash653 IEEE Langkawi MalaysiaDecember 2012

[37] Y Zhan D Halliday P Jiang X Liu and J Feng ldquoDetectingtime-dependent coherence between non-stationary electro-physiological signalsmdasha combined statistical and time-freq-uency approachrdquo Journal of Neuroscience Methods vol 156 no1-2 pp 322ndash332 2006

[38] E R Ferrara and BWidrow ldquoFetal electrocardiogram enhance-ment by time-sequenced adaptive filteringrdquo IEEE Transactionson Biomedical Engineering vol 29 no 6 pp 458ndash460 1982

[39] R H Shumway ldquoTime-frequency clustering and discriminantanalysisrdquo Statistics amp Probability Letters vol 63 no 3 pp 307ndash314 2003

[40] D Korosec ldquoParametric estimation of the continuous non-stationary spectrum and its dynamics in surface EMG studiesrdquo

10 BioMed Research International

International Journal of Medical Informatics vol 58-59 pp 59ndash69 2000

[41] G Morren S Van Huffel I Helon et al ldquoEffects of non-nutritive sucking on heart rate respiration and oxygenationa model-based signal processing approachrdquo Comparative Bio-chemistry and PhysiologymdashAMolecular and Integrative Physiol-ogy vol 132 no 1 pp 97ndash106 2002

[42] R Latif E Aassif M Laaboubi A Moudden and G MazeldquoDetermination of the cut-off frequency of the anti-symmetriccircumferential waves using the time-varying autoregressiveTVARrdquo in Proceedings of the 3rd International Symposium onCommunications Control and Signal Processing (ISCCSP rsquo08)pp 357ndash361 IEEE March 2008

[43] R Latif E Aassif M Laaboubi and G Maze ldquoDeterminationof the group velocity using the time-varying autoregressiveTVARrdquo in Proceedings of the 9th International Symposium onSignal Processing and Its Applications (ISSPA rsquo07) pp 1ndash4 IEEESharjah United Arab Emirates February 2007

[44] A Al Zaman X Luo M Ferdjallah and A Khamayseh ldquoA newTVARmodeling in cascaded form for nonstationary signalsrdquo inProceedings of the IEEE International Conference on AcousticsSpeech and Signal Processing (ICASSP rsquo06) pp 353ndash356 May2006

[45] A Al Zaman M A Ahad M Ferdjallah and J J Wertsch ldquoAnew approach for muscle fatigue analysis in young adults atdifferent MVC levelsrdquo in Proceedings of the IEEE International48th Midwest Symposium on Circuits and Systems (MWSCASrsquo05) pp 499ndash502 August 2005

[46] Z G Zhang H T Liu S C Chan K D K Luk and Y HuldquoTime-dependent power spectral density estimation of surfaceelectromyography during isometric muscle contraction meth-ods and comparisonsrdquo Journal of Electromyography and Kinesi-ology vol 20 no 1 pp 89ndash101 2010

[47] AAl ZamanM Ferdjallah andAKhamayseh ldquoMuscle fatigueanalysis for healthy adults using TVAR model with instanta-neous frequency estimationrdquo in Proceedings of the 38th South-eastern Symposium on SystemTheory pp 44ndash47 March 2006

[48] K Englehart B Hudgins P A Parker and M StevensonldquoClassification of the myoelectric signal using time-frequencybased representationsrdquoMedical Engineering and Physics vol 21no 6-7 pp 431ndash438 1999

[49] N Selvaraj J Lee and K H Chon ldquoTime-varying methods forcharac-terizing nonstationary dynamics of physiological sys-temsrdquo Methods of Information in Medicine vol 49 no 5 pp435ndash442 2010

[50] D Tkach H Huang and T A Kuiken ldquoStudy of stability oftime-domain features for electromyographic pattern recogni-tionrdquo Journal of NeuroEngineering and Rehabilitation vol 7article 21 13 pages 2010

[51] S Du and M Vuskovic ldquoTemporal vs spectral approach tofeature extraction from prehensile EMG signalsrdquo in Proceedingsof the IEEE International Conference on Information Reuseand Integration (IRI rsquo04) pp 344ndash350 Las Vegas Nev USANovember 2004

[52] R M Enoka and J Duchateau ldquoMuscle fatigue what why andhow it influences muscle functionrdquo Journal of Physiology vol586 no 1 pp 11ndash23 2008

[53] W J Williams and J Jeong ldquoKernel design for reduced interfer-ence distributionsrdquo IEEE Transactions on Signal Processing vol40 no 2 pp 402ndash412 1992

[54] M B I Reaz M S Hussain and F Mohd-Yasin ldquoTechniquesof EMG signal analysis detection processing classification and

applicationsrdquo Biological Procedures Online vol 8 no 1 pp 11ndash35 2006

[55] R A Malinzak S M Colby D T Kirkendall B Yu and W EGarrett ldquoA comparison of knee joint motion patterns betweenmen and women in selected athletic tasksrdquo Clinical Biomechan-ics vol 16 no 5 pp 438ndash445 2001

[56] T I Arabadzhiev V G Dimitrov N A Dimitrova and G VDimitrov ldquoInterpretation of EMG integral or RMS and esti-mates of lsquoneuromuscular efficiencyrsquo can bemisleading in fatigu-ing contractionrdquo Journal of Electromyography and Kinesiologyvol 20 no 2 pp 223ndash232 2010

[57] C Suetta P Aagaard A Rosted et al ldquoTraining-inducedchanges in muscle CSA muscle strength EMG and rate offorce development in elderly subjects after long-term unilateraldisuserdquo Journal of Applied Physiology vol 97 no 5 pp 1954ndash1961 2004

[58] M A Oskoei andHHu ldquoSupport vectormachine-based classi-fication scheme for myoelectric control applied to upper limbrdquoIEEE Transactions on Biomedical Engineering vol 55 no 8 pp1956ndash1965 2008

[59] J J Villarejo A Frizera T F Bastos and J F Sarmiento ldquoPatternrecognition of hand movements with low density sEMG forprosthesis control purposesrdquo in Proceedings of the IEEE 13thInternational Conference onRehabilitation Robotics (ICORR rsquo13)Seattle Wash USA June 2013

[60] A Phinyomark C Limsakul and P Phukpattaranont ldquoA novelfeature extraction for robust EMG pattern recognitionrdquo Journalof Computing vol 1 no 1 pp 71ndash80 2009

[61] M Zardoshti-Kermani B C Wheeler K Badie and R MHashemi ldquoEMG Feature Evaluation for Movement Control ofUpper Extremity Prosthesesrdquo IEEE Transactions on Rehabilita-tion Engineering vol 3 no 4 pp 324ndash333 1995

[62] D Tkach H Huang and T A Kuiken ldquoStudy of stability oftime-domain features for electromyographic pattern recogni-tionrdquo Journal of NeuroEngineering and Rehabilitation vol 7 no1 article no 21 2010

[63] K Kiguchi S Kariya K Watanabe K Izumi and T FukudaldquoAn exoskeletal robot for human elbowmotion supportmdashsensorfusion adaptation and controlrdquo IEEE Transactions on SystemsMan and Cybernetics Part B Cybernetics vol 31 no 3 pp 353ndash361 2001

[64] F Al Omari J Hui C Mei and G Liu ldquoPattern recognition ofeight hand motions using feature extraction of forearm EMGsignalrdquo Proceedings of the National Academy of Sciences IndiaSection AmdashPhysical Sciences vol 84 no 3 pp 473ndash480 2014

[65] A R Siddiqi S N Sidek andAKhorshidtalab ldquoSignal process-ing of EMG signal for continuous thumb-angle estimationrdquo inProceedings of the 41st Annual Conference of the IEEE IndustrialElectronics Society (IECON rsquo15) pp 374ndash379 Yokohama JapanNovember 2015

[66] A Kilbom G M Hagg and C Kall ldquoOne-handed loadcarryingmdashcardiovascular muscular and subjective indices ofendurance and fatiguerdquo European Journal of Applied Physiologyand Occupational Physiology vol 65 no 1 pp 52ndash58 1992

[67] F AlOmari and G Liu ldquoAnalysis of extracted forearm sEMGsignal using LDA QDA K-NN classification algorithmsrdquoOpenAutomation and Control Systems Journal vol 6 no 1 pp 108ndash116 2014

[68] A Phinyomark G Chujit P Phukpattaranont C Limsakuland H Hu ldquoA preliminary study assessing time-domain EMGfeatures of classifying exercises in preventing falls in the elderlyrdquoin Proceedings of the 9th International Conference on Electri-cal EngineeringElectronics Computer Telecommunications and

BioMed Research International 11

Information Technology (ECTI-CON rsquo12) pp 1ndash4 IEEE Phetch-aburi Thailand May 2012

[69] D R Rogers and D T MacIsaac ldquoEMG-based muscle fatigueassessment during dynamic contractions using principal com-ponent analysisrdquo Journal of Electromyography and Kinesiologyvol 21 no 5 pp 811ndash818 2011

[70] K Buchenrieder ldquoDimensionality reduction for the controlof powered upper limb prosthesesrdquo in Proceedings of the 14thAnnual IEEE International Conference and Workshops on theEngineering of Computer-Based Systems (ECBS rsquo07) pp 327ndash333IEEE Tucson Ariz USA March 2007

[71] G Venugopal M Navaneethakrishna and S RamakrishnanldquoExtraction and analysis of multiple time window featuresassociated with muscle fatigue conditions using sEMG signalsrdquoExpert Systems with Applications vol 41 no 6 pp 2652ndash26592014

[72] A Phinyomark C Limsakul and P Phukpattaranont ldquoEMGfeature extraction for tolerance of white Gaussian noiserdquo inProceedings of the International Workshop and Symposium onScience and Technology (I-SEEC rsquo08) pp 178ndash183 December2008

[73] M S Al-Quraishi A J Ishak S A Ahmad and M K HasanldquoImpact of feature extraction techniques on classification accu-racy for EMG based ankle joint movementsrdquo in Proceedings ofthe 10th Asian Control Conference (ASCC rsquo15) pp 1ndash5 SabahMalaysia June 2015

[74] G-C Chang W-J Kang L Jer-Junn et al ldquoReal-time imple-mentation of electromyogram pattern recognition as a controlcommand of man-machine interfacerdquoMedical Engineering andPhysics vol 18 no 7 pp 529ndash537 1996

[75] A Georgakis L K Stergioulas and G Giakas ldquoFatigue analysisof the surface EMG signal in isometric constant force con-tractions using the averaged instantaneous frequencyrdquo IEEETransactions on Biomedical Engineering vol 50 no 2 pp 262ndash265 2003

[76] R Merletti and L R Lo Conte ldquoSurface EMG signal processingduring isometric contractionsrdquo Journal of Electromyography andKinesiology vol 7 no 4 pp 241ndash250 1997

[77] C J De Luca ldquoUse of the surface EMG signal for performanceevaluation of backmusclesrdquoMuscle and Nerve vol 16 no 2 pp210ndash216 1993

[78] F Khanam and M Ahmad ldquoFrequency based EMG powerspectrum analysis of Salat associated muscle contractionrdquo inProceedings of the 1st International Conference on Electricaland Electronic Engineering (ICEEE rsquo15) Rajshahi BangladeshNovember 2015

[79] M M Altaf B M Elbagoury F Alraddady and M RoushdyldquoExtended case-based behavior control for multi-humanoidrobotsrdquo International Journal of Humanoid Robotics vol 13 no2 2016

[80] A Phinyomark A Nuidod P Phukpattaranont and C Lim-sakul ldquoFeature extraction and reduction of wavelet transformcoefficients for EMG pattern classificationrdquo Elektronika ir Elek-trotechnika vol 122 no 6 pp 27ndash32 2012

[81] M Gonzalez-Izal A Malanda I Navarro-Amezqueta et alldquoEMG spectral indices and muscle power fatigue duringdynamic contractionsrdquo Journal of Electromyography and Kine-siology vol 20 no 2 pp 233ndash240 2010

[82] G V Dimitrov T I Arabadzhiev K N Mileva J L Bowtell NCrichton andNADimitrova ldquoMuscle fatigue during dynamiccontractions assessed by new spectral indicesrdquo Medicine andScience in Sports and Exercise vol 38 no 11 pp 1971ndash1979 2006

[83] MVitor-Costa H Bortolotti T V Camata et al ldquoEMG spectralanalysis of incremental exercise in cyclists and non-cyclistsusing fourier and wavelet transformsrdquo Revista Brasileira deCineantropometria e Desempenho Humano vol 14 no 6 pp660ndash670 2012

[84] M Wacker and H Witte ldquoTime-frequency techniques in bio-medical signal analysis a tutorial review of similarities anddifferencesrdquoMethods of Information in Medicine vol 52 no 4pp 279ndash296 2013

[85] S Karlsson J Yu and M Akay ldquoTime-frequency analysis ofmyoelectric signals during dynamic contractions a compara-tive studyrdquo IEEE Transactions on Biomedical Engineering vol47 no 2 pp 228ndash238 2000

[86] A L Martinez-Herrera L M Ledesma-Carrillo M Lopez-Ramirez S Salazar-Colores E Cabal-Yepez and A Garcia-Perez ldquoGabor and theWigner-Ville transforms for broken rotorbars detection in induction motorsrdquo in Proceedings of the 24thInternational Conference on Electronics Communications andComputers (CONIELECOMP rsquo14) pp 83ndash87 IEEE CholulaMexico February 2014

[87] D MacIsaac P A Parker and R N Scott ldquoThe short-timeFourier transform and muscle fatigue assessment in dynamiccontractionsrdquo Journal of Electromyography and Kinesiology vol11 no 6 pp 439ndash449 2001

[88] E F Shair A R Abdullah T N S Tengku Zawawi S AAhmad and SMohamad Saleh ldquoAuto-segmentation analysis ofEMG signal for lifting muscle contraction activitiesrdquo Journal ofTelecommunication Electronic and Computer Engineering vol8 no 7 pp 17ndash22 2016

[89] MM Jankovic andD B Popovic ldquoAnEMGsystem for studyingmotor control strategies and fatiguerdquo in Proceedings of the 10thSymposium on Neural Network Applications in Electrical Engi-neering (NEUREL-rsquo10) pp 15ndash18 Belgrade Serbia September2010

[90] T N S T Zawawi A R Abdullah I Halim E F Shair andS M Salleh ldquoApplication of spectrogram in analysing elec-tromyography (EMG) signals of manual liftingrdquo ARPN Journalof Engineering andApplied Sciences vol 11 no 6 pp 3603ndash36092016

[91] M Yochum T Bakir R Lepers and S Binczak ldquoEstimationof muscular fatigue under electromyostimulation using CWTrdquoIEEE Transactions on Biomedical Engineering vol 59 no 12 pp3372ndash3378 2012

[92] J L Dantas T V Camata M A O C Brunetto A C MoraesT Abrao and L R Altimari ldquoFourier and Wavelet spectralanalysis of EMG signals in isometric and dynamic maximaleffort exerciserdquo in Proceedings of the 32nd Annual InternationalConference of the IEEEEngineering inMedicine andBiology Soci-ety (EMBC rsquo10) pp 5979ndash5982 IEEE Buenos Aires ArgentinaSeptember 2010

[93] L Ai J Nan and J Shen ldquoApplication of S-transform to drivingfatigue in EEG analysisrdquo in Proceedings of the IEEE InternationalConference on Intelligent Computing and Intelligent Systems(ICIS rsquo10) pp 814ndash817 IEEE Guilin China October 2010

[94] R Upadhyay P K Padhy and P K Kankar ldquoEEG artifactremoval and noise suppression by discrete orthonormal S-transform denoisingrdquo Computers and Electrical Engineeringvol 53 pp 125ndash142 2016

[95] C Dora and P K Biswal ldquoRobust ECG artifact removalfrom EEG using continuous wavelet transformation and linearregressionrdquo in 2016 International Conference on Signal Process-ing and Communications (SPCOM) pp 1ndash5 Bangalore IndiaJune 2016

12 BioMed Research International

[96] S Assous and B Boashash ldquoEvaluation of themodified S-trans-form for time-frequency synchrony analysis and source locali-sationrdquo Eurasip Journal on Advances in Signal Processing vol2012 article 49 2012

[97] A L Ricamato R G Absher M T Moffroid and J P Tra-nowski ldquoA time-frequency approach to evaluate electromyo-graphic recordingsrdquo in Proceedings of the 5th Annual IEEESymposium on Computer-Based Medical Systems pp 520ndash527IEEE Durham NC USA 1992

[98] M Khezri and M Jahed ldquoAn inventive quadratic time-freq-uency scheme based on Wigner-Ville distribution for classifi-cation of sEMG signalsrdquo in Proceedings of the 6th InternationalSpecial Topic Conference on Information TechnologyApplicationsin Biomedicine pp 261ndash264 IEEE Tokyo Japan November2007

[99] A Subasi and M K Kiymik ldquoMuscle fatigue detection in EMGusing time-frequency methods ICA and neural networksrdquoJournal of Medical Systems vol 34 no 4 pp 777ndash785 2010

[100] W Martin and P Flandrin ldquoWigner-Ville Spectral Analysisof Nonstationary Processesrdquo IEEE Transactions on AcousticsSpeech and Signal Processing vol 33 no 6 pp 1461ndash1470 1985

[101] H-I Choi and W J Williams ldquoImproved time-frequency rep-resentation of multicomponent signals using exponential ker-nelsrdquo IEEE Transactions on Acoustics Speech and Signal Pro-cessing vol 37 no 6 pp 862ndash871 1989

[102] G R Pereira L F de Oliveira and J Nadal ldquoReducing crossterms effects in the Choi-Williams transform of mioelectricsignalsrdquo Computer Methods and Programs in Biomedicine vol111 no 3 pp 685ndash692 2013

[103] M Alemu D K Kumar and A Bradley ldquoTime-frequency anal-ysis of SEMGmdashwith special consideration to the interelectrodespacingrdquo IEEE Transactions on Neural Systems and Rehabilita-tion Engineering vol 11 no 4 pp 341ndash345 2003

[104] P A Karthick G Venugopal and S Ramakrishnan ldquoAnalysis ofsurface EMG signals under fatigue and non-fatigue conditionsusing B-distribution based quadratic time frequency distribu-tionrdquo Journal of Mechanics in Medicine and Biology vol 15 no2 Article ID 1540028 2015

[105] V Sucic B Barkat and B Boashash ldquoPerformance evaluationof the B-distributionrdquo in Proceedings of the 5th InternationalSymposium on Signal Processing and Its Applications (ISSPA rsquo99)pp 267ndash270 Queensland Australia August 1999

[106] PAKarthick and S Ramakrishnan ldquoSurface electromyographybasedmuscle fatigue progression analysis usingmodified B dis-tribution time-frequency featuresrdquo Biomedical Signal Processingand Control vol 26 pp 42ndash51 2016

[107] B Boashash and V Sucic ldquoResolution measure criteria for theobjective assessment of the performance of quadratic time-frequency distributionsrdquo IEEE Transactions on Signal Process-ing vol 51 no 5 pp 1253ndash1263 2003

[108] S Dong G Azemi and B Boashash ldquoImproved characteriza-tion of HRV signals based on instantaneous frequency featuresestimated from quadratic time-frequency distributions withdata-adapted kernelsrdquoBiomedical Signal Processing andControlvol 10 no 1 pp 153ndash165 2014

[109] B Boashash and T Ben-Jabeur ldquoDesign of a high-resolutionseparable-kernel quadratic TFD for improving newborn healthoutcomes using fetal movement detectionrdquo in Proceedings ofthe 11th International Conference on Information Science SignalProcessing and their Applications (ISSPA rsquo12) pp 354ndash359Quebec Canada July 2012

Submit your manuscripts athttpswwwhindawicom

Stem CellsInternational

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MEDIATORSINFLAMMATION

of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Behavioural Neurology

EndocrinologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Disease Markers

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

BioMed Research International

OncologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Oxidative Medicine and Cellular Longevity

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

PPAR Research

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Immunology ResearchHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

ObesityJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Computational and Mathematical Methods in Medicine

OphthalmologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Diabetes ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Research and TreatmentAIDS

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Gastroenterology Research and Practice

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Parkinsonrsquos Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttpwwwhindawicom

Page 6: EMG Processing Based Measures of Fatigue Assessment during …downloads.hindawi.com/journals/bmri/2017/3937254.pdf · 2019-07-30 · BioMedResearchInternational 3 5. Electromyography

6 BioMed Research International

signal (with a phase factor) which will provide excellent timelocalization but no frequency localization

Research byMacIsaac et al usesMNFas fatigue index andhas drawn three important conclusions First the nonstation-arities in EMGdo not seem to affect theMNF values obtainedfrom Fourier transform Secondly the STFT method tomeasure MNF has successfully midpoints out the impacts ofnonstationarities in dynamic compressions Third the STFTis capable of identifying a descending pattern with fatigue indynamic contraction as long as the range of motion remainsconsistent across sequential time interval [87]

Since the STFT is basic in idea and execution and isequipped for distinguishing a pattern in muscle fatigue itis an undeniable choice for applications that require simpleanalysis with acceptable results

Spectrogram The squared magnitude of the STFT is calledspectrogram It can be expressed as

119878119909 (119905 119891) =10038161003816100381610038161003816100381610038161003816intinfin

minusinfin119909 (120591)119908 (120591 minus 119905) 119890minus2120587119891120591

100381610038161003816100381610038161003816100381610038162

119889120591 (3)

where 119909(120591) is the EMG signal 119908(120591 minus 119905) is the observationwindow and 119905 is the variable that slides the window over thesignal 119909(120591)

Spectrogram can be used to obtain the power distributionand energy distribution of the signal along the frequencydirection at a given time In EMG processing the instan-taneous energy is useful to separate muscle activation fromthe baseline which is called the segmentation process Seg-mentation is important in reducing the high computationalcomplexity of TFDs The muscle activation segmentationprocess by using spectrogram has been proposed by Shair etal and resulted in a mean absolute percentage error (MAPE)of just 1404 [88]

For both STFT and spectrogram there is a compromisebetween the time-based and frequency-based perspective ofa signal Both time and frequency are represented in limitedprecision where precision is controlled by the span of thewindow and the size of the window chosen will be the samefor all frequencies

In 2010 Jankovic and Popovic have presented in theirpaper the use of spectrogram in the analysis ofmuscle fatigueThey highlighted that even though the traditional MDFtechnique provides good sensitivity for fatigue indicationhowever there is a significant slope error particularly for lowactivity of coinitiated muscles They proposed a hybrid ofalternative methods (spectrogram and scalogram) that cansuccessfully provide complete information but at the sametime produce less error prediction for low-level coinitiatedmuscles [89] As a recommendation for improvement theyalso proposed further investigation on the robustness of thismethod in dynamic exercise contractions

Later in 2016 Zawawi et al came out with a detailedanalysis of EMG signals by using spectrogram for manuallifting application By taking the average instantaneous RMSvoltage (119881rms(119905)) as fatigue index she concludes that asthe lifting height is increased the average 119881rms(119905) is alsoincreased However as the average 119881rms(119905) decreased thenumber of lifting repetitions is increased [90]

Wavelet Transform (WT) A wavelet is a waveform of effec-tively limited duration that has an average value of zero Someof its properties are short-time localized waves with zerointegral value the possibility of time shifting and flexibilityWavelet analysis produces a time-scale view of the signalTheresult of a continuous wavelet transform (CWT) is waveletcoefficients Multiplying each coefficient with the appropri-ately scaled and shifted wavelet yields the constituent waveletof the original signal Equation (4) shows the formula forCWT

CWT119909 = intinfin

minusinfin119909 (120591) 1radic|119886|Ψ (

120591 minus 119905119886 ) 119889120591 (4)

where 119905 is the translation 119886 is the scale parameter and Ψ isthe mother wavelet

Yochum et al have proposed a new fatigue index basedon the continuous wavelet transform (CWT) named 119868CWTand have compared it to other fatigue indices from literaturein terms of their sensitivity to noiseThe results show that thisnew fatigue index quantifies the EMG signal elongation dur-ing a contraction and thus makes it a suitable fatigue index119868CWT is less subjected to noise and less truncation dependentcompared to other fatigue indices based on the frequency likeMNF and MDF [91]

A comparison has been made by Dantas et al betweenSTFT and CWT in evaluating muscle fatigue in isometricand dynamic contractions The after effects of this studyexhibit that CWT and STFT analysis give comparable fatigueestimates (slope of MDF) in isometric and dynamic contrac-tions However the aftereffects of CWT for both contractionsindicate less variability (higher accuracy) in EMG signalanalysis contrasted with STFT [92]

S-Transform Another TFD technique exists is the S-transform This technique has been widely used in diverseareas of telecommunication power quality geophysics andbiomedicine due to its obvious advantages in processingnonstationary signals [93] In the area of EEG and ECGdespite the use of S-transform in the time-frequency analysisthis technique is also a good alternative for denoising andremoval of artifacts that exist in ECG and EEG signals [94]Even though it has been explored in the areas of EEG andECG there are no available researches conducted in utilizingthis technique for EMG signal

S-transform was introduced in 1996 by StockwellMansinha and Lowe It is a hybrid of two advanced signalprocessing techniques which are STFT and WT [95] Dueto this it inherits the good qualities from both techniquesIt provides good resolution in both time and frequency andallows users to assess any frequency component in the time-frequency domain without the need of using any digital filterEven though S-transform uses a variable window length tomaintain a good time-frequency resolutions for all frequen-cies it still retains the phase information using a Fourierkernel

Expression for S-transform is shown in

ST119909 (119905 119891) = intinfin

minusinfin119909 (120591)

10038161003816100381610038161198911003816100381610038161003816radic2120587119890

minus(120591minus119905)212059122119890minus1198952120587119891120591119889120591 (5)

BioMed Research International 7

where 120591 and119891 denote the time of the spectral localization andFourier frequency respectively

The development of the S-transform is led by the urge toovercome the low resolution of STFT and the absence of thephase information in CWT Since the features had existed inthe S-transform it is sometimes viewed as a variable slidingwindow STFT or a phase-corrected CWT [96]

Although S-transform has better time-frequency resolu-tion than STFT the resolution is still far from perfect andneeds improvement To date there are several improved S-transform introduced by researchers such as the modified S-transform [96] and discrete orthonormal S-transform [94]

Bilinear TFD

Wigner-Ville Distribution (WVD) Wigner first introducedWVD in the area of quantum mechanics in the year 1932and later in 1948 Ville developed and applied the sametransformation to signal processing and spectral analysis[97]

Compared to STFT and WT WVD does not contain awindowing function and thus frees WVD from the smearingeffect due to the windowing function As a result it providesbest possible temporal and frequency resolution in the time-frequency plane It possesses several interesting propertiessuch as energy conservation real-valued marginal prop-erties translation covariance dilation covariance instanta-neous frequency and group delay

However because of its quadratic nature when dealingwith signals with several frequency components WVD suf-fers from the so-called cross components (inference terms)which represent significant defects in this method

These interference terms are troublesome since they mayoverlap with autoterms (signal terms) and in this manner itis hard to visually interpret the WVD image It creates theimpression that these terms must be available or the goodproperties of WVD (localization group delay instantaneousfrequency marginal properties etc) cannot be fulfilledThere is also a trade-off between the amount of interferencesand the number of good properties These are known issueswith the WVD spectrum and there are several ways to com-pensate these cross terms that has been discussed by previousresearchers [98ndash100] Equation (6) shows the formula forWVD

119882119909 (119905 119891) = intinfin

minusinfin119909(119905 + 1205912) 119909

lowast (120591 minus 1205912) 119890minus2119895120587119891120591119889120591 (6)

where 119909(119905 + 1205912)119909lowast(120591 minus 1205912) is the instantaneous autocorre-lation function and lowast shows conjugate operation

In the occasion that time smoothing window 119892(119905) anda frequency smoothing window ℎ(119905) are connected toWVDWVDwill then transform into the smoothed-pseudo-Wigner-Ville distribution (SPWVD) as composed in

SPWVD119909 (119905 119891)= ∬ℎ (119905 minus 120591) 119892 (119891 minus 120576)119882119909 (120591 120576) 119889120591 119889120576

(7)

where119882 is the WVD

A research by Subasi and Kiymik in 2010 has comparedthe performance between STFT SPWVD and CWT in termsof accuracy specificity and sensitivity The results suggestedthat all three methods provide almost similar performanceand are found to be satisfactory despite the fact that there arelittle differences between the results [99]

Choi-William Distribution (CWD) CWD is a member ofCohenrsquos class distribution function and was proposed in theyear 1989 by [101] It is able to avoid one of the main problemsof WVD which is the presence of interference in regionswhere one would expect zero power values Adoption of theexponential kernel in the distribution helps to suppress theeffect of cross term [102] CWD is given by

CWD119909 (119905 119891)

= intinfin

minusinfinintinfin

minusinfin119860119909 (120578 120591)Φ (120578 120591) 119890(1198952120587(120578119905minus120591119891))119889120578 119889120591

(8)

where

119860119909 (120578 120591) = intinfin

minusinfin119909(119905 + 1205912) 119909

lowast (119905 minus 1205912) 119890minus1198952120587119905120578119889119905 (9)

and the kernel function is given by

Φ(120578 120591) = 119890minus120572(120578120591)2 (10)

By studying the influence of muscle contraction towardsthe frequency content of EMG signal usingWVD and CWDAlemu et al observed that cross terms existing in WVD aregreatly reduced by CWD thus resulting in easier interpreta-tion with no loss of definition due to cross terms [103] Thereduction of the cross terms is due to the smoothing param-eters introduced in CWD However this still depends on thesmoothing parameter value selected As observed from theresult itself when the smoothing parameter decreased thereduction of the cross terms is better This however affectsthe time and frequency resolutions by reducing it and leadsto more signal loss As the smoothing parameter approachesinfin it resembles WVDmore [103]

B-Distribution Karthick et al introduced B-distribution in2015 to attenuate the cross terms exist in WVD and CWDThis is realized by introducing a smoothing function referredto as quadratic time-frequency distribution (QTFD) [104]Expression of the QTFD is given as follows [105]

119901 [119899 119896] = 2 sum|119898|lt1198722

119866 [119899119898] lowast 119911 [119899 + 119898] 119911

lowast [119899 minus 119898] 119890minus1198952120587119896119898119872(11)

where 119866[119899119898] is the smoothing kernel and operator lowastrepresents convolution

Kernel of the B-distribution is given by

119866 [119899119898] = ( |2119898|cosh2119899)

120573

lowast (sin 119888119898) (12)

where 120573 is a smoothing parameter

8 BioMed Research International

Three fatigue indices namely instantaneous median fre-quency (IMDF) instantaneousmean frequency (IMNF) andinstantaneous spectral entropy (ISPEn) were used alongsidethe B-distribution and it is found that all three features aredistinct in both fatigue and nonfatigue conditions The studyconcludes that B-distribution may be useful in EMG analysisunder various clinical and normal conditions

Modified B-Distribution A year later the same author cameout with a study using modified B-distribution to monitorthe progression of muscle fatigue Since the performance oftime-frequency distribution is assessed based on its abilityto reduce cross term and provide closely spaced frequencycomponents representation this modified version success-fully removes cross term interference while maintaininghigh time-frequency resolution [106] It is demonstratedthat modified B-distribution based time-frequency distribu-tion performs better in suppressing cross terms with goodtime and frequency resolution for multicomponent signalscompared with other above-mentioned techniques [107]Recently this technique has been widely used to analyzenonstationarities related to EEG signals heart rate signalsand accelerometer data based on fetal movements [108 109]

7 Discussion and Conclusion

The use of EMG processing to assess muscle fatigue duringmanual lifting was reviewed and several fatigue indices wereintroduced Research in the area of EMG signal currentlyevolves around the sensitivity variability and repeatability ofthe fatigue indices as well as the best processing techniqueswith high efficiency and less computational complexityDespite the use of traditional methods such as time distribu-tion and frequency distribution time-frequency distributionis found to be more superior in terms of monitoring since itenables the user to observe the progress of the signal in timeand frequency This is very important as at the point whenmuscle fatigue sets in frequency compression can happenUnlike the conventional frequency distribution techniquetime-frequency analysis demonstrates the time when anychanges in frequencies occur

In recent years researchers had identified that the bilinearTFD could perform better than the linear TFD such asSTFT spectrogram and WT due to the fact that it doesnot suffer from the smearing effects cause by windowingfunction Nonetheless on account of its quadratic naturebilinear TFD suffers the cross term effectsWith the existenceof high resolution Cohen class TFD such as the modified B-distribution introduced in 2016 the cross term effects canbe removed and at the same time maintain the high time-frequency resolution

Other than the processing techniques listed in this paperthere are still several techniques such as Hilbert-Huangtransform which have been used to analyze the EMG signalsThere are also already well established techniques in otherresearch areas and are proven to be effective and have yetto be implemented in analyzing EMG signals such as the S-transform This technique may be suitable for nonstationarysignals like EMG itself due to its good nature including

linearity lossless reversibility multiple resolution good time-frequency resolution and simple algorithm In conclusion itcan be seen that there are many gaps to be filled in this areain either finding new and better fatigue indices or constantlydeveloping new and improved processing techniques

Competing Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

The authors acknowledge and thank the Control and SignalProcessing Group of Universiti Putra Malaysia and Rehabili-tation andAssistive Technology ResearchGroup ofUniversitiTeknikal Malaysia Melaka for their continuous support inthe research This research is fully funded by the Ministry ofHigher Education Malaysia (MOHE)

References

[1] S Gallagher and J R Heberger ldquoExamining the interactionof force and repetition on musculoskeletal disorder risk asystematic literature reviewrdquo Human Factors vol 55 no 1 pp108ndash124 2013

[2] E M Eatough J D Way and C-H Chang ldquoUnderstandingthe link between psychosocial work stressors and work-relatedmusculoskeletal complaintsrdquo Applied Ergonomics vol 43 no 3pp 554ndash563 2012

[3] T RWaters V Putz-Anderson A Garg and L J Fine ldquoRevisedNIOSH equation for the design and evaluation ofmanual liftingtasksrdquo Ergonomics vol 36 no 7 pp 749ndash776 1993

[4] Y Roquelaure C Ha A Leclerc et al ldquoEpidemiologic surveil-lance of upper-extremity musculoskeletal disorders in theworking populationrdquo Arthritis Care and Research vol 55 no5 pp 765ndash778 2006

[5] H Arif M S E Kosnan K Jusoff et al ldquoFinite element analysisof conceptual lumbar spine for different lifting positionrdquoWorldApplied Sciences Journal vol 21 pp 68ndash75 2013

[6] C J Colloca and R N Hinrichs ldquoThe biomechanical and clin-ical significance of the lumbar erector spinae flexion-relaxationphenomenon a review of literaturerdquo Journal of Manipulativeand Physiological Therapeutics vol 28 no 8 pp 623ndash631 2005

[7] N H Kamarudin S A Ahmad M K Hassan R M Yusoffand S Z Dawal ldquoMuscle contraction analysis during liftingtaskrdquo in Proceedings of the 3rd IEEE Conference on BiomedicalEngineering and Sciences (IECBES rsquo14) pp 452ndash457 IEEE KualaLumpur Malaysia December 2014

[8] K R Voge and J B Dingwell ldquoRelative timing of changes inmuscle fatigue and movement coordination during a repetitiveone-hand lifting taskrdquo in Proceedings of the A New Beginningfor Human Health Proceddings of the 25th Annual InternationalConference of the IEEE Engineering in Medicine and BiologySociety (EMBS rsquo03) pp 1807ndash1810 IEEE Cancun MexicoSeptember 2003

[9] TN S T Zawawi A R Abdullah E F Shair I Halim and SMSaleh ldquoEMG signal analysis of fatiguemuscle activity inmanualliftingrdquo Journal of Electrical Systems vol 11 no 3 pp 319ndash3252015

BioMed Research International 9

[10] R A Al-Ashaik M Z Ramadan K S Al-Saleh and T M Kha-laf ldquoEffect of safety shoes type lifting frequency and ambienttemperature on subjectrsquos MAWL and physiological responsesrdquoInternational Journal of Industrial Ergonomics vol 50 pp 43ndash512015

[11] K P Granata and W S Marras ldquoAn EMG-assisted model oftrunk loading during free-dynamic liftingrdquo Journal of Biome-chanics vol 28 no 11 pp 1309ndash1317 1995

[12] S H Roy P Bonato and M Knaflitz ldquoEMG assessment ofback muscle function during cyclical liftingrdquo Journal of Elec-tromyography and Kinesiology vol 8 no 4 pp 233ndash245 1998

[13] R E Seroussi andM H Pope ldquoThe relationship between trunkmuscle electromyography and lifting moments in the sagittaland frontal planesrdquo Journal of Biomechanics vol 20 no 2 pp135ndash146 1987

[14] R B Graham M J Agnew and J M Stevenson ldquoEffectivenessof an on-body lifting aid at reducing low back physical demandsduring an automotive assembly task assessment of EMG res-ponse and user acceptabilityrdquo Applied Ergonomics vol 40 no5 pp 936ndash942 2009

[15] D Gagnon C Lariviere and P Loisel ldquoComparative abilityof EMG optimization and hybrid modelling approaches topredict trunk muscle forces and lumbar spine loading duringdynamic sagittal plane liftingrdquoClinical Biomechanics vol 16 no5 pp 359ndash372 2001

[16] P Dolan and M A Adams ldquoThe relationship between EMGactivity and extensor moment generation in the erector spinaemuscles during bending and lifting activitiesrdquo Journal of Biome-chanics vol 26 no 4-5 pp 513ndash522 1993

[17] P Bonato P Boissy U Della Croce and S H Roy ldquoChanges inthe surface EMG signal and the biomechanics of motion duringa repetitive lifting taskrdquo IEEE Transactions on Neural Systemsand Rehabilitation Engineering vol 10 no 1 pp 38ndash47 2002

[18] J R Potvin ldquoMechanically corrected EMG for the continuousestimation of erector spinae muscle loading during repetitiveliftingrdquo European Journal of Applied Physiology and Occupa-tional Physiology vol 74 no 1-2 pp 119ndash132 1996

[19] I Kingma and J H Van Dieen ldquoLifting over an obstacle effectsof one-handed lifting and hand support on trunk kinematicsand low back loadingrdquo Journal of Biomechanics vol 37 no 2pp 249ndash255 2004

[20] H-J Shin and J-Y Kim ldquoMeasurement of trunk muscle fatigueduring dynamic lifting and lowering as recovery time changesrdquoInternational Journal of Industrial Ergonomics vol 37 no 6 pp545ndash551 2007

[21] J Cholewicki S M McGill and R W Norman ldquoComparisonof muscle forces and joint load from an optimization and EMGassisted lumbar spine model towards development of a hybridapproachrdquo Journal of Biomechanics vol 28 no 3 pp 321ndash3311995

[22] L Straker ldquoEvidence to support using squat semi-squat andstoop techniques to lift low-lying objectsrdquo International Journalof Industrial Ergonomics vol 31 no 3 pp 149ndash160 2003

[23] C K Anderson and D B Chaffin ldquoA biomechanical evaluationof five lifting techniquesrdquo Applied Ergonomics vol 17 no 1 pp2ndash8 1986

[24] R Burgess-Limerick ldquoSquat stoop or something in betweenrdquoInternational Journal of Industrial Ergonomics vol 31 no 3 pp143ndash148 2003

[25] S M Hsiang G E Brogmus and T K Courtney ldquoLow backpain (LBP) and lifting techniquemdasha reviewrdquo International Jour-nal of Industrial Ergonomics vol 19 no 1 pp 59ndash74 1997

[26] A Merlo and I Campanini ldquoTechnical aspects of surface elec-tromyography for cliniciansrdquo The Open Rehabilitation Journalvol 3 no 1 pp 98ndash109 2010

[27] E N Kamavuako J C Rosenvang R Horup W Jensen DFarina and K B Englehart ldquoSurface versus untargeted intra-muscular EMG based classification of simultaneous and dyna-mically changing movementsrdquo IEEE Transactions on NeuralSystems and Rehabilitation Engineering vol 21 no 6 pp 992ndash998 2013

[28] L H Smith and L J Hargrove ldquoComparison of surface andintramuscular EMG pattern recognition for simultaneouswristhand motion classificationrdquo in Proceedings of the 35thAnnual International Conference of the IEEE Engineering inMedicine and Biology Society (EMBC rsquo13) pp 4223ndash4226Osaka Japan July 2013

[29] L J Hargrove K Englehart and B Hudgins ldquoA comparisonof surface and intramuscular myoelectric signal classificationrdquoIEEE Transactions on Biomedical Engineering vol 54 no 5 pp847ndash853 2007

[30] A Fratini M Cesarelli P Bifulco and M Romano ldquoRelevanceof motion artifact in electromyography recordings duringvibration treatmentrdquo Journal of Electromyography and Kinesiol-ogy vol 19 no 4 pp 710ndash718 2009

[31] C J De Luca L Donald Gilmore M Kuznetsov and S HRoy ldquoFiltering the surface EMG signal movement artifact andbaseline noise contaminationrdquo Journal of Biomechanics vol 43no 8 pp 1573ndash1579 2010

[32] H L Butler R Newell C L Hubley-Kozey and J W KozeyldquoThe interpretation of abdominal wall muscle recruitment stra-tegies change when the electrocardiogram (ECG) is removedfrom the electromyogram (EMG)rdquo Journal of Electromyographyand Kinesiology vol 19 no 2 pp e102ndashe113 2009

[33] G Lu J-S Brittain P Holland et al ldquoRemoving ECG noisefrom surface EMG signals using adaptive filteringrdquo Neuro-science Letters vol 462 no 1 pp 14ndash19 2009

[34] N W Willigenburg A Daffertshofer I Kingma and J H vanDieen ldquoRemoving ECG contamination from EMG recordingsa comparison of ICA-based and other filtering proceduresrdquoJournal of Electromyography and Kinesiology vol 22 no 3 pp485ndash493 2012

[35] M I Ibrahimy M R Ahsan and O O Khalifa ldquoDesign andoptimization of Levenberg-Marquardt based neural networkclassifier for EMG signals to identify hand motionsrdquo Measure-ment Science Review vol 13 no 3 pp 142ndash151 2013

[36] S N Sidek and A J Haja Mohideen ldquoMapping of EMG signalto hand grip force at varying wrist anglesrdquo in Proceedings ofthe 2nd IEEE-EMBS Conference on Biomedical Engineering andSciences (IECBES rsquo12) pp 648ndash653 IEEE Langkawi MalaysiaDecember 2012

[37] Y Zhan D Halliday P Jiang X Liu and J Feng ldquoDetectingtime-dependent coherence between non-stationary electro-physiological signalsmdasha combined statistical and time-freq-uency approachrdquo Journal of Neuroscience Methods vol 156 no1-2 pp 322ndash332 2006

[38] E R Ferrara and BWidrow ldquoFetal electrocardiogram enhance-ment by time-sequenced adaptive filteringrdquo IEEE Transactionson Biomedical Engineering vol 29 no 6 pp 458ndash460 1982

[39] R H Shumway ldquoTime-frequency clustering and discriminantanalysisrdquo Statistics amp Probability Letters vol 63 no 3 pp 307ndash314 2003

[40] D Korosec ldquoParametric estimation of the continuous non-stationary spectrum and its dynamics in surface EMG studiesrdquo

10 BioMed Research International

International Journal of Medical Informatics vol 58-59 pp 59ndash69 2000

[41] G Morren S Van Huffel I Helon et al ldquoEffects of non-nutritive sucking on heart rate respiration and oxygenationa model-based signal processing approachrdquo Comparative Bio-chemistry and PhysiologymdashAMolecular and Integrative Physiol-ogy vol 132 no 1 pp 97ndash106 2002

[42] R Latif E Aassif M Laaboubi A Moudden and G MazeldquoDetermination of the cut-off frequency of the anti-symmetriccircumferential waves using the time-varying autoregressiveTVARrdquo in Proceedings of the 3rd International Symposium onCommunications Control and Signal Processing (ISCCSP rsquo08)pp 357ndash361 IEEE March 2008

[43] R Latif E Aassif M Laaboubi and G Maze ldquoDeterminationof the group velocity using the time-varying autoregressiveTVARrdquo in Proceedings of the 9th International Symposium onSignal Processing and Its Applications (ISSPA rsquo07) pp 1ndash4 IEEESharjah United Arab Emirates February 2007

[44] A Al Zaman X Luo M Ferdjallah and A Khamayseh ldquoA newTVARmodeling in cascaded form for nonstationary signalsrdquo inProceedings of the IEEE International Conference on AcousticsSpeech and Signal Processing (ICASSP rsquo06) pp 353ndash356 May2006

[45] A Al Zaman M A Ahad M Ferdjallah and J J Wertsch ldquoAnew approach for muscle fatigue analysis in young adults atdifferent MVC levelsrdquo in Proceedings of the IEEE International48th Midwest Symposium on Circuits and Systems (MWSCASrsquo05) pp 499ndash502 August 2005

[46] Z G Zhang H T Liu S C Chan K D K Luk and Y HuldquoTime-dependent power spectral density estimation of surfaceelectromyography during isometric muscle contraction meth-ods and comparisonsrdquo Journal of Electromyography and Kinesi-ology vol 20 no 1 pp 89ndash101 2010

[47] AAl ZamanM Ferdjallah andAKhamayseh ldquoMuscle fatigueanalysis for healthy adults using TVAR model with instanta-neous frequency estimationrdquo in Proceedings of the 38th South-eastern Symposium on SystemTheory pp 44ndash47 March 2006

[48] K Englehart B Hudgins P A Parker and M StevensonldquoClassification of the myoelectric signal using time-frequencybased representationsrdquoMedical Engineering and Physics vol 21no 6-7 pp 431ndash438 1999

[49] N Selvaraj J Lee and K H Chon ldquoTime-varying methods forcharac-terizing nonstationary dynamics of physiological sys-temsrdquo Methods of Information in Medicine vol 49 no 5 pp435ndash442 2010

[50] D Tkach H Huang and T A Kuiken ldquoStudy of stability oftime-domain features for electromyographic pattern recogni-tionrdquo Journal of NeuroEngineering and Rehabilitation vol 7article 21 13 pages 2010

[51] S Du and M Vuskovic ldquoTemporal vs spectral approach tofeature extraction from prehensile EMG signalsrdquo in Proceedingsof the IEEE International Conference on Information Reuseand Integration (IRI rsquo04) pp 344ndash350 Las Vegas Nev USANovember 2004

[52] R M Enoka and J Duchateau ldquoMuscle fatigue what why andhow it influences muscle functionrdquo Journal of Physiology vol586 no 1 pp 11ndash23 2008

[53] W J Williams and J Jeong ldquoKernel design for reduced interfer-ence distributionsrdquo IEEE Transactions on Signal Processing vol40 no 2 pp 402ndash412 1992

[54] M B I Reaz M S Hussain and F Mohd-Yasin ldquoTechniquesof EMG signal analysis detection processing classification and

applicationsrdquo Biological Procedures Online vol 8 no 1 pp 11ndash35 2006

[55] R A Malinzak S M Colby D T Kirkendall B Yu and W EGarrett ldquoA comparison of knee joint motion patterns betweenmen and women in selected athletic tasksrdquo Clinical Biomechan-ics vol 16 no 5 pp 438ndash445 2001

[56] T I Arabadzhiev V G Dimitrov N A Dimitrova and G VDimitrov ldquoInterpretation of EMG integral or RMS and esti-mates of lsquoneuromuscular efficiencyrsquo can bemisleading in fatigu-ing contractionrdquo Journal of Electromyography and Kinesiologyvol 20 no 2 pp 223ndash232 2010

[57] C Suetta P Aagaard A Rosted et al ldquoTraining-inducedchanges in muscle CSA muscle strength EMG and rate offorce development in elderly subjects after long-term unilateraldisuserdquo Journal of Applied Physiology vol 97 no 5 pp 1954ndash1961 2004

[58] M A Oskoei andHHu ldquoSupport vectormachine-based classi-fication scheme for myoelectric control applied to upper limbrdquoIEEE Transactions on Biomedical Engineering vol 55 no 8 pp1956ndash1965 2008

[59] J J Villarejo A Frizera T F Bastos and J F Sarmiento ldquoPatternrecognition of hand movements with low density sEMG forprosthesis control purposesrdquo in Proceedings of the IEEE 13thInternational Conference onRehabilitation Robotics (ICORR rsquo13)Seattle Wash USA June 2013

[60] A Phinyomark C Limsakul and P Phukpattaranont ldquoA novelfeature extraction for robust EMG pattern recognitionrdquo Journalof Computing vol 1 no 1 pp 71ndash80 2009

[61] M Zardoshti-Kermani B C Wheeler K Badie and R MHashemi ldquoEMG Feature Evaluation for Movement Control ofUpper Extremity Prosthesesrdquo IEEE Transactions on Rehabilita-tion Engineering vol 3 no 4 pp 324ndash333 1995

[62] D Tkach H Huang and T A Kuiken ldquoStudy of stability oftime-domain features for electromyographic pattern recogni-tionrdquo Journal of NeuroEngineering and Rehabilitation vol 7 no1 article no 21 2010

[63] K Kiguchi S Kariya K Watanabe K Izumi and T FukudaldquoAn exoskeletal robot for human elbowmotion supportmdashsensorfusion adaptation and controlrdquo IEEE Transactions on SystemsMan and Cybernetics Part B Cybernetics vol 31 no 3 pp 353ndash361 2001

[64] F Al Omari J Hui C Mei and G Liu ldquoPattern recognition ofeight hand motions using feature extraction of forearm EMGsignalrdquo Proceedings of the National Academy of Sciences IndiaSection AmdashPhysical Sciences vol 84 no 3 pp 473ndash480 2014

[65] A R Siddiqi S N Sidek andAKhorshidtalab ldquoSignal process-ing of EMG signal for continuous thumb-angle estimationrdquo inProceedings of the 41st Annual Conference of the IEEE IndustrialElectronics Society (IECON rsquo15) pp 374ndash379 Yokohama JapanNovember 2015

[66] A Kilbom G M Hagg and C Kall ldquoOne-handed loadcarryingmdashcardiovascular muscular and subjective indices ofendurance and fatiguerdquo European Journal of Applied Physiologyand Occupational Physiology vol 65 no 1 pp 52ndash58 1992

[67] F AlOmari and G Liu ldquoAnalysis of extracted forearm sEMGsignal using LDA QDA K-NN classification algorithmsrdquoOpenAutomation and Control Systems Journal vol 6 no 1 pp 108ndash116 2014

[68] A Phinyomark G Chujit P Phukpattaranont C Limsakuland H Hu ldquoA preliminary study assessing time-domain EMGfeatures of classifying exercises in preventing falls in the elderlyrdquoin Proceedings of the 9th International Conference on Electri-cal EngineeringElectronics Computer Telecommunications and

BioMed Research International 11

Information Technology (ECTI-CON rsquo12) pp 1ndash4 IEEE Phetch-aburi Thailand May 2012

[69] D R Rogers and D T MacIsaac ldquoEMG-based muscle fatigueassessment during dynamic contractions using principal com-ponent analysisrdquo Journal of Electromyography and Kinesiologyvol 21 no 5 pp 811ndash818 2011

[70] K Buchenrieder ldquoDimensionality reduction for the controlof powered upper limb prosthesesrdquo in Proceedings of the 14thAnnual IEEE International Conference and Workshops on theEngineering of Computer-Based Systems (ECBS rsquo07) pp 327ndash333IEEE Tucson Ariz USA March 2007

[71] G Venugopal M Navaneethakrishna and S RamakrishnanldquoExtraction and analysis of multiple time window featuresassociated with muscle fatigue conditions using sEMG signalsrdquoExpert Systems with Applications vol 41 no 6 pp 2652ndash26592014

[72] A Phinyomark C Limsakul and P Phukpattaranont ldquoEMGfeature extraction for tolerance of white Gaussian noiserdquo inProceedings of the International Workshop and Symposium onScience and Technology (I-SEEC rsquo08) pp 178ndash183 December2008

[73] M S Al-Quraishi A J Ishak S A Ahmad and M K HasanldquoImpact of feature extraction techniques on classification accu-racy for EMG based ankle joint movementsrdquo in Proceedings ofthe 10th Asian Control Conference (ASCC rsquo15) pp 1ndash5 SabahMalaysia June 2015

[74] G-C Chang W-J Kang L Jer-Junn et al ldquoReal-time imple-mentation of electromyogram pattern recognition as a controlcommand of man-machine interfacerdquoMedical Engineering andPhysics vol 18 no 7 pp 529ndash537 1996

[75] A Georgakis L K Stergioulas and G Giakas ldquoFatigue analysisof the surface EMG signal in isometric constant force con-tractions using the averaged instantaneous frequencyrdquo IEEETransactions on Biomedical Engineering vol 50 no 2 pp 262ndash265 2003

[76] R Merletti and L R Lo Conte ldquoSurface EMG signal processingduring isometric contractionsrdquo Journal of Electromyography andKinesiology vol 7 no 4 pp 241ndash250 1997

[77] C J De Luca ldquoUse of the surface EMG signal for performanceevaluation of backmusclesrdquoMuscle and Nerve vol 16 no 2 pp210ndash216 1993

[78] F Khanam and M Ahmad ldquoFrequency based EMG powerspectrum analysis of Salat associated muscle contractionrdquo inProceedings of the 1st International Conference on Electricaland Electronic Engineering (ICEEE rsquo15) Rajshahi BangladeshNovember 2015

[79] M M Altaf B M Elbagoury F Alraddady and M RoushdyldquoExtended case-based behavior control for multi-humanoidrobotsrdquo International Journal of Humanoid Robotics vol 13 no2 2016

[80] A Phinyomark A Nuidod P Phukpattaranont and C Lim-sakul ldquoFeature extraction and reduction of wavelet transformcoefficients for EMG pattern classificationrdquo Elektronika ir Elek-trotechnika vol 122 no 6 pp 27ndash32 2012

[81] M Gonzalez-Izal A Malanda I Navarro-Amezqueta et alldquoEMG spectral indices and muscle power fatigue duringdynamic contractionsrdquo Journal of Electromyography and Kine-siology vol 20 no 2 pp 233ndash240 2010

[82] G V Dimitrov T I Arabadzhiev K N Mileva J L Bowtell NCrichton andNADimitrova ldquoMuscle fatigue during dynamiccontractions assessed by new spectral indicesrdquo Medicine andScience in Sports and Exercise vol 38 no 11 pp 1971ndash1979 2006

[83] MVitor-Costa H Bortolotti T V Camata et al ldquoEMG spectralanalysis of incremental exercise in cyclists and non-cyclistsusing fourier and wavelet transformsrdquo Revista Brasileira deCineantropometria e Desempenho Humano vol 14 no 6 pp660ndash670 2012

[84] M Wacker and H Witte ldquoTime-frequency techniques in bio-medical signal analysis a tutorial review of similarities anddifferencesrdquoMethods of Information in Medicine vol 52 no 4pp 279ndash296 2013

[85] S Karlsson J Yu and M Akay ldquoTime-frequency analysis ofmyoelectric signals during dynamic contractions a compara-tive studyrdquo IEEE Transactions on Biomedical Engineering vol47 no 2 pp 228ndash238 2000

[86] A L Martinez-Herrera L M Ledesma-Carrillo M Lopez-Ramirez S Salazar-Colores E Cabal-Yepez and A Garcia-Perez ldquoGabor and theWigner-Ville transforms for broken rotorbars detection in induction motorsrdquo in Proceedings of the 24thInternational Conference on Electronics Communications andComputers (CONIELECOMP rsquo14) pp 83ndash87 IEEE CholulaMexico February 2014

[87] D MacIsaac P A Parker and R N Scott ldquoThe short-timeFourier transform and muscle fatigue assessment in dynamiccontractionsrdquo Journal of Electromyography and Kinesiology vol11 no 6 pp 439ndash449 2001

[88] E F Shair A R Abdullah T N S Tengku Zawawi S AAhmad and SMohamad Saleh ldquoAuto-segmentation analysis ofEMG signal for lifting muscle contraction activitiesrdquo Journal ofTelecommunication Electronic and Computer Engineering vol8 no 7 pp 17ndash22 2016

[89] MM Jankovic andD B Popovic ldquoAnEMGsystem for studyingmotor control strategies and fatiguerdquo in Proceedings of the 10thSymposium on Neural Network Applications in Electrical Engi-neering (NEUREL-rsquo10) pp 15ndash18 Belgrade Serbia September2010

[90] T N S T Zawawi A R Abdullah I Halim E F Shair andS M Salleh ldquoApplication of spectrogram in analysing elec-tromyography (EMG) signals of manual liftingrdquo ARPN Journalof Engineering andApplied Sciences vol 11 no 6 pp 3603ndash36092016

[91] M Yochum T Bakir R Lepers and S Binczak ldquoEstimationof muscular fatigue under electromyostimulation using CWTrdquoIEEE Transactions on Biomedical Engineering vol 59 no 12 pp3372ndash3378 2012

[92] J L Dantas T V Camata M A O C Brunetto A C MoraesT Abrao and L R Altimari ldquoFourier and Wavelet spectralanalysis of EMG signals in isometric and dynamic maximaleffort exerciserdquo in Proceedings of the 32nd Annual InternationalConference of the IEEEEngineering inMedicine andBiology Soci-ety (EMBC rsquo10) pp 5979ndash5982 IEEE Buenos Aires ArgentinaSeptember 2010

[93] L Ai J Nan and J Shen ldquoApplication of S-transform to drivingfatigue in EEG analysisrdquo in Proceedings of the IEEE InternationalConference on Intelligent Computing and Intelligent Systems(ICIS rsquo10) pp 814ndash817 IEEE Guilin China October 2010

[94] R Upadhyay P K Padhy and P K Kankar ldquoEEG artifactremoval and noise suppression by discrete orthonormal S-transform denoisingrdquo Computers and Electrical Engineeringvol 53 pp 125ndash142 2016

[95] C Dora and P K Biswal ldquoRobust ECG artifact removalfrom EEG using continuous wavelet transformation and linearregressionrdquo in 2016 International Conference on Signal Process-ing and Communications (SPCOM) pp 1ndash5 Bangalore IndiaJune 2016

12 BioMed Research International

[96] S Assous and B Boashash ldquoEvaluation of themodified S-trans-form for time-frequency synchrony analysis and source locali-sationrdquo Eurasip Journal on Advances in Signal Processing vol2012 article 49 2012

[97] A L Ricamato R G Absher M T Moffroid and J P Tra-nowski ldquoA time-frequency approach to evaluate electromyo-graphic recordingsrdquo in Proceedings of the 5th Annual IEEESymposium on Computer-Based Medical Systems pp 520ndash527IEEE Durham NC USA 1992

[98] M Khezri and M Jahed ldquoAn inventive quadratic time-freq-uency scheme based on Wigner-Ville distribution for classifi-cation of sEMG signalsrdquo in Proceedings of the 6th InternationalSpecial Topic Conference on Information TechnologyApplicationsin Biomedicine pp 261ndash264 IEEE Tokyo Japan November2007

[99] A Subasi and M K Kiymik ldquoMuscle fatigue detection in EMGusing time-frequency methods ICA and neural networksrdquoJournal of Medical Systems vol 34 no 4 pp 777ndash785 2010

[100] W Martin and P Flandrin ldquoWigner-Ville Spectral Analysisof Nonstationary Processesrdquo IEEE Transactions on AcousticsSpeech and Signal Processing vol 33 no 6 pp 1461ndash1470 1985

[101] H-I Choi and W J Williams ldquoImproved time-frequency rep-resentation of multicomponent signals using exponential ker-nelsrdquo IEEE Transactions on Acoustics Speech and Signal Pro-cessing vol 37 no 6 pp 862ndash871 1989

[102] G R Pereira L F de Oliveira and J Nadal ldquoReducing crossterms effects in the Choi-Williams transform of mioelectricsignalsrdquo Computer Methods and Programs in Biomedicine vol111 no 3 pp 685ndash692 2013

[103] M Alemu D K Kumar and A Bradley ldquoTime-frequency anal-ysis of SEMGmdashwith special consideration to the interelectrodespacingrdquo IEEE Transactions on Neural Systems and Rehabilita-tion Engineering vol 11 no 4 pp 341ndash345 2003

[104] P A Karthick G Venugopal and S Ramakrishnan ldquoAnalysis ofsurface EMG signals under fatigue and non-fatigue conditionsusing B-distribution based quadratic time frequency distribu-tionrdquo Journal of Mechanics in Medicine and Biology vol 15 no2 Article ID 1540028 2015

[105] V Sucic B Barkat and B Boashash ldquoPerformance evaluationof the B-distributionrdquo in Proceedings of the 5th InternationalSymposium on Signal Processing and Its Applications (ISSPA rsquo99)pp 267ndash270 Queensland Australia August 1999

[106] PAKarthick and S Ramakrishnan ldquoSurface electromyographybasedmuscle fatigue progression analysis usingmodified B dis-tribution time-frequency featuresrdquo Biomedical Signal Processingand Control vol 26 pp 42ndash51 2016

[107] B Boashash and V Sucic ldquoResolution measure criteria for theobjective assessment of the performance of quadratic time-frequency distributionsrdquo IEEE Transactions on Signal Process-ing vol 51 no 5 pp 1253ndash1263 2003

[108] S Dong G Azemi and B Boashash ldquoImproved characteriza-tion of HRV signals based on instantaneous frequency featuresestimated from quadratic time-frequency distributions withdata-adapted kernelsrdquoBiomedical Signal Processing andControlvol 10 no 1 pp 153ndash165 2014

[109] B Boashash and T Ben-Jabeur ldquoDesign of a high-resolutionseparable-kernel quadratic TFD for improving newborn healthoutcomes using fetal movement detectionrdquo in Proceedings ofthe 11th International Conference on Information Science SignalProcessing and their Applications (ISSPA rsquo12) pp 354ndash359Quebec Canada July 2012

Submit your manuscripts athttpswwwhindawicom

Stem CellsInternational

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MEDIATORSINFLAMMATION

of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Behavioural Neurology

EndocrinologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Disease Markers

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

BioMed Research International

OncologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Oxidative Medicine and Cellular Longevity

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

PPAR Research

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Immunology ResearchHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

ObesityJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Computational and Mathematical Methods in Medicine

OphthalmologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Diabetes ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Research and TreatmentAIDS

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Gastroenterology Research and Practice

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Parkinsonrsquos Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttpwwwhindawicom

Page 7: EMG Processing Based Measures of Fatigue Assessment during …downloads.hindawi.com/journals/bmri/2017/3937254.pdf · 2019-07-30 · BioMedResearchInternational 3 5. Electromyography

BioMed Research International 7

where 120591 and119891 denote the time of the spectral localization andFourier frequency respectively

The development of the S-transform is led by the urge toovercome the low resolution of STFT and the absence of thephase information in CWT Since the features had existed inthe S-transform it is sometimes viewed as a variable slidingwindow STFT or a phase-corrected CWT [96]

Although S-transform has better time-frequency resolu-tion than STFT the resolution is still far from perfect andneeds improvement To date there are several improved S-transform introduced by researchers such as the modified S-transform [96] and discrete orthonormal S-transform [94]

Bilinear TFD

Wigner-Ville Distribution (WVD) Wigner first introducedWVD in the area of quantum mechanics in the year 1932and later in 1948 Ville developed and applied the sametransformation to signal processing and spectral analysis[97]

Compared to STFT and WT WVD does not contain awindowing function and thus frees WVD from the smearingeffect due to the windowing function As a result it providesbest possible temporal and frequency resolution in the time-frequency plane It possesses several interesting propertiessuch as energy conservation real-valued marginal prop-erties translation covariance dilation covariance instanta-neous frequency and group delay

However because of its quadratic nature when dealingwith signals with several frequency components WVD suf-fers from the so-called cross components (inference terms)which represent significant defects in this method

These interference terms are troublesome since they mayoverlap with autoterms (signal terms) and in this manner itis hard to visually interpret the WVD image It creates theimpression that these terms must be available or the goodproperties of WVD (localization group delay instantaneousfrequency marginal properties etc) cannot be fulfilledThere is also a trade-off between the amount of interferencesand the number of good properties These are known issueswith the WVD spectrum and there are several ways to com-pensate these cross terms that has been discussed by previousresearchers [98ndash100] Equation (6) shows the formula forWVD

119882119909 (119905 119891) = intinfin

minusinfin119909(119905 + 1205912) 119909

lowast (120591 minus 1205912) 119890minus2119895120587119891120591119889120591 (6)

where 119909(119905 + 1205912)119909lowast(120591 minus 1205912) is the instantaneous autocorre-lation function and lowast shows conjugate operation

In the occasion that time smoothing window 119892(119905) anda frequency smoothing window ℎ(119905) are connected toWVDWVDwill then transform into the smoothed-pseudo-Wigner-Ville distribution (SPWVD) as composed in

SPWVD119909 (119905 119891)= ∬ℎ (119905 minus 120591) 119892 (119891 minus 120576)119882119909 (120591 120576) 119889120591 119889120576

(7)

where119882 is the WVD

A research by Subasi and Kiymik in 2010 has comparedthe performance between STFT SPWVD and CWT in termsof accuracy specificity and sensitivity The results suggestedthat all three methods provide almost similar performanceand are found to be satisfactory despite the fact that there arelittle differences between the results [99]

Choi-William Distribution (CWD) CWD is a member ofCohenrsquos class distribution function and was proposed in theyear 1989 by [101] It is able to avoid one of the main problemsof WVD which is the presence of interference in regionswhere one would expect zero power values Adoption of theexponential kernel in the distribution helps to suppress theeffect of cross term [102] CWD is given by

CWD119909 (119905 119891)

= intinfin

minusinfinintinfin

minusinfin119860119909 (120578 120591)Φ (120578 120591) 119890(1198952120587(120578119905minus120591119891))119889120578 119889120591

(8)

where

119860119909 (120578 120591) = intinfin

minusinfin119909(119905 + 1205912) 119909

lowast (119905 minus 1205912) 119890minus1198952120587119905120578119889119905 (9)

and the kernel function is given by

Φ(120578 120591) = 119890minus120572(120578120591)2 (10)

By studying the influence of muscle contraction towardsthe frequency content of EMG signal usingWVD and CWDAlemu et al observed that cross terms existing in WVD aregreatly reduced by CWD thus resulting in easier interpreta-tion with no loss of definition due to cross terms [103] Thereduction of the cross terms is due to the smoothing param-eters introduced in CWD However this still depends on thesmoothing parameter value selected As observed from theresult itself when the smoothing parameter decreased thereduction of the cross terms is better This however affectsthe time and frequency resolutions by reducing it and leadsto more signal loss As the smoothing parameter approachesinfin it resembles WVDmore [103]

B-Distribution Karthick et al introduced B-distribution in2015 to attenuate the cross terms exist in WVD and CWDThis is realized by introducing a smoothing function referredto as quadratic time-frequency distribution (QTFD) [104]Expression of the QTFD is given as follows [105]

119901 [119899 119896] = 2 sum|119898|lt1198722

119866 [119899119898] lowast 119911 [119899 + 119898] 119911

lowast [119899 minus 119898] 119890minus1198952120587119896119898119872(11)

where 119866[119899119898] is the smoothing kernel and operator lowastrepresents convolution

Kernel of the B-distribution is given by

119866 [119899119898] = ( |2119898|cosh2119899)

120573

lowast (sin 119888119898) (12)

where 120573 is a smoothing parameter

8 BioMed Research International

Three fatigue indices namely instantaneous median fre-quency (IMDF) instantaneousmean frequency (IMNF) andinstantaneous spectral entropy (ISPEn) were used alongsidethe B-distribution and it is found that all three features aredistinct in both fatigue and nonfatigue conditions The studyconcludes that B-distribution may be useful in EMG analysisunder various clinical and normal conditions

Modified B-Distribution A year later the same author cameout with a study using modified B-distribution to monitorthe progression of muscle fatigue Since the performance oftime-frequency distribution is assessed based on its abilityto reduce cross term and provide closely spaced frequencycomponents representation this modified version success-fully removes cross term interference while maintaininghigh time-frequency resolution [106] It is demonstratedthat modified B-distribution based time-frequency distribu-tion performs better in suppressing cross terms with goodtime and frequency resolution for multicomponent signalscompared with other above-mentioned techniques [107]Recently this technique has been widely used to analyzenonstationarities related to EEG signals heart rate signalsand accelerometer data based on fetal movements [108 109]

7 Discussion and Conclusion

The use of EMG processing to assess muscle fatigue duringmanual lifting was reviewed and several fatigue indices wereintroduced Research in the area of EMG signal currentlyevolves around the sensitivity variability and repeatability ofthe fatigue indices as well as the best processing techniqueswith high efficiency and less computational complexityDespite the use of traditional methods such as time distribu-tion and frequency distribution time-frequency distributionis found to be more superior in terms of monitoring since itenables the user to observe the progress of the signal in timeand frequency This is very important as at the point whenmuscle fatigue sets in frequency compression can happenUnlike the conventional frequency distribution techniquetime-frequency analysis demonstrates the time when anychanges in frequencies occur

In recent years researchers had identified that the bilinearTFD could perform better than the linear TFD such asSTFT spectrogram and WT due to the fact that it doesnot suffer from the smearing effects cause by windowingfunction Nonetheless on account of its quadratic naturebilinear TFD suffers the cross term effectsWith the existenceof high resolution Cohen class TFD such as the modified B-distribution introduced in 2016 the cross term effects canbe removed and at the same time maintain the high time-frequency resolution

Other than the processing techniques listed in this paperthere are still several techniques such as Hilbert-Huangtransform which have been used to analyze the EMG signalsThere are also already well established techniques in otherresearch areas and are proven to be effective and have yetto be implemented in analyzing EMG signals such as the S-transform This technique may be suitable for nonstationarysignals like EMG itself due to its good nature including

linearity lossless reversibility multiple resolution good time-frequency resolution and simple algorithm In conclusion itcan be seen that there are many gaps to be filled in this areain either finding new and better fatigue indices or constantlydeveloping new and improved processing techniques

Competing Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

The authors acknowledge and thank the Control and SignalProcessing Group of Universiti Putra Malaysia and Rehabili-tation andAssistive Technology ResearchGroup ofUniversitiTeknikal Malaysia Melaka for their continuous support inthe research This research is fully funded by the Ministry ofHigher Education Malaysia (MOHE)

References

[1] S Gallagher and J R Heberger ldquoExamining the interactionof force and repetition on musculoskeletal disorder risk asystematic literature reviewrdquo Human Factors vol 55 no 1 pp108ndash124 2013

[2] E M Eatough J D Way and C-H Chang ldquoUnderstandingthe link between psychosocial work stressors and work-relatedmusculoskeletal complaintsrdquo Applied Ergonomics vol 43 no 3pp 554ndash563 2012

[3] T RWaters V Putz-Anderson A Garg and L J Fine ldquoRevisedNIOSH equation for the design and evaluation ofmanual liftingtasksrdquo Ergonomics vol 36 no 7 pp 749ndash776 1993

[4] Y Roquelaure C Ha A Leclerc et al ldquoEpidemiologic surveil-lance of upper-extremity musculoskeletal disorders in theworking populationrdquo Arthritis Care and Research vol 55 no5 pp 765ndash778 2006

[5] H Arif M S E Kosnan K Jusoff et al ldquoFinite element analysisof conceptual lumbar spine for different lifting positionrdquoWorldApplied Sciences Journal vol 21 pp 68ndash75 2013

[6] C J Colloca and R N Hinrichs ldquoThe biomechanical and clin-ical significance of the lumbar erector spinae flexion-relaxationphenomenon a review of literaturerdquo Journal of Manipulativeand Physiological Therapeutics vol 28 no 8 pp 623ndash631 2005

[7] N H Kamarudin S A Ahmad M K Hassan R M Yusoffand S Z Dawal ldquoMuscle contraction analysis during liftingtaskrdquo in Proceedings of the 3rd IEEE Conference on BiomedicalEngineering and Sciences (IECBES rsquo14) pp 452ndash457 IEEE KualaLumpur Malaysia December 2014

[8] K R Voge and J B Dingwell ldquoRelative timing of changes inmuscle fatigue and movement coordination during a repetitiveone-hand lifting taskrdquo in Proceedings of the A New Beginningfor Human Health Proceddings of the 25th Annual InternationalConference of the IEEE Engineering in Medicine and BiologySociety (EMBS rsquo03) pp 1807ndash1810 IEEE Cancun MexicoSeptember 2003

[9] TN S T Zawawi A R Abdullah E F Shair I Halim and SMSaleh ldquoEMG signal analysis of fatiguemuscle activity inmanualliftingrdquo Journal of Electrical Systems vol 11 no 3 pp 319ndash3252015

BioMed Research International 9

[10] R A Al-Ashaik M Z Ramadan K S Al-Saleh and T M Kha-laf ldquoEffect of safety shoes type lifting frequency and ambienttemperature on subjectrsquos MAWL and physiological responsesrdquoInternational Journal of Industrial Ergonomics vol 50 pp 43ndash512015

[11] K P Granata and W S Marras ldquoAn EMG-assisted model oftrunk loading during free-dynamic liftingrdquo Journal of Biome-chanics vol 28 no 11 pp 1309ndash1317 1995

[12] S H Roy P Bonato and M Knaflitz ldquoEMG assessment ofback muscle function during cyclical liftingrdquo Journal of Elec-tromyography and Kinesiology vol 8 no 4 pp 233ndash245 1998

[13] R E Seroussi andM H Pope ldquoThe relationship between trunkmuscle electromyography and lifting moments in the sagittaland frontal planesrdquo Journal of Biomechanics vol 20 no 2 pp135ndash146 1987

[14] R B Graham M J Agnew and J M Stevenson ldquoEffectivenessof an on-body lifting aid at reducing low back physical demandsduring an automotive assembly task assessment of EMG res-ponse and user acceptabilityrdquo Applied Ergonomics vol 40 no5 pp 936ndash942 2009

[15] D Gagnon C Lariviere and P Loisel ldquoComparative abilityof EMG optimization and hybrid modelling approaches topredict trunk muscle forces and lumbar spine loading duringdynamic sagittal plane liftingrdquoClinical Biomechanics vol 16 no5 pp 359ndash372 2001

[16] P Dolan and M A Adams ldquoThe relationship between EMGactivity and extensor moment generation in the erector spinaemuscles during bending and lifting activitiesrdquo Journal of Biome-chanics vol 26 no 4-5 pp 513ndash522 1993

[17] P Bonato P Boissy U Della Croce and S H Roy ldquoChanges inthe surface EMG signal and the biomechanics of motion duringa repetitive lifting taskrdquo IEEE Transactions on Neural Systemsand Rehabilitation Engineering vol 10 no 1 pp 38ndash47 2002

[18] J R Potvin ldquoMechanically corrected EMG for the continuousestimation of erector spinae muscle loading during repetitiveliftingrdquo European Journal of Applied Physiology and Occupa-tional Physiology vol 74 no 1-2 pp 119ndash132 1996

[19] I Kingma and J H Van Dieen ldquoLifting over an obstacle effectsof one-handed lifting and hand support on trunk kinematicsand low back loadingrdquo Journal of Biomechanics vol 37 no 2pp 249ndash255 2004

[20] H-J Shin and J-Y Kim ldquoMeasurement of trunk muscle fatigueduring dynamic lifting and lowering as recovery time changesrdquoInternational Journal of Industrial Ergonomics vol 37 no 6 pp545ndash551 2007

[21] J Cholewicki S M McGill and R W Norman ldquoComparisonof muscle forces and joint load from an optimization and EMGassisted lumbar spine model towards development of a hybridapproachrdquo Journal of Biomechanics vol 28 no 3 pp 321ndash3311995

[22] L Straker ldquoEvidence to support using squat semi-squat andstoop techniques to lift low-lying objectsrdquo International Journalof Industrial Ergonomics vol 31 no 3 pp 149ndash160 2003

[23] C K Anderson and D B Chaffin ldquoA biomechanical evaluationof five lifting techniquesrdquo Applied Ergonomics vol 17 no 1 pp2ndash8 1986

[24] R Burgess-Limerick ldquoSquat stoop or something in betweenrdquoInternational Journal of Industrial Ergonomics vol 31 no 3 pp143ndash148 2003

[25] S M Hsiang G E Brogmus and T K Courtney ldquoLow backpain (LBP) and lifting techniquemdasha reviewrdquo International Jour-nal of Industrial Ergonomics vol 19 no 1 pp 59ndash74 1997

[26] A Merlo and I Campanini ldquoTechnical aspects of surface elec-tromyography for cliniciansrdquo The Open Rehabilitation Journalvol 3 no 1 pp 98ndash109 2010

[27] E N Kamavuako J C Rosenvang R Horup W Jensen DFarina and K B Englehart ldquoSurface versus untargeted intra-muscular EMG based classification of simultaneous and dyna-mically changing movementsrdquo IEEE Transactions on NeuralSystems and Rehabilitation Engineering vol 21 no 6 pp 992ndash998 2013

[28] L H Smith and L J Hargrove ldquoComparison of surface andintramuscular EMG pattern recognition for simultaneouswristhand motion classificationrdquo in Proceedings of the 35thAnnual International Conference of the IEEE Engineering inMedicine and Biology Society (EMBC rsquo13) pp 4223ndash4226Osaka Japan July 2013

[29] L J Hargrove K Englehart and B Hudgins ldquoA comparisonof surface and intramuscular myoelectric signal classificationrdquoIEEE Transactions on Biomedical Engineering vol 54 no 5 pp847ndash853 2007

[30] A Fratini M Cesarelli P Bifulco and M Romano ldquoRelevanceof motion artifact in electromyography recordings duringvibration treatmentrdquo Journal of Electromyography and Kinesiol-ogy vol 19 no 4 pp 710ndash718 2009

[31] C J De Luca L Donald Gilmore M Kuznetsov and S HRoy ldquoFiltering the surface EMG signal movement artifact andbaseline noise contaminationrdquo Journal of Biomechanics vol 43no 8 pp 1573ndash1579 2010

[32] H L Butler R Newell C L Hubley-Kozey and J W KozeyldquoThe interpretation of abdominal wall muscle recruitment stra-tegies change when the electrocardiogram (ECG) is removedfrom the electromyogram (EMG)rdquo Journal of Electromyographyand Kinesiology vol 19 no 2 pp e102ndashe113 2009

[33] G Lu J-S Brittain P Holland et al ldquoRemoving ECG noisefrom surface EMG signals using adaptive filteringrdquo Neuro-science Letters vol 462 no 1 pp 14ndash19 2009

[34] N W Willigenburg A Daffertshofer I Kingma and J H vanDieen ldquoRemoving ECG contamination from EMG recordingsa comparison of ICA-based and other filtering proceduresrdquoJournal of Electromyography and Kinesiology vol 22 no 3 pp485ndash493 2012

[35] M I Ibrahimy M R Ahsan and O O Khalifa ldquoDesign andoptimization of Levenberg-Marquardt based neural networkclassifier for EMG signals to identify hand motionsrdquo Measure-ment Science Review vol 13 no 3 pp 142ndash151 2013

[36] S N Sidek and A J Haja Mohideen ldquoMapping of EMG signalto hand grip force at varying wrist anglesrdquo in Proceedings ofthe 2nd IEEE-EMBS Conference on Biomedical Engineering andSciences (IECBES rsquo12) pp 648ndash653 IEEE Langkawi MalaysiaDecember 2012

[37] Y Zhan D Halliday P Jiang X Liu and J Feng ldquoDetectingtime-dependent coherence between non-stationary electro-physiological signalsmdasha combined statistical and time-freq-uency approachrdquo Journal of Neuroscience Methods vol 156 no1-2 pp 322ndash332 2006

[38] E R Ferrara and BWidrow ldquoFetal electrocardiogram enhance-ment by time-sequenced adaptive filteringrdquo IEEE Transactionson Biomedical Engineering vol 29 no 6 pp 458ndash460 1982

[39] R H Shumway ldquoTime-frequency clustering and discriminantanalysisrdquo Statistics amp Probability Letters vol 63 no 3 pp 307ndash314 2003

[40] D Korosec ldquoParametric estimation of the continuous non-stationary spectrum and its dynamics in surface EMG studiesrdquo

10 BioMed Research International

International Journal of Medical Informatics vol 58-59 pp 59ndash69 2000

[41] G Morren S Van Huffel I Helon et al ldquoEffects of non-nutritive sucking on heart rate respiration and oxygenationa model-based signal processing approachrdquo Comparative Bio-chemistry and PhysiologymdashAMolecular and Integrative Physiol-ogy vol 132 no 1 pp 97ndash106 2002

[42] R Latif E Aassif M Laaboubi A Moudden and G MazeldquoDetermination of the cut-off frequency of the anti-symmetriccircumferential waves using the time-varying autoregressiveTVARrdquo in Proceedings of the 3rd International Symposium onCommunications Control and Signal Processing (ISCCSP rsquo08)pp 357ndash361 IEEE March 2008

[43] R Latif E Aassif M Laaboubi and G Maze ldquoDeterminationof the group velocity using the time-varying autoregressiveTVARrdquo in Proceedings of the 9th International Symposium onSignal Processing and Its Applications (ISSPA rsquo07) pp 1ndash4 IEEESharjah United Arab Emirates February 2007

[44] A Al Zaman X Luo M Ferdjallah and A Khamayseh ldquoA newTVARmodeling in cascaded form for nonstationary signalsrdquo inProceedings of the IEEE International Conference on AcousticsSpeech and Signal Processing (ICASSP rsquo06) pp 353ndash356 May2006

[45] A Al Zaman M A Ahad M Ferdjallah and J J Wertsch ldquoAnew approach for muscle fatigue analysis in young adults atdifferent MVC levelsrdquo in Proceedings of the IEEE International48th Midwest Symposium on Circuits and Systems (MWSCASrsquo05) pp 499ndash502 August 2005

[46] Z G Zhang H T Liu S C Chan K D K Luk and Y HuldquoTime-dependent power spectral density estimation of surfaceelectromyography during isometric muscle contraction meth-ods and comparisonsrdquo Journal of Electromyography and Kinesi-ology vol 20 no 1 pp 89ndash101 2010

[47] AAl ZamanM Ferdjallah andAKhamayseh ldquoMuscle fatigueanalysis for healthy adults using TVAR model with instanta-neous frequency estimationrdquo in Proceedings of the 38th South-eastern Symposium on SystemTheory pp 44ndash47 March 2006

[48] K Englehart B Hudgins P A Parker and M StevensonldquoClassification of the myoelectric signal using time-frequencybased representationsrdquoMedical Engineering and Physics vol 21no 6-7 pp 431ndash438 1999

[49] N Selvaraj J Lee and K H Chon ldquoTime-varying methods forcharac-terizing nonstationary dynamics of physiological sys-temsrdquo Methods of Information in Medicine vol 49 no 5 pp435ndash442 2010

[50] D Tkach H Huang and T A Kuiken ldquoStudy of stability oftime-domain features for electromyographic pattern recogni-tionrdquo Journal of NeuroEngineering and Rehabilitation vol 7article 21 13 pages 2010

[51] S Du and M Vuskovic ldquoTemporal vs spectral approach tofeature extraction from prehensile EMG signalsrdquo in Proceedingsof the IEEE International Conference on Information Reuseand Integration (IRI rsquo04) pp 344ndash350 Las Vegas Nev USANovember 2004

[52] R M Enoka and J Duchateau ldquoMuscle fatigue what why andhow it influences muscle functionrdquo Journal of Physiology vol586 no 1 pp 11ndash23 2008

[53] W J Williams and J Jeong ldquoKernel design for reduced interfer-ence distributionsrdquo IEEE Transactions on Signal Processing vol40 no 2 pp 402ndash412 1992

[54] M B I Reaz M S Hussain and F Mohd-Yasin ldquoTechniquesof EMG signal analysis detection processing classification and

applicationsrdquo Biological Procedures Online vol 8 no 1 pp 11ndash35 2006

[55] R A Malinzak S M Colby D T Kirkendall B Yu and W EGarrett ldquoA comparison of knee joint motion patterns betweenmen and women in selected athletic tasksrdquo Clinical Biomechan-ics vol 16 no 5 pp 438ndash445 2001

[56] T I Arabadzhiev V G Dimitrov N A Dimitrova and G VDimitrov ldquoInterpretation of EMG integral or RMS and esti-mates of lsquoneuromuscular efficiencyrsquo can bemisleading in fatigu-ing contractionrdquo Journal of Electromyography and Kinesiologyvol 20 no 2 pp 223ndash232 2010

[57] C Suetta P Aagaard A Rosted et al ldquoTraining-inducedchanges in muscle CSA muscle strength EMG and rate offorce development in elderly subjects after long-term unilateraldisuserdquo Journal of Applied Physiology vol 97 no 5 pp 1954ndash1961 2004

[58] M A Oskoei andHHu ldquoSupport vectormachine-based classi-fication scheme for myoelectric control applied to upper limbrdquoIEEE Transactions on Biomedical Engineering vol 55 no 8 pp1956ndash1965 2008

[59] J J Villarejo A Frizera T F Bastos and J F Sarmiento ldquoPatternrecognition of hand movements with low density sEMG forprosthesis control purposesrdquo in Proceedings of the IEEE 13thInternational Conference onRehabilitation Robotics (ICORR rsquo13)Seattle Wash USA June 2013

[60] A Phinyomark C Limsakul and P Phukpattaranont ldquoA novelfeature extraction for robust EMG pattern recognitionrdquo Journalof Computing vol 1 no 1 pp 71ndash80 2009

[61] M Zardoshti-Kermani B C Wheeler K Badie and R MHashemi ldquoEMG Feature Evaluation for Movement Control ofUpper Extremity Prosthesesrdquo IEEE Transactions on Rehabilita-tion Engineering vol 3 no 4 pp 324ndash333 1995

[62] D Tkach H Huang and T A Kuiken ldquoStudy of stability oftime-domain features for electromyographic pattern recogni-tionrdquo Journal of NeuroEngineering and Rehabilitation vol 7 no1 article no 21 2010

[63] K Kiguchi S Kariya K Watanabe K Izumi and T FukudaldquoAn exoskeletal robot for human elbowmotion supportmdashsensorfusion adaptation and controlrdquo IEEE Transactions on SystemsMan and Cybernetics Part B Cybernetics vol 31 no 3 pp 353ndash361 2001

[64] F Al Omari J Hui C Mei and G Liu ldquoPattern recognition ofeight hand motions using feature extraction of forearm EMGsignalrdquo Proceedings of the National Academy of Sciences IndiaSection AmdashPhysical Sciences vol 84 no 3 pp 473ndash480 2014

[65] A R Siddiqi S N Sidek andAKhorshidtalab ldquoSignal process-ing of EMG signal for continuous thumb-angle estimationrdquo inProceedings of the 41st Annual Conference of the IEEE IndustrialElectronics Society (IECON rsquo15) pp 374ndash379 Yokohama JapanNovember 2015

[66] A Kilbom G M Hagg and C Kall ldquoOne-handed loadcarryingmdashcardiovascular muscular and subjective indices ofendurance and fatiguerdquo European Journal of Applied Physiologyand Occupational Physiology vol 65 no 1 pp 52ndash58 1992

[67] F AlOmari and G Liu ldquoAnalysis of extracted forearm sEMGsignal using LDA QDA K-NN classification algorithmsrdquoOpenAutomation and Control Systems Journal vol 6 no 1 pp 108ndash116 2014

[68] A Phinyomark G Chujit P Phukpattaranont C Limsakuland H Hu ldquoA preliminary study assessing time-domain EMGfeatures of classifying exercises in preventing falls in the elderlyrdquoin Proceedings of the 9th International Conference on Electri-cal EngineeringElectronics Computer Telecommunications and

BioMed Research International 11

Information Technology (ECTI-CON rsquo12) pp 1ndash4 IEEE Phetch-aburi Thailand May 2012

[69] D R Rogers and D T MacIsaac ldquoEMG-based muscle fatigueassessment during dynamic contractions using principal com-ponent analysisrdquo Journal of Electromyography and Kinesiologyvol 21 no 5 pp 811ndash818 2011

[70] K Buchenrieder ldquoDimensionality reduction for the controlof powered upper limb prosthesesrdquo in Proceedings of the 14thAnnual IEEE International Conference and Workshops on theEngineering of Computer-Based Systems (ECBS rsquo07) pp 327ndash333IEEE Tucson Ariz USA March 2007

[71] G Venugopal M Navaneethakrishna and S RamakrishnanldquoExtraction and analysis of multiple time window featuresassociated with muscle fatigue conditions using sEMG signalsrdquoExpert Systems with Applications vol 41 no 6 pp 2652ndash26592014

[72] A Phinyomark C Limsakul and P Phukpattaranont ldquoEMGfeature extraction for tolerance of white Gaussian noiserdquo inProceedings of the International Workshop and Symposium onScience and Technology (I-SEEC rsquo08) pp 178ndash183 December2008

[73] M S Al-Quraishi A J Ishak S A Ahmad and M K HasanldquoImpact of feature extraction techniques on classification accu-racy for EMG based ankle joint movementsrdquo in Proceedings ofthe 10th Asian Control Conference (ASCC rsquo15) pp 1ndash5 SabahMalaysia June 2015

[74] G-C Chang W-J Kang L Jer-Junn et al ldquoReal-time imple-mentation of electromyogram pattern recognition as a controlcommand of man-machine interfacerdquoMedical Engineering andPhysics vol 18 no 7 pp 529ndash537 1996

[75] A Georgakis L K Stergioulas and G Giakas ldquoFatigue analysisof the surface EMG signal in isometric constant force con-tractions using the averaged instantaneous frequencyrdquo IEEETransactions on Biomedical Engineering vol 50 no 2 pp 262ndash265 2003

[76] R Merletti and L R Lo Conte ldquoSurface EMG signal processingduring isometric contractionsrdquo Journal of Electromyography andKinesiology vol 7 no 4 pp 241ndash250 1997

[77] C J De Luca ldquoUse of the surface EMG signal for performanceevaluation of backmusclesrdquoMuscle and Nerve vol 16 no 2 pp210ndash216 1993

[78] F Khanam and M Ahmad ldquoFrequency based EMG powerspectrum analysis of Salat associated muscle contractionrdquo inProceedings of the 1st International Conference on Electricaland Electronic Engineering (ICEEE rsquo15) Rajshahi BangladeshNovember 2015

[79] M M Altaf B M Elbagoury F Alraddady and M RoushdyldquoExtended case-based behavior control for multi-humanoidrobotsrdquo International Journal of Humanoid Robotics vol 13 no2 2016

[80] A Phinyomark A Nuidod P Phukpattaranont and C Lim-sakul ldquoFeature extraction and reduction of wavelet transformcoefficients for EMG pattern classificationrdquo Elektronika ir Elek-trotechnika vol 122 no 6 pp 27ndash32 2012

[81] M Gonzalez-Izal A Malanda I Navarro-Amezqueta et alldquoEMG spectral indices and muscle power fatigue duringdynamic contractionsrdquo Journal of Electromyography and Kine-siology vol 20 no 2 pp 233ndash240 2010

[82] G V Dimitrov T I Arabadzhiev K N Mileva J L Bowtell NCrichton andNADimitrova ldquoMuscle fatigue during dynamiccontractions assessed by new spectral indicesrdquo Medicine andScience in Sports and Exercise vol 38 no 11 pp 1971ndash1979 2006

[83] MVitor-Costa H Bortolotti T V Camata et al ldquoEMG spectralanalysis of incremental exercise in cyclists and non-cyclistsusing fourier and wavelet transformsrdquo Revista Brasileira deCineantropometria e Desempenho Humano vol 14 no 6 pp660ndash670 2012

[84] M Wacker and H Witte ldquoTime-frequency techniques in bio-medical signal analysis a tutorial review of similarities anddifferencesrdquoMethods of Information in Medicine vol 52 no 4pp 279ndash296 2013

[85] S Karlsson J Yu and M Akay ldquoTime-frequency analysis ofmyoelectric signals during dynamic contractions a compara-tive studyrdquo IEEE Transactions on Biomedical Engineering vol47 no 2 pp 228ndash238 2000

[86] A L Martinez-Herrera L M Ledesma-Carrillo M Lopez-Ramirez S Salazar-Colores E Cabal-Yepez and A Garcia-Perez ldquoGabor and theWigner-Ville transforms for broken rotorbars detection in induction motorsrdquo in Proceedings of the 24thInternational Conference on Electronics Communications andComputers (CONIELECOMP rsquo14) pp 83ndash87 IEEE CholulaMexico February 2014

[87] D MacIsaac P A Parker and R N Scott ldquoThe short-timeFourier transform and muscle fatigue assessment in dynamiccontractionsrdquo Journal of Electromyography and Kinesiology vol11 no 6 pp 439ndash449 2001

[88] E F Shair A R Abdullah T N S Tengku Zawawi S AAhmad and SMohamad Saleh ldquoAuto-segmentation analysis ofEMG signal for lifting muscle contraction activitiesrdquo Journal ofTelecommunication Electronic and Computer Engineering vol8 no 7 pp 17ndash22 2016

[89] MM Jankovic andD B Popovic ldquoAnEMGsystem for studyingmotor control strategies and fatiguerdquo in Proceedings of the 10thSymposium on Neural Network Applications in Electrical Engi-neering (NEUREL-rsquo10) pp 15ndash18 Belgrade Serbia September2010

[90] T N S T Zawawi A R Abdullah I Halim E F Shair andS M Salleh ldquoApplication of spectrogram in analysing elec-tromyography (EMG) signals of manual liftingrdquo ARPN Journalof Engineering andApplied Sciences vol 11 no 6 pp 3603ndash36092016

[91] M Yochum T Bakir R Lepers and S Binczak ldquoEstimationof muscular fatigue under electromyostimulation using CWTrdquoIEEE Transactions on Biomedical Engineering vol 59 no 12 pp3372ndash3378 2012

[92] J L Dantas T V Camata M A O C Brunetto A C MoraesT Abrao and L R Altimari ldquoFourier and Wavelet spectralanalysis of EMG signals in isometric and dynamic maximaleffort exerciserdquo in Proceedings of the 32nd Annual InternationalConference of the IEEEEngineering inMedicine andBiology Soci-ety (EMBC rsquo10) pp 5979ndash5982 IEEE Buenos Aires ArgentinaSeptember 2010

[93] L Ai J Nan and J Shen ldquoApplication of S-transform to drivingfatigue in EEG analysisrdquo in Proceedings of the IEEE InternationalConference on Intelligent Computing and Intelligent Systems(ICIS rsquo10) pp 814ndash817 IEEE Guilin China October 2010

[94] R Upadhyay P K Padhy and P K Kankar ldquoEEG artifactremoval and noise suppression by discrete orthonormal S-transform denoisingrdquo Computers and Electrical Engineeringvol 53 pp 125ndash142 2016

[95] C Dora and P K Biswal ldquoRobust ECG artifact removalfrom EEG using continuous wavelet transformation and linearregressionrdquo in 2016 International Conference on Signal Process-ing and Communications (SPCOM) pp 1ndash5 Bangalore IndiaJune 2016

12 BioMed Research International

[96] S Assous and B Boashash ldquoEvaluation of themodified S-trans-form for time-frequency synchrony analysis and source locali-sationrdquo Eurasip Journal on Advances in Signal Processing vol2012 article 49 2012

[97] A L Ricamato R G Absher M T Moffroid and J P Tra-nowski ldquoA time-frequency approach to evaluate electromyo-graphic recordingsrdquo in Proceedings of the 5th Annual IEEESymposium on Computer-Based Medical Systems pp 520ndash527IEEE Durham NC USA 1992

[98] M Khezri and M Jahed ldquoAn inventive quadratic time-freq-uency scheme based on Wigner-Ville distribution for classifi-cation of sEMG signalsrdquo in Proceedings of the 6th InternationalSpecial Topic Conference on Information TechnologyApplicationsin Biomedicine pp 261ndash264 IEEE Tokyo Japan November2007

[99] A Subasi and M K Kiymik ldquoMuscle fatigue detection in EMGusing time-frequency methods ICA and neural networksrdquoJournal of Medical Systems vol 34 no 4 pp 777ndash785 2010

[100] W Martin and P Flandrin ldquoWigner-Ville Spectral Analysisof Nonstationary Processesrdquo IEEE Transactions on AcousticsSpeech and Signal Processing vol 33 no 6 pp 1461ndash1470 1985

[101] H-I Choi and W J Williams ldquoImproved time-frequency rep-resentation of multicomponent signals using exponential ker-nelsrdquo IEEE Transactions on Acoustics Speech and Signal Pro-cessing vol 37 no 6 pp 862ndash871 1989

[102] G R Pereira L F de Oliveira and J Nadal ldquoReducing crossterms effects in the Choi-Williams transform of mioelectricsignalsrdquo Computer Methods and Programs in Biomedicine vol111 no 3 pp 685ndash692 2013

[103] M Alemu D K Kumar and A Bradley ldquoTime-frequency anal-ysis of SEMGmdashwith special consideration to the interelectrodespacingrdquo IEEE Transactions on Neural Systems and Rehabilita-tion Engineering vol 11 no 4 pp 341ndash345 2003

[104] P A Karthick G Venugopal and S Ramakrishnan ldquoAnalysis ofsurface EMG signals under fatigue and non-fatigue conditionsusing B-distribution based quadratic time frequency distribu-tionrdquo Journal of Mechanics in Medicine and Biology vol 15 no2 Article ID 1540028 2015

[105] V Sucic B Barkat and B Boashash ldquoPerformance evaluationof the B-distributionrdquo in Proceedings of the 5th InternationalSymposium on Signal Processing and Its Applications (ISSPA rsquo99)pp 267ndash270 Queensland Australia August 1999

[106] PAKarthick and S Ramakrishnan ldquoSurface electromyographybasedmuscle fatigue progression analysis usingmodified B dis-tribution time-frequency featuresrdquo Biomedical Signal Processingand Control vol 26 pp 42ndash51 2016

[107] B Boashash and V Sucic ldquoResolution measure criteria for theobjective assessment of the performance of quadratic time-frequency distributionsrdquo IEEE Transactions on Signal Process-ing vol 51 no 5 pp 1253ndash1263 2003

[108] S Dong G Azemi and B Boashash ldquoImproved characteriza-tion of HRV signals based on instantaneous frequency featuresestimated from quadratic time-frequency distributions withdata-adapted kernelsrdquoBiomedical Signal Processing andControlvol 10 no 1 pp 153ndash165 2014

[109] B Boashash and T Ben-Jabeur ldquoDesign of a high-resolutionseparable-kernel quadratic TFD for improving newborn healthoutcomes using fetal movement detectionrdquo in Proceedings ofthe 11th International Conference on Information Science SignalProcessing and their Applications (ISSPA rsquo12) pp 354ndash359Quebec Canada July 2012

Submit your manuscripts athttpswwwhindawicom

Stem CellsInternational

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MEDIATORSINFLAMMATION

of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Behavioural Neurology

EndocrinologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Disease Markers

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

BioMed Research International

OncologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Oxidative Medicine and Cellular Longevity

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

PPAR Research

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Immunology ResearchHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

ObesityJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Computational and Mathematical Methods in Medicine

OphthalmologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Diabetes ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Research and TreatmentAIDS

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Gastroenterology Research and Practice

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Parkinsonrsquos Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttpwwwhindawicom

Page 8: EMG Processing Based Measures of Fatigue Assessment during …downloads.hindawi.com/journals/bmri/2017/3937254.pdf · 2019-07-30 · BioMedResearchInternational 3 5. Electromyography

8 BioMed Research International

Three fatigue indices namely instantaneous median fre-quency (IMDF) instantaneousmean frequency (IMNF) andinstantaneous spectral entropy (ISPEn) were used alongsidethe B-distribution and it is found that all three features aredistinct in both fatigue and nonfatigue conditions The studyconcludes that B-distribution may be useful in EMG analysisunder various clinical and normal conditions

Modified B-Distribution A year later the same author cameout with a study using modified B-distribution to monitorthe progression of muscle fatigue Since the performance oftime-frequency distribution is assessed based on its abilityto reduce cross term and provide closely spaced frequencycomponents representation this modified version success-fully removes cross term interference while maintaininghigh time-frequency resolution [106] It is demonstratedthat modified B-distribution based time-frequency distribu-tion performs better in suppressing cross terms with goodtime and frequency resolution for multicomponent signalscompared with other above-mentioned techniques [107]Recently this technique has been widely used to analyzenonstationarities related to EEG signals heart rate signalsand accelerometer data based on fetal movements [108 109]

7 Discussion and Conclusion

The use of EMG processing to assess muscle fatigue duringmanual lifting was reviewed and several fatigue indices wereintroduced Research in the area of EMG signal currentlyevolves around the sensitivity variability and repeatability ofthe fatigue indices as well as the best processing techniqueswith high efficiency and less computational complexityDespite the use of traditional methods such as time distribu-tion and frequency distribution time-frequency distributionis found to be more superior in terms of monitoring since itenables the user to observe the progress of the signal in timeand frequency This is very important as at the point whenmuscle fatigue sets in frequency compression can happenUnlike the conventional frequency distribution techniquetime-frequency analysis demonstrates the time when anychanges in frequencies occur

In recent years researchers had identified that the bilinearTFD could perform better than the linear TFD such asSTFT spectrogram and WT due to the fact that it doesnot suffer from the smearing effects cause by windowingfunction Nonetheless on account of its quadratic naturebilinear TFD suffers the cross term effectsWith the existenceof high resolution Cohen class TFD such as the modified B-distribution introduced in 2016 the cross term effects canbe removed and at the same time maintain the high time-frequency resolution

Other than the processing techniques listed in this paperthere are still several techniques such as Hilbert-Huangtransform which have been used to analyze the EMG signalsThere are also already well established techniques in otherresearch areas and are proven to be effective and have yetto be implemented in analyzing EMG signals such as the S-transform This technique may be suitable for nonstationarysignals like EMG itself due to its good nature including

linearity lossless reversibility multiple resolution good time-frequency resolution and simple algorithm In conclusion itcan be seen that there are many gaps to be filled in this areain either finding new and better fatigue indices or constantlydeveloping new and improved processing techniques

Competing Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

The authors acknowledge and thank the Control and SignalProcessing Group of Universiti Putra Malaysia and Rehabili-tation andAssistive Technology ResearchGroup ofUniversitiTeknikal Malaysia Melaka for their continuous support inthe research This research is fully funded by the Ministry ofHigher Education Malaysia (MOHE)

References

[1] S Gallagher and J R Heberger ldquoExamining the interactionof force and repetition on musculoskeletal disorder risk asystematic literature reviewrdquo Human Factors vol 55 no 1 pp108ndash124 2013

[2] E M Eatough J D Way and C-H Chang ldquoUnderstandingthe link between psychosocial work stressors and work-relatedmusculoskeletal complaintsrdquo Applied Ergonomics vol 43 no 3pp 554ndash563 2012

[3] T RWaters V Putz-Anderson A Garg and L J Fine ldquoRevisedNIOSH equation for the design and evaluation ofmanual liftingtasksrdquo Ergonomics vol 36 no 7 pp 749ndash776 1993

[4] Y Roquelaure C Ha A Leclerc et al ldquoEpidemiologic surveil-lance of upper-extremity musculoskeletal disorders in theworking populationrdquo Arthritis Care and Research vol 55 no5 pp 765ndash778 2006

[5] H Arif M S E Kosnan K Jusoff et al ldquoFinite element analysisof conceptual lumbar spine for different lifting positionrdquoWorldApplied Sciences Journal vol 21 pp 68ndash75 2013

[6] C J Colloca and R N Hinrichs ldquoThe biomechanical and clin-ical significance of the lumbar erector spinae flexion-relaxationphenomenon a review of literaturerdquo Journal of Manipulativeand Physiological Therapeutics vol 28 no 8 pp 623ndash631 2005

[7] N H Kamarudin S A Ahmad M K Hassan R M Yusoffand S Z Dawal ldquoMuscle contraction analysis during liftingtaskrdquo in Proceedings of the 3rd IEEE Conference on BiomedicalEngineering and Sciences (IECBES rsquo14) pp 452ndash457 IEEE KualaLumpur Malaysia December 2014

[8] K R Voge and J B Dingwell ldquoRelative timing of changes inmuscle fatigue and movement coordination during a repetitiveone-hand lifting taskrdquo in Proceedings of the A New Beginningfor Human Health Proceddings of the 25th Annual InternationalConference of the IEEE Engineering in Medicine and BiologySociety (EMBS rsquo03) pp 1807ndash1810 IEEE Cancun MexicoSeptember 2003

[9] TN S T Zawawi A R Abdullah E F Shair I Halim and SMSaleh ldquoEMG signal analysis of fatiguemuscle activity inmanualliftingrdquo Journal of Electrical Systems vol 11 no 3 pp 319ndash3252015

BioMed Research International 9

[10] R A Al-Ashaik M Z Ramadan K S Al-Saleh and T M Kha-laf ldquoEffect of safety shoes type lifting frequency and ambienttemperature on subjectrsquos MAWL and physiological responsesrdquoInternational Journal of Industrial Ergonomics vol 50 pp 43ndash512015

[11] K P Granata and W S Marras ldquoAn EMG-assisted model oftrunk loading during free-dynamic liftingrdquo Journal of Biome-chanics vol 28 no 11 pp 1309ndash1317 1995

[12] S H Roy P Bonato and M Knaflitz ldquoEMG assessment ofback muscle function during cyclical liftingrdquo Journal of Elec-tromyography and Kinesiology vol 8 no 4 pp 233ndash245 1998

[13] R E Seroussi andM H Pope ldquoThe relationship between trunkmuscle electromyography and lifting moments in the sagittaland frontal planesrdquo Journal of Biomechanics vol 20 no 2 pp135ndash146 1987

[14] R B Graham M J Agnew and J M Stevenson ldquoEffectivenessof an on-body lifting aid at reducing low back physical demandsduring an automotive assembly task assessment of EMG res-ponse and user acceptabilityrdquo Applied Ergonomics vol 40 no5 pp 936ndash942 2009

[15] D Gagnon C Lariviere and P Loisel ldquoComparative abilityof EMG optimization and hybrid modelling approaches topredict trunk muscle forces and lumbar spine loading duringdynamic sagittal plane liftingrdquoClinical Biomechanics vol 16 no5 pp 359ndash372 2001

[16] P Dolan and M A Adams ldquoThe relationship between EMGactivity and extensor moment generation in the erector spinaemuscles during bending and lifting activitiesrdquo Journal of Biome-chanics vol 26 no 4-5 pp 513ndash522 1993

[17] P Bonato P Boissy U Della Croce and S H Roy ldquoChanges inthe surface EMG signal and the biomechanics of motion duringa repetitive lifting taskrdquo IEEE Transactions on Neural Systemsand Rehabilitation Engineering vol 10 no 1 pp 38ndash47 2002

[18] J R Potvin ldquoMechanically corrected EMG for the continuousestimation of erector spinae muscle loading during repetitiveliftingrdquo European Journal of Applied Physiology and Occupa-tional Physiology vol 74 no 1-2 pp 119ndash132 1996

[19] I Kingma and J H Van Dieen ldquoLifting over an obstacle effectsof one-handed lifting and hand support on trunk kinematicsand low back loadingrdquo Journal of Biomechanics vol 37 no 2pp 249ndash255 2004

[20] H-J Shin and J-Y Kim ldquoMeasurement of trunk muscle fatigueduring dynamic lifting and lowering as recovery time changesrdquoInternational Journal of Industrial Ergonomics vol 37 no 6 pp545ndash551 2007

[21] J Cholewicki S M McGill and R W Norman ldquoComparisonof muscle forces and joint load from an optimization and EMGassisted lumbar spine model towards development of a hybridapproachrdquo Journal of Biomechanics vol 28 no 3 pp 321ndash3311995

[22] L Straker ldquoEvidence to support using squat semi-squat andstoop techniques to lift low-lying objectsrdquo International Journalof Industrial Ergonomics vol 31 no 3 pp 149ndash160 2003

[23] C K Anderson and D B Chaffin ldquoA biomechanical evaluationof five lifting techniquesrdquo Applied Ergonomics vol 17 no 1 pp2ndash8 1986

[24] R Burgess-Limerick ldquoSquat stoop or something in betweenrdquoInternational Journal of Industrial Ergonomics vol 31 no 3 pp143ndash148 2003

[25] S M Hsiang G E Brogmus and T K Courtney ldquoLow backpain (LBP) and lifting techniquemdasha reviewrdquo International Jour-nal of Industrial Ergonomics vol 19 no 1 pp 59ndash74 1997

[26] A Merlo and I Campanini ldquoTechnical aspects of surface elec-tromyography for cliniciansrdquo The Open Rehabilitation Journalvol 3 no 1 pp 98ndash109 2010

[27] E N Kamavuako J C Rosenvang R Horup W Jensen DFarina and K B Englehart ldquoSurface versus untargeted intra-muscular EMG based classification of simultaneous and dyna-mically changing movementsrdquo IEEE Transactions on NeuralSystems and Rehabilitation Engineering vol 21 no 6 pp 992ndash998 2013

[28] L H Smith and L J Hargrove ldquoComparison of surface andintramuscular EMG pattern recognition for simultaneouswristhand motion classificationrdquo in Proceedings of the 35thAnnual International Conference of the IEEE Engineering inMedicine and Biology Society (EMBC rsquo13) pp 4223ndash4226Osaka Japan July 2013

[29] L J Hargrove K Englehart and B Hudgins ldquoA comparisonof surface and intramuscular myoelectric signal classificationrdquoIEEE Transactions on Biomedical Engineering vol 54 no 5 pp847ndash853 2007

[30] A Fratini M Cesarelli P Bifulco and M Romano ldquoRelevanceof motion artifact in electromyography recordings duringvibration treatmentrdquo Journal of Electromyography and Kinesiol-ogy vol 19 no 4 pp 710ndash718 2009

[31] C J De Luca L Donald Gilmore M Kuznetsov and S HRoy ldquoFiltering the surface EMG signal movement artifact andbaseline noise contaminationrdquo Journal of Biomechanics vol 43no 8 pp 1573ndash1579 2010

[32] H L Butler R Newell C L Hubley-Kozey and J W KozeyldquoThe interpretation of abdominal wall muscle recruitment stra-tegies change when the electrocardiogram (ECG) is removedfrom the electromyogram (EMG)rdquo Journal of Electromyographyand Kinesiology vol 19 no 2 pp e102ndashe113 2009

[33] G Lu J-S Brittain P Holland et al ldquoRemoving ECG noisefrom surface EMG signals using adaptive filteringrdquo Neuro-science Letters vol 462 no 1 pp 14ndash19 2009

[34] N W Willigenburg A Daffertshofer I Kingma and J H vanDieen ldquoRemoving ECG contamination from EMG recordingsa comparison of ICA-based and other filtering proceduresrdquoJournal of Electromyography and Kinesiology vol 22 no 3 pp485ndash493 2012

[35] M I Ibrahimy M R Ahsan and O O Khalifa ldquoDesign andoptimization of Levenberg-Marquardt based neural networkclassifier for EMG signals to identify hand motionsrdquo Measure-ment Science Review vol 13 no 3 pp 142ndash151 2013

[36] S N Sidek and A J Haja Mohideen ldquoMapping of EMG signalto hand grip force at varying wrist anglesrdquo in Proceedings ofthe 2nd IEEE-EMBS Conference on Biomedical Engineering andSciences (IECBES rsquo12) pp 648ndash653 IEEE Langkawi MalaysiaDecember 2012

[37] Y Zhan D Halliday P Jiang X Liu and J Feng ldquoDetectingtime-dependent coherence between non-stationary electro-physiological signalsmdasha combined statistical and time-freq-uency approachrdquo Journal of Neuroscience Methods vol 156 no1-2 pp 322ndash332 2006

[38] E R Ferrara and BWidrow ldquoFetal electrocardiogram enhance-ment by time-sequenced adaptive filteringrdquo IEEE Transactionson Biomedical Engineering vol 29 no 6 pp 458ndash460 1982

[39] R H Shumway ldquoTime-frequency clustering and discriminantanalysisrdquo Statistics amp Probability Letters vol 63 no 3 pp 307ndash314 2003

[40] D Korosec ldquoParametric estimation of the continuous non-stationary spectrum and its dynamics in surface EMG studiesrdquo

10 BioMed Research International

International Journal of Medical Informatics vol 58-59 pp 59ndash69 2000

[41] G Morren S Van Huffel I Helon et al ldquoEffects of non-nutritive sucking on heart rate respiration and oxygenationa model-based signal processing approachrdquo Comparative Bio-chemistry and PhysiologymdashAMolecular and Integrative Physiol-ogy vol 132 no 1 pp 97ndash106 2002

[42] R Latif E Aassif M Laaboubi A Moudden and G MazeldquoDetermination of the cut-off frequency of the anti-symmetriccircumferential waves using the time-varying autoregressiveTVARrdquo in Proceedings of the 3rd International Symposium onCommunications Control and Signal Processing (ISCCSP rsquo08)pp 357ndash361 IEEE March 2008

[43] R Latif E Aassif M Laaboubi and G Maze ldquoDeterminationof the group velocity using the time-varying autoregressiveTVARrdquo in Proceedings of the 9th International Symposium onSignal Processing and Its Applications (ISSPA rsquo07) pp 1ndash4 IEEESharjah United Arab Emirates February 2007

[44] A Al Zaman X Luo M Ferdjallah and A Khamayseh ldquoA newTVARmodeling in cascaded form for nonstationary signalsrdquo inProceedings of the IEEE International Conference on AcousticsSpeech and Signal Processing (ICASSP rsquo06) pp 353ndash356 May2006

[45] A Al Zaman M A Ahad M Ferdjallah and J J Wertsch ldquoAnew approach for muscle fatigue analysis in young adults atdifferent MVC levelsrdquo in Proceedings of the IEEE International48th Midwest Symposium on Circuits and Systems (MWSCASrsquo05) pp 499ndash502 August 2005

[46] Z G Zhang H T Liu S C Chan K D K Luk and Y HuldquoTime-dependent power spectral density estimation of surfaceelectromyography during isometric muscle contraction meth-ods and comparisonsrdquo Journal of Electromyography and Kinesi-ology vol 20 no 1 pp 89ndash101 2010

[47] AAl ZamanM Ferdjallah andAKhamayseh ldquoMuscle fatigueanalysis for healthy adults using TVAR model with instanta-neous frequency estimationrdquo in Proceedings of the 38th South-eastern Symposium on SystemTheory pp 44ndash47 March 2006

[48] K Englehart B Hudgins P A Parker and M StevensonldquoClassification of the myoelectric signal using time-frequencybased representationsrdquoMedical Engineering and Physics vol 21no 6-7 pp 431ndash438 1999

[49] N Selvaraj J Lee and K H Chon ldquoTime-varying methods forcharac-terizing nonstationary dynamics of physiological sys-temsrdquo Methods of Information in Medicine vol 49 no 5 pp435ndash442 2010

[50] D Tkach H Huang and T A Kuiken ldquoStudy of stability oftime-domain features for electromyographic pattern recogni-tionrdquo Journal of NeuroEngineering and Rehabilitation vol 7article 21 13 pages 2010

[51] S Du and M Vuskovic ldquoTemporal vs spectral approach tofeature extraction from prehensile EMG signalsrdquo in Proceedingsof the IEEE International Conference on Information Reuseand Integration (IRI rsquo04) pp 344ndash350 Las Vegas Nev USANovember 2004

[52] R M Enoka and J Duchateau ldquoMuscle fatigue what why andhow it influences muscle functionrdquo Journal of Physiology vol586 no 1 pp 11ndash23 2008

[53] W J Williams and J Jeong ldquoKernel design for reduced interfer-ence distributionsrdquo IEEE Transactions on Signal Processing vol40 no 2 pp 402ndash412 1992

[54] M B I Reaz M S Hussain and F Mohd-Yasin ldquoTechniquesof EMG signal analysis detection processing classification and

applicationsrdquo Biological Procedures Online vol 8 no 1 pp 11ndash35 2006

[55] R A Malinzak S M Colby D T Kirkendall B Yu and W EGarrett ldquoA comparison of knee joint motion patterns betweenmen and women in selected athletic tasksrdquo Clinical Biomechan-ics vol 16 no 5 pp 438ndash445 2001

[56] T I Arabadzhiev V G Dimitrov N A Dimitrova and G VDimitrov ldquoInterpretation of EMG integral or RMS and esti-mates of lsquoneuromuscular efficiencyrsquo can bemisleading in fatigu-ing contractionrdquo Journal of Electromyography and Kinesiologyvol 20 no 2 pp 223ndash232 2010

[57] C Suetta P Aagaard A Rosted et al ldquoTraining-inducedchanges in muscle CSA muscle strength EMG and rate offorce development in elderly subjects after long-term unilateraldisuserdquo Journal of Applied Physiology vol 97 no 5 pp 1954ndash1961 2004

[58] M A Oskoei andHHu ldquoSupport vectormachine-based classi-fication scheme for myoelectric control applied to upper limbrdquoIEEE Transactions on Biomedical Engineering vol 55 no 8 pp1956ndash1965 2008

[59] J J Villarejo A Frizera T F Bastos and J F Sarmiento ldquoPatternrecognition of hand movements with low density sEMG forprosthesis control purposesrdquo in Proceedings of the IEEE 13thInternational Conference onRehabilitation Robotics (ICORR rsquo13)Seattle Wash USA June 2013

[60] A Phinyomark C Limsakul and P Phukpattaranont ldquoA novelfeature extraction for robust EMG pattern recognitionrdquo Journalof Computing vol 1 no 1 pp 71ndash80 2009

[61] M Zardoshti-Kermani B C Wheeler K Badie and R MHashemi ldquoEMG Feature Evaluation for Movement Control ofUpper Extremity Prosthesesrdquo IEEE Transactions on Rehabilita-tion Engineering vol 3 no 4 pp 324ndash333 1995

[62] D Tkach H Huang and T A Kuiken ldquoStudy of stability oftime-domain features for electromyographic pattern recogni-tionrdquo Journal of NeuroEngineering and Rehabilitation vol 7 no1 article no 21 2010

[63] K Kiguchi S Kariya K Watanabe K Izumi and T FukudaldquoAn exoskeletal robot for human elbowmotion supportmdashsensorfusion adaptation and controlrdquo IEEE Transactions on SystemsMan and Cybernetics Part B Cybernetics vol 31 no 3 pp 353ndash361 2001

[64] F Al Omari J Hui C Mei and G Liu ldquoPattern recognition ofeight hand motions using feature extraction of forearm EMGsignalrdquo Proceedings of the National Academy of Sciences IndiaSection AmdashPhysical Sciences vol 84 no 3 pp 473ndash480 2014

[65] A R Siddiqi S N Sidek andAKhorshidtalab ldquoSignal process-ing of EMG signal for continuous thumb-angle estimationrdquo inProceedings of the 41st Annual Conference of the IEEE IndustrialElectronics Society (IECON rsquo15) pp 374ndash379 Yokohama JapanNovember 2015

[66] A Kilbom G M Hagg and C Kall ldquoOne-handed loadcarryingmdashcardiovascular muscular and subjective indices ofendurance and fatiguerdquo European Journal of Applied Physiologyand Occupational Physiology vol 65 no 1 pp 52ndash58 1992

[67] F AlOmari and G Liu ldquoAnalysis of extracted forearm sEMGsignal using LDA QDA K-NN classification algorithmsrdquoOpenAutomation and Control Systems Journal vol 6 no 1 pp 108ndash116 2014

[68] A Phinyomark G Chujit P Phukpattaranont C Limsakuland H Hu ldquoA preliminary study assessing time-domain EMGfeatures of classifying exercises in preventing falls in the elderlyrdquoin Proceedings of the 9th International Conference on Electri-cal EngineeringElectronics Computer Telecommunications and

BioMed Research International 11

Information Technology (ECTI-CON rsquo12) pp 1ndash4 IEEE Phetch-aburi Thailand May 2012

[69] D R Rogers and D T MacIsaac ldquoEMG-based muscle fatigueassessment during dynamic contractions using principal com-ponent analysisrdquo Journal of Electromyography and Kinesiologyvol 21 no 5 pp 811ndash818 2011

[70] K Buchenrieder ldquoDimensionality reduction for the controlof powered upper limb prosthesesrdquo in Proceedings of the 14thAnnual IEEE International Conference and Workshops on theEngineering of Computer-Based Systems (ECBS rsquo07) pp 327ndash333IEEE Tucson Ariz USA March 2007

[71] G Venugopal M Navaneethakrishna and S RamakrishnanldquoExtraction and analysis of multiple time window featuresassociated with muscle fatigue conditions using sEMG signalsrdquoExpert Systems with Applications vol 41 no 6 pp 2652ndash26592014

[72] A Phinyomark C Limsakul and P Phukpattaranont ldquoEMGfeature extraction for tolerance of white Gaussian noiserdquo inProceedings of the International Workshop and Symposium onScience and Technology (I-SEEC rsquo08) pp 178ndash183 December2008

[73] M S Al-Quraishi A J Ishak S A Ahmad and M K HasanldquoImpact of feature extraction techniques on classification accu-racy for EMG based ankle joint movementsrdquo in Proceedings ofthe 10th Asian Control Conference (ASCC rsquo15) pp 1ndash5 SabahMalaysia June 2015

[74] G-C Chang W-J Kang L Jer-Junn et al ldquoReal-time imple-mentation of electromyogram pattern recognition as a controlcommand of man-machine interfacerdquoMedical Engineering andPhysics vol 18 no 7 pp 529ndash537 1996

[75] A Georgakis L K Stergioulas and G Giakas ldquoFatigue analysisof the surface EMG signal in isometric constant force con-tractions using the averaged instantaneous frequencyrdquo IEEETransactions on Biomedical Engineering vol 50 no 2 pp 262ndash265 2003

[76] R Merletti and L R Lo Conte ldquoSurface EMG signal processingduring isometric contractionsrdquo Journal of Electromyography andKinesiology vol 7 no 4 pp 241ndash250 1997

[77] C J De Luca ldquoUse of the surface EMG signal for performanceevaluation of backmusclesrdquoMuscle and Nerve vol 16 no 2 pp210ndash216 1993

[78] F Khanam and M Ahmad ldquoFrequency based EMG powerspectrum analysis of Salat associated muscle contractionrdquo inProceedings of the 1st International Conference on Electricaland Electronic Engineering (ICEEE rsquo15) Rajshahi BangladeshNovember 2015

[79] M M Altaf B M Elbagoury F Alraddady and M RoushdyldquoExtended case-based behavior control for multi-humanoidrobotsrdquo International Journal of Humanoid Robotics vol 13 no2 2016

[80] A Phinyomark A Nuidod P Phukpattaranont and C Lim-sakul ldquoFeature extraction and reduction of wavelet transformcoefficients for EMG pattern classificationrdquo Elektronika ir Elek-trotechnika vol 122 no 6 pp 27ndash32 2012

[81] M Gonzalez-Izal A Malanda I Navarro-Amezqueta et alldquoEMG spectral indices and muscle power fatigue duringdynamic contractionsrdquo Journal of Electromyography and Kine-siology vol 20 no 2 pp 233ndash240 2010

[82] G V Dimitrov T I Arabadzhiev K N Mileva J L Bowtell NCrichton andNADimitrova ldquoMuscle fatigue during dynamiccontractions assessed by new spectral indicesrdquo Medicine andScience in Sports and Exercise vol 38 no 11 pp 1971ndash1979 2006

[83] MVitor-Costa H Bortolotti T V Camata et al ldquoEMG spectralanalysis of incremental exercise in cyclists and non-cyclistsusing fourier and wavelet transformsrdquo Revista Brasileira deCineantropometria e Desempenho Humano vol 14 no 6 pp660ndash670 2012

[84] M Wacker and H Witte ldquoTime-frequency techniques in bio-medical signal analysis a tutorial review of similarities anddifferencesrdquoMethods of Information in Medicine vol 52 no 4pp 279ndash296 2013

[85] S Karlsson J Yu and M Akay ldquoTime-frequency analysis ofmyoelectric signals during dynamic contractions a compara-tive studyrdquo IEEE Transactions on Biomedical Engineering vol47 no 2 pp 228ndash238 2000

[86] A L Martinez-Herrera L M Ledesma-Carrillo M Lopez-Ramirez S Salazar-Colores E Cabal-Yepez and A Garcia-Perez ldquoGabor and theWigner-Ville transforms for broken rotorbars detection in induction motorsrdquo in Proceedings of the 24thInternational Conference on Electronics Communications andComputers (CONIELECOMP rsquo14) pp 83ndash87 IEEE CholulaMexico February 2014

[87] D MacIsaac P A Parker and R N Scott ldquoThe short-timeFourier transform and muscle fatigue assessment in dynamiccontractionsrdquo Journal of Electromyography and Kinesiology vol11 no 6 pp 439ndash449 2001

[88] E F Shair A R Abdullah T N S Tengku Zawawi S AAhmad and SMohamad Saleh ldquoAuto-segmentation analysis ofEMG signal for lifting muscle contraction activitiesrdquo Journal ofTelecommunication Electronic and Computer Engineering vol8 no 7 pp 17ndash22 2016

[89] MM Jankovic andD B Popovic ldquoAnEMGsystem for studyingmotor control strategies and fatiguerdquo in Proceedings of the 10thSymposium on Neural Network Applications in Electrical Engi-neering (NEUREL-rsquo10) pp 15ndash18 Belgrade Serbia September2010

[90] T N S T Zawawi A R Abdullah I Halim E F Shair andS M Salleh ldquoApplication of spectrogram in analysing elec-tromyography (EMG) signals of manual liftingrdquo ARPN Journalof Engineering andApplied Sciences vol 11 no 6 pp 3603ndash36092016

[91] M Yochum T Bakir R Lepers and S Binczak ldquoEstimationof muscular fatigue under electromyostimulation using CWTrdquoIEEE Transactions on Biomedical Engineering vol 59 no 12 pp3372ndash3378 2012

[92] J L Dantas T V Camata M A O C Brunetto A C MoraesT Abrao and L R Altimari ldquoFourier and Wavelet spectralanalysis of EMG signals in isometric and dynamic maximaleffort exerciserdquo in Proceedings of the 32nd Annual InternationalConference of the IEEEEngineering inMedicine andBiology Soci-ety (EMBC rsquo10) pp 5979ndash5982 IEEE Buenos Aires ArgentinaSeptember 2010

[93] L Ai J Nan and J Shen ldquoApplication of S-transform to drivingfatigue in EEG analysisrdquo in Proceedings of the IEEE InternationalConference on Intelligent Computing and Intelligent Systems(ICIS rsquo10) pp 814ndash817 IEEE Guilin China October 2010

[94] R Upadhyay P K Padhy and P K Kankar ldquoEEG artifactremoval and noise suppression by discrete orthonormal S-transform denoisingrdquo Computers and Electrical Engineeringvol 53 pp 125ndash142 2016

[95] C Dora and P K Biswal ldquoRobust ECG artifact removalfrom EEG using continuous wavelet transformation and linearregressionrdquo in 2016 International Conference on Signal Process-ing and Communications (SPCOM) pp 1ndash5 Bangalore IndiaJune 2016

12 BioMed Research International

[96] S Assous and B Boashash ldquoEvaluation of themodified S-trans-form for time-frequency synchrony analysis and source locali-sationrdquo Eurasip Journal on Advances in Signal Processing vol2012 article 49 2012

[97] A L Ricamato R G Absher M T Moffroid and J P Tra-nowski ldquoA time-frequency approach to evaluate electromyo-graphic recordingsrdquo in Proceedings of the 5th Annual IEEESymposium on Computer-Based Medical Systems pp 520ndash527IEEE Durham NC USA 1992

[98] M Khezri and M Jahed ldquoAn inventive quadratic time-freq-uency scheme based on Wigner-Ville distribution for classifi-cation of sEMG signalsrdquo in Proceedings of the 6th InternationalSpecial Topic Conference on Information TechnologyApplicationsin Biomedicine pp 261ndash264 IEEE Tokyo Japan November2007

[99] A Subasi and M K Kiymik ldquoMuscle fatigue detection in EMGusing time-frequency methods ICA and neural networksrdquoJournal of Medical Systems vol 34 no 4 pp 777ndash785 2010

[100] W Martin and P Flandrin ldquoWigner-Ville Spectral Analysisof Nonstationary Processesrdquo IEEE Transactions on AcousticsSpeech and Signal Processing vol 33 no 6 pp 1461ndash1470 1985

[101] H-I Choi and W J Williams ldquoImproved time-frequency rep-resentation of multicomponent signals using exponential ker-nelsrdquo IEEE Transactions on Acoustics Speech and Signal Pro-cessing vol 37 no 6 pp 862ndash871 1989

[102] G R Pereira L F de Oliveira and J Nadal ldquoReducing crossterms effects in the Choi-Williams transform of mioelectricsignalsrdquo Computer Methods and Programs in Biomedicine vol111 no 3 pp 685ndash692 2013

[103] M Alemu D K Kumar and A Bradley ldquoTime-frequency anal-ysis of SEMGmdashwith special consideration to the interelectrodespacingrdquo IEEE Transactions on Neural Systems and Rehabilita-tion Engineering vol 11 no 4 pp 341ndash345 2003

[104] P A Karthick G Venugopal and S Ramakrishnan ldquoAnalysis ofsurface EMG signals under fatigue and non-fatigue conditionsusing B-distribution based quadratic time frequency distribu-tionrdquo Journal of Mechanics in Medicine and Biology vol 15 no2 Article ID 1540028 2015

[105] V Sucic B Barkat and B Boashash ldquoPerformance evaluationof the B-distributionrdquo in Proceedings of the 5th InternationalSymposium on Signal Processing and Its Applications (ISSPA rsquo99)pp 267ndash270 Queensland Australia August 1999

[106] PAKarthick and S Ramakrishnan ldquoSurface electromyographybasedmuscle fatigue progression analysis usingmodified B dis-tribution time-frequency featuresrdquo Biomedical Signal Processingand Control vol 26 pp 42ndash51 2016

[107] B Boashash and V Sucic ldquoResolution measure criteria for theobjective assessment of the performance of quadratic time-frequency distributionsrdquo IEEE Transactions on Signal Process-ing vol 51 no 5 pp 1253ndash1263 2003

[108] S Dong G Azemi and B Boashash ldquoImproved characteriza-tion of HRV signals based on instantaneous frequency featuresestimated from quadratic time-frequency distributions withdata-adapted kernelsrdquoBiomedical Signal Processing andControlvol 10 no 1 pp 153ndash165 2014

[109] B Boashash and T Ben-Jabeur ldquoDesign of a high-resolutionseparable-kernel quadratic TFD for improving newborn healthoutcomes using fetal movement detectionrdquo in Proceedings ofthe 11th International Conference on Information Science SignalProcessing and their Applications (ISSPA rsquo12) pp 354ndash359Quebec Canada July 2012

Submit your manuscripts athttpswwwhindawicom

Stem CellsInternational

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MEDIATORSINFLAMMATION

of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Behavioural Neurology

EndocrinologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Disease Markers

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

BioMed Research International

OncologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Oxidative Medicine and Cellular Longevity

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

PPAR Research

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Immunology ResearchHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

ObesityJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Computational and Mathematical Methods in Medicine

OphthalmologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Diabetes ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Research and TreatmentAIDS

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Gastroenterology Research and Practice

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Parkinsonrsquos Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttpwwwhindawicom

Page 9: EMG Processing Based Measures of Fatigue Assessment during …downloads.hindawi.com/journals/bmri/2017/3937254.pdf · 2019-07-30 · BioMedResearchInternational 3 5. Electromyography

BioMed Research International 9

[10] R A Al-Ashaik M Z Ramadan K S Al-Saleh and T M Kha-laf ldquoEffect of safety shoes type lifting frequency and ambienttemperature on subjectrsquos MAWL and physiological responsesrdquoInternational Journal of Industrial Ergonomics vol 50 pp 43ndash512015

[11] K P Granata and W S Marras ldquoAn EMG-assisted model oftrunk loading during free-dynamic liftingrdquo Journal of Biome-chanics vol 28 no 11 pp 1309ndash1317 1995

[12] S H Roy P Bonato and M Knaflitz ldquoEMG assessment ofback muscle function during cyclical liftingrdquo Journal of Elec-tromyography and Kinesiology vol 8 no 4 pp 233ndash245 1998

[13] R E Seroussi andM H Pope ldquoThe relationship between trunkmuscle electromyography and lifting moments in the sagittaland frontal planesrdquo Journal of Biomechanics vol 20 no 2 pp135ndash146 1987

[14] R B Graham M J Agnew and J M Stevenson ldquoEffectivenessof an on-body lifting aid at reducing low back physical demandsduring an automotive assembly task assessment of EMG res-ponse and user acceptabilityrdquo Applied Ergonomics vol 40 no5 pp 936ndash942 2009

[15] D Gagnon C Lariviere and P Loisel ldquoComparative abilityof EMG optimization and hybrid modelling approaches topredict trunk muscle forces and lumbar spine loading duringdynamic sagittal plane liftingrdquoClinical Biomechanics vol 16 no5 pp 359ndash372 2001

[16] P Dolan and M A Adams ldquoThe relationship between EMGactivity and extensor moment generation in the erector spinaemuscles during bending and lifting activitiesrdquo Journal of Biome-chanics vol 26 no 4-5 pp 513ndash522 1993

[17] P Bonato P Boissy U Della Croce and S H Roy ldquoChanges inthe surface EMG signal and the biomechanics of motion duringa repetitive lifting taskrdquo IEEE Transactions on Neural Systemsand Rehabilitation Engineering vol 10 no 1 pp 38ndash47 2002

[18] J R Potvin ldquoMechanically corrected EMG for the continuousestimation of erector spinae muscle loading during repetitiveliftingrdquo European Journal of Applied Physiology and Occupa-tional Physiology vol 74 no 1-2 pp 119ndash132 1996

[19] I Kingma and J H Van Dieen ldquoLifting over an obstacle effectsof one-handed lifting and hand support on trunk kinematicsand low back loadingrdquo Journal of Biomechanics vol 37 no 2pp 249ndash255 2004

[20] H-J Shin and J-Y Kim ldquoMeasurement of trunk muscle fatigueduring dynamic lifting and lowering as recovery time changesrdquoInternational Journal of Industrial Ergonomics vol 37 no 6 pp545ndash551 2007

[21] J Cholewicki S M McGill and R W Norman ldquoComparisonof muscle forces and joint load from an optimization and EMGassisted lumbar spine model towards development of a hybridapproachrdquo Journal of Biomechanics vol 28 no 3 pp 321ndash3311995

[22] L Straker ldquoEvidence to support using squat semi-squat andstoop techniques to lift low-lying objectsrdquo International Journalof Industrial Ergonomics vol 31 no 3 pp 149ndash160 2003

[23] C K Anderson and D B Chaffin ldquoA biomechanical evaluationof five lifting techniquesrdquo Applied Ergonomics vol 17 no 1 pp2ndash8 1986

[24] R Burgess-Limerick ldquoSquat stoop or something in betweenrdquoInternational Journal of Industrial Ergonomics vol 31 no 3 pp143ndash148 2003

[25] S M Hsiang G E Brogmus and T K Courtney ldquoLow backpain (LBP) and lifting techniquemdasha reviewrdquo International Jour-nal of Industrial Ergonomics vol 19 no 1 pp 59ndash74 1997

[26] A Merlo and I Campanini ldquoTechnical aspects of surface elec-tromyography for cliniciansrdquo The Open Rehabilitation Journalvol 3 no 1 pp 98ndash109 2010

[27] E N Kamavuako J C Rosenvang R Horup W Jensen DFarina and K B Englehart ldquoSurface versus untargeted intra-muscular EMG based classification of simultaneous and dyna-mically changing movementsrdquo IEEE Transactions on NeuralSystems and Rehabilitation Engineering vol 21 no 6 pp 992ndash998 2013

[28] L H Smith and L J Hargrove ldquoComparison of surface andintramuscular EMG pattern recognition for simultaneouswristhand motion classificationrdquo in Proceedings of the 35thAnnual International Conference of the IEEE Engineering inMedicine and Biology Society (EMBC rsquo13) pp 4223ndash4226Osaka Japan July 2013

[29] L J Hargrove K Englehart and B Hudgins ldquoA comparisonof surface and intramuscular myoelectric signal classificationrdquoIEEE Transactions on Biomedical Engineering vol 54 no 5 pp847ndash853 2007

[30] A Fratini M Cesarelli P Bifulco and M Romano ldquoRelevanceof motion artifact in electromyography recordings duringvibration treatmentrdquo Journal of Electromyography and Kinesiol-ogy vol 19 no 4 pp 710ndash718 2009

[31] C J De Luca L Donald Gilmore M Kuznetsov and S HRoy ldquoFiltering the surface EMG signal movement artifact andbaseline noise contaminationrdquo Journal of Biomechanics vol 43no 8 pp 1573ndash1579 2010

[32] H L Butler R Newell C L Hubley-Kozey and J W KozeyldquoThe interpretation of abdominal wall muscle recruitment stra-tegies change when the electrocardiogram (ECG) is removedfrom the electromyogram (EMG)rdquo Journal of Electromyographyand Kinesiology vol 19 no 2 pp e102ndashe113 2009

[33] G Lu J-S Brittain P Holland et al ldquoRemoving ECG noisefrom surface EMG signals using adaptive filteringrdquo Neuro-science Letters vol 462 no 1 pp 14ndash19 2009

[34] N W Willigenburg A Daffertshofer I Kingma and J H vanDieen ldquoRemoving ECG contamination from EMG recordingsa comparison of ICA-based and other filtering proceduresrdquoJournal of Electromyography and Kinesiology vol 22 no 3 pp485ndash493 2012

[35] M I Ibrahimy M R Ahsan and O O Khalifa ldquoDesign andoptimization of Levenberg-Marquardt based neural networkclassifier for EMG signals to identify hand motionsrdquo Measure-ment Science Review vol 13 no 3 pp 142ndash151 2013

[36] S N Sidek and A J Haja Mohideen ldquoMapping of EMG signalto hand grip force at varying wrist anglesrdquo in Proceedings ofthe 2nd IEEE-EMBS Conference on Biomedical Engineering andSciences (IECBES rsquo12) pp 648ndash653 IEEE Langkawi MalaysiaDecember 2012

[37] Y Zhan D Halliday P Jiang X Liu and J Feng ldquoDetectingtime-dependent coherence between non-stationary electro-physiological signalsmdasha combined statistical and time-freq-uency approachrdquo Journal of Neuroscience Methods vol 156 no1-2 pp 322ndash332 2006

[38] E R Ferrara and BWidrow ldquoFetal electrocardiogram enhance-ment by time-sequenced adaptive filteringrdquo IEEE Transactionson Biomedical Engineering vol 29 no 6 pp 458ndash460 1982

[39] R H Shumway ldquoTime-frequency clustering and discriminantanalysisrdquo Statistics amp Probability Letters vol 63 no 3 pp 307ndash314 2003

[40] D Korosec ldquoParametric estimation of the continuous non-stationary spectrum and its dynamics in surface EMG studiesrdquo

10 BioMed Research International

International Journal of Medical Informatics vol 58-59 pp 59ndash69 2000

[41] G Morren S Van Huffel I Helon et al ldquoEffects of non-nutritive sucking on heart rate respiration and oxygenationa model-based signal processing approachrdquo Comparative Bio-chemistry and PhysiologymdashAMolecular and Integrative Physiol-ogy vol 132 no 1 pp 97ndash106 2002

[42] R Latif E Aassif M Laaboubi A Moudden and G MazeldquoDetermination of the cut-off frequency of the anti-symmetriccircumferential waves using the time-varying autoregressiveTVARrdquo in Proceedings of the 3rd International Symposium onCommunications Control and Signal Processing (ISCCSP rsquo08)pp 357ndash361 IEEE March 2008

[43] R Latif E Aassif M Laaboubi and G Maze ldquoDeterminationof the group velocity using the time-varying autoregressiveTVARrdquo in Proceedings of the 9th International Symposium onSignal Processing and Its Applications (ISSPA rsquo07) pp 1ndash4 IEEESharjah United Arab Emirates February 2007

[44] A Al Zaman X Luo M Ferdjallah and A Khamayseh ldquoA newTVARmodeling in cascaded form for nonstationary signalsrdquo inProceedings of the IEEE International Conference on AcousticsSpeech and Signal Processing (ICASSP rsquo06) pp 353ndash356 May2006

[45] A Al Zaman M A Ahad M Ferdjallah and J J Wertsch ldquoAnew approach for muscle fatigue analysis in young adults atdifferent MVC levelsrdquo in Proceedings of the IEEE International48th Midwest Symposium on Circuits and Systems (MWSCASrsquo05) pp 499ndash502 August 2005

[46] Z G Zhang H T Liu S C Chan K D K Luk and Y HuldquoTime-dependent power spectral density estimation of surfaceelectromyography during isometric muscle contraction meth-ods and comparisonsrdquo Journal of Electromyography and Kinesi-ology vol 20 no 1 pp 89ndash101 2010

[47] AAl ZamanM Ferdjallah andAKhamayseh ldquoMuscle fatigueanalysis for healthy adults using TVAR model with instanta-neous frequency estimationrdquo in Proceedings of the 38th South-eastern Symposium on SystemTheory pp 44ndash47 March 2006

[48] K Englehart B Hudgins P A Parker and M StevensonldquoClassification of the myoelectric signal using time-frequencybased representationsrdquoMedical Engineering and Physics vol 21no 6-7 pp 431ndash438 1999

[49] N Selvaraj J Lee and K H Chon ldquoTime-varying methods forcharac-terizing nonstationary dynamics of physiological sys-temsrdquo Methods of Information in Medicine vol 49 no 5 pp435ndash442 2010

[50] D Tkach H Huang and T A Kuiken ldquoStudy of stability oftime-domain features for electromyographic pattern recogni-tionrdquo Journal of NeuroEngineering and Rehabilitation vol 7article 21 13 pages 2010

[51] S Du and M Vuskovic ldquoTemporal vs spectral approach tofeature extraction from prehensile EMG signalsrdquo in Proceedingsof the IEEE International Conference on Information Reuseand Integration (IRI rsquo04) pp 344ndash350 Las Vegas Nev USANovember 2004

[52] R M Enoka and J Duchateau ldquoMuscle fatigue what why andhow it influences muscle functionrdquo Journal of Physiology vol586 no 1 pp 11ndash23 2008

[53] W J Williams and J Jeong ldquoKernel design for reduced interfer-ence distributionsrdquo IEEE Transactions on Signal Processing vol40 no 2 pp 402ndash412 1992

[54] M B I Reaz M S Hussain and F Mohd-Yasin ldquoTechniquesof EMG signal analysis detection processing classification and

applicationsrdquo Biological Procedures Online vol 8 no 1 pp 11ndash35 2006

[55] R A Malinzak S M Colby D T Kirkendall B Yu and W EGarrett ldquoA comparison of knee joint motion patterns betweenmen and women in selected athletic tasksrdquo Clinical Biomechan-ics vol 16 no 5 pp 438ndash445 2001

[56] T I Arabadzhiev V G Dimitrov N A Dimitrova and G VDimitrov ldquoInterpretation of EMG integral or RMS and esti-mates of lsquoneuromuscular efficiencyrsquo can bemisleading in fatigu-ing contractionrdquo Journal of Electromyography and Kinesiologyvol 20 no 2 pp 223ndash232 2010

[57] C Suetta P Aagaard A Rosted et al ldquoTraining-inducedchanges in muscle CSA muscle strength EMG and rate offorce development in elderly subjects after long-term unilateraldisuserdquo Journal of Applied Physiology vol 97 no 5 pp 1954ndash1961 2004

[58] M A Oskoei andHHu ldquoSupport vectormachine-based classi-fication scheme for myoelectric control applied to upper limbrdquoIEEE Transactions on Biomedical Engineering vol 55 no 8 pp1956ndash1965 2008

[59] J J Villarejo A Frizera T F Bastos and J F Sarmiento ldquoPatternrecognition of hand movements with low density sEMG forprosthesis control purposesrdquo in Proceedings of the IEEE 13thInternational Conference onRehabilitation Robotics (ICORR rsquo13)Seattle Wash USA June 2013

[60] A Phinyomark C Limsakul and P Phukpattaranont ldquoA novelfeature extraction for robust EMG pattern recognitionrdquo Journalof Computing vol 1 no 1 pp 71ndash80 2009

[61] M Zardoshti-Kermani B C Wheeler K Badie and R MHashemi ldquoEMG Feature Evaluation for Movement Control ofUpper Extremity Prosthesesrdquo IEEE Transactions on Rehabilita-tion Engineering vol 3 no 4 pp 324ndash333 1995

[62] D Tkach H Huang and T A Kuiken ldquoStudy of stability oftime-domain features for electromyographic pattern recogni-tionrdquo Journal of NeuroEngineering and Rehabilitation vol 7 no1 article no 21 2010

[63] K Kiguchi S Kariya K Watanabe K Izumi and T FukudaldquoAn exoskeletal robot for human elbowmotion supportmdashsensorfusion adaptation and controlrdquo IEEE Transactions on SystemsMan and Cybernetics Part B Cybernetics vol 31 no 3 pp 353ndash361 2001

[64] F Al Omari J Hui C Mei and G Liu ldquoPattern recognition ofeight hand motions using feature extraction of forearm EMGsignalrdquo Proceedings of the National Academy of Sciences IndiaSection AmdashPhysical Sciences vol 84 no 3 pp 473ndash480 2014

[65] A R Siddiqi S N Sidek andAKhorshidtalab ldquoSignal process-ing of EMG signal for continuous thumb-angle estimationrdquo inProceedings of the 41st Annual Conference of the IEEE IndustrialElectronics Society (IECON rsquo15) pp 374ndash379 Yokohama JapanNovember 2015

[66] A Kilbom G M Hagg and C Kall ldquoOne-handed loadcarryingmdashcardiovascular muscular and subjective indices ofendurance and fatiguerdquo European Journal of Applied Physiologyand Occupational Physiology vol 65 no 1 pp 52ndash58 1992

[67] F AlOmari and G Liu ldquoAnalysis of extracted forearm sEMGsignal using LDA QDA K-NN classification algorithmsrdquoOpenAutomation and Control Systems Journal vol 6 no 1 pp 108ndash116 2014

[68] A Phinyomark G Chujit P Phukpattaranont C Limsakuland H Hu ldquoA preliminary study assessing time-domain EMGfeatures of classifying exercises in preventing falls in the elderlyrdquoin Proceedings of the 9th International Conference on Electri-cal EngineeringElectronics Computer Telecommunications and

BioMed Research International 11

Information Technology (ECTI-CON rsquo12) pp 1ndash4 IEEE Phetch-aburi Thailand May 2012

[69] D R Rogers and D T MacIsaac ldquoEMG-based muscle fatigueassessment during dynamic contractions using principal com-ponent analysisrdquo Journal of Electromyography and Kinesiologyvol 21 no 5 pp 811ndash818 2011

[70] K Buchenrieder ldquoDimensionality reduction for the controlof powered upper limb prosthesesrdquo in Proceedings of the 14thAnnual IEEE International Conference and Workshops on theEngineering of Computer-Based Systems (ECBS rsquo07) pp 327ndash333IEEE Tucson Ariz USA March 2007

[71] G Venugopal M Navaneethakrishna and S RamakrishnanldquoExtraction and analysis of multiple time window featuresassociated with muscle fatigue conditions using sEMG signalsrdquoExpert Systems with Applications vol 41 no 6 pp 2652ndash26592014

[72] A Phinyomark C Limsakul and P Phukpattaranont ldquoEMGfeature extraction for tolerance of white Gaussian noiserdquo inProceedings of the International Workshop and Symposium onScience and Technology (I-SEEC rsquo08) pp 178ndash183 December2008

[73] M S Al-Quraishi A J Ishak S A Ahmad and M K HasanldquoImpact of feature extraction techniques on classification accu-racy for EMG based ankle joint movementsrdquo in Proceedings ofthe 10th Asian Control Conference (ASCC rsquo15) pp 1ndash5 SabahMalaysia June 2015

[74] G-C Chang W-J Kang L Jer-Junn et al ldquoReal-time imple-mentation of electromyogram pattern recognition as a controlcommand of man-machine interfacerdquoMedical Engineering andPhysics vol 18 no 7 pp 529ndash537 1996

[75] A Georgakis L K Stergioulas and G Giakas ldquoFatigue analysisof the surface EMG signal in isometric constant force con-tractions using the averaged instantaneous frequencyrdquo IEEETransactions on Biomedical Engineering vol 50 no 2 pp 262ndash265 2003

[76] R Merletti and L R Lo Conte ldquoSurface EMG signal processingduring isometric contractionsrdquo Journal of Electromyography andKinesiology vol 7 no 4 pp 241ndash250 1997

[77] C J De Luca ldquoUse of the surface EMG signal for performanceevaluation of backmusclesrdquoMuscle and Nerve vol 16 no 2 pp210ndash216 1993

[78] F Khanam and M Ahmad ldquoFrequency based EMG powerspectrum analysis of Salat associated muscle contractionrdquo inProceedings of the 1st International Conference on Electricaland Electronic Engineering (ICEEE rsquo15) Rajshahi BangladeshNovember 2015

[79] M M Altaf B M Elbagoury F Alraddady and M RoushdyldquoExtended case-based behavior control for multi-humanoidrobotsrdquo International Journal of Humanoid Robotics vol 13 no2 2016

[80] A Phinyomark A Nuidod P Phukpattaranont and C Lim-sakul ldquoFeature extraction and reduction of wavelet transformcoefficients for EMG pattern classificationrdquo Elektronika ir Elek-trotechnika vol 122 no 6 pp 27ndash32 2012

[81] M Gonzalez-Izal A Malanda I Navarro-Amezqueta et alldquoEMG spectral indices and muscle power fatigue duringdynamic contractionsrdquo Journal of Electromyography and Kine-siology vol 20 no 2 pp 233ndash240 2010

[82] G V Dimitrov T I Arabadzhiev K N Mileva J L Bowtell NCrichton andNADimitrova ldquoMuscle fatigue during dynamiccontractions assessed by new spectral indicesrdquo Medicine andScience in Sports and Exercise vol 38 no 11 pp 1971ndash1979 2006

[83] MVitor-Costa H Bortolotti T V Camata et al ldquoEMG spectralanalysis of incremental exercise in cyclists and non-cyclistsusing fourier and wavelet transformsrdquo Revista Brasileira deCineantropometria e Desempenho Humano vol 14 no 6 pp660ndash670 2012

[84] M Wacker and H Witte ldquoTime-frequency techniques in bio-medical signal analysis a tutorial review of similarities anddifferencesrdquoMethods of Information in Medicine vol 52 no 4pp 279ndash296 2013

[85] S Karlsson J Yu and M Akay ldquoTime-frequency analysis ofmyoelectric signals during dynamic contractions a compara-tive studyrdquo IEEE Transactions on Biomedical Engineering vol47 no 2 pp 228ndash238 2000

[86] A L Martinez-Herrera L M Ledesma-Carrillo M Lopez-Ramirez S Salazar-Colores E Cabal-Yepez and A Garcia-Perez ldquoGabor and theWigner-Ville transforms for broken rotorbars detection in induction motorsrdquo in Proceedings of the 24thInternational Conference on Electronics Communications andComputers (CONIELECOMP rsquo14) pp 83ndash87 IEEE CholulaMexico February 2014

[87] D MacIsaac P A Parker and R N Scott ldquoThe short-timeFourier transform and muscle fatigue assessment in dynamiccontractionsrdquo Journal of Electromyography and Kinesiology vol11 no 6 pp 439ndash449 2001

[88] E F Shair A R Abdullah T N S Tengku Zawawi S AAhmad and SMohamad Saleh ldquoAuto-segmentation analysis ofEMG signal for lifting muscle contraction activitiesrdquo Journal ofTelecommunication Electronic and Computer Engineering vol8 no 7 pp 17ndash22 2016

[89] MM Jankovic andD B Popovic ldquoAnEMGsystem for studyingmotor control strategies and fatiguerdquo in Proceedings of the 10thSymposium on Neural Network Applications in Electrical Engi-neering (NEUREL-rsquo10) pp 15ndash18 Belgrade Serbia September2010

[90] T N S T Zawawi A R Abdullah I Halim E F Shair andS M Salleh ldquoApplication of spectrogram in analysing elec-tromyography (EMG) signals of manual liftingrdquo ARPN Journalof Engineering andApplied Sciences vol 11 no 6 pp 3603ndash36092016

[91] M Yochum T Bakir R Lepers and S Binczak ldquoEstimationof muscular fatigue under electromyostimulation using CWTrdquoIEEE Transactions on Biomedical Engineering vol 59 no 12 pp3372ndash3378 2012

[92] J L Dantas T V Camata M A O C Brunetto A C MoraesT Abrao and L R Altimari ldquoFourier and Wavelet spectralanalysis of EMG signals in isometric and dynamic maximaleffort exerciserdquo in Proceedings of the 32nd Annual InternationalConference of the IEEEEngineering inMedicine andBiology Soci-ety (EMBC rsquo10) pp 5979ndash5982 IEEE Buenos Aires ArgentinaSeptember 2010

[93] L Ai J Nan and J Shen ldquoApplication of S-transform to drivingfatigue in EEG analysisrdquo in Proceedings of the IEEE InternationalConference on Intelligent Computing and Intelligent Systems(ICIS rsquo10) pp 814ndash817 IEEE Guilin China October 2010

[94] R Upadhyay P K Padhy and P K Kankar ldquoEEG artifactremoval and noise suppression by discrete orthonormal S-transform denoisingrdquo Computers and Electrical Engineeringvol 53 pp 125ndash142 2016

[95] C Dora and P K Biswal ldquoRobust ECG artifact removalfrom EEG using continuous wavelet transformation and linearregressionrdquo in 2016 International Conference on Signal Process-ing and Communications (SPCOM) pp 1ndash5 Bangalore IndiaJune 2016

12 BioMed Research International

[96] S Assous and B Boashash ldquoEvaluation of themodified S-trans-form for time-frequency synchrony analysis and source locali-sationrdquo Eurasip Journal on Advances in Signal Processing vol2012 article 49 2012

[97] A L Ricamato R G Absher M T Moffroid and J P Tra-nowski ldquoA time-frequency approach to evaluate electromyo-graphic recordingsrdquo in Proceedings of the 5th Annual IEEESymposium on Computer-Based Medical Systems pp 520ndash527IEEE Durham NC USA 1992

[98] M Khezri and M Jahed ldquoAn inventive quadratic time-freq-uency scheme based on Wigner-Ville distribution for classifi-cation of sEMG signalsrdquo in Proceedings of the 6th InternationalSpecial Topic Conference on Information TechnologyApplicationsin Biomedicine pp 261ndash264 IEEE Tokyo Japan November2007

[99] A Subasi and M K Kiymik ldquoMuscle fatigue detection in EMGusing time-frequency methods ICA and neural networksrdquoJournal of Medical Systems vol 34 no 4 pp 777ndash785 2010

[100] W Martin and P Flandrin ldquoWigner-Ville Spectral Analysisof Nonstationary Processesrdquo IEEE Transactions on AcousticsSpeech and Signal Processing vol 33 no 6 pp 1461ndash1470 1985

[101] H-I Choi and W J Williams ldquoImproved time-frequency rep-resentation of multicomponent signals using exponential ker-nelsrdquo IEEE Transactions on Acoustics Speech and Signal Pro-cessing vol 37 no 6 pp 862ndash871 1989

[102] G R Pereira L F de Oliveira and J Nadal ldquoReducing crossterms effects in the Choi-Williams transform of mioelectricsignalsrdquo Computer Methods and Programs in Biomedicine vol111 no 3 pp 685ndash692 2013

[103] M Alemu D K Kumar and A Bradley ldquoTime-frequency anal-ysis of SEMGmdashwith special consideration to the interelectrodespacingrdquo IEEE Transactions on Neural Systems and Rehabilita-tion Engineering vol 11 no 4 pp 341ndash345 2003

[104] P A Karthick G Venugopal and S Ramakrishnan ldquoAnalysis ofsurface EMG signals under fatigue and non-fatigue conditionsusing B-distribution based quadratic time frequency distribu-tionrdquo Journal of Mechanics in Medicine and Biology vol 15 no2 Article ID 1540028 2015

[105] V Sucic B Barkat and B Boashash ldquoPerformance evaluationof the B-distributionrdquo in Proceedings of the 5th InternationalSymposium on Signal Processing and Its Applications (ISSPA rsquo99)pp 267ndash270 Queensland Australia August 1999

[106] PAKarthick and S Ramakrishnan ldquoSurface electromyographybasedmuscle fatigue progression analysis usingmodified B dis-tribution time-frequency featuresrdquo Biomedical Signal Processingand Control vol 26 pp 42ndash51 2016

[107] B Boashash and V Sucic ldquoResolution measure criteria for theobjective assessment of the performance of quadratic time-frequency distributionsrdquo IEEE Transactions on Signal Process-ing vol 51 no 5 pp 1253ndash1263 2003

[108] S Dong G Azemi and B Boashash ldquoImproved characteriza-tion of HRV signals based on instantaneous frequency featuresestimated from quadratic time-frequency distributions withdata-adapted kernelsrdquoBiomedical Signal Processing andControlvol 10 no 1 pp 153ndash165 2014

[109] B Boashash and T Ben-Jabeur ldquoDesign of a high-resolutionseparable-kernel quadratic TFD for improving newborn healthoutcomes using fetal movement detectionrdquo in Proceedings ofthe 11th International Conference on Information Science SignalProcessing and their Applications (ISSPA rsquo12) pp 354ndash359Quebec Canada July 2012

Submit your manuscripts athttpswwwhindawicom

Stem CellsInternational

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MEDIATORSINFLAMMATION

of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Behavioural Neurology

EndocrinologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Disease Markers

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

BioMed Research International

OncologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Oxidative Medicine and Cellular Longevity

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

PPAR Research

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Immunology ResearchHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

ObesityJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Computational and Mathematical Methods in Medicine

OphthalmologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Diabetes ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Research and TreatmentAIDS

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Gastroenterology Research and Practice

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Parkinsonrsquos Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttpwwwhindawicom

Page 10: EMG Processing Based Measures of Fatigue Assessment during …downloads.hindawi.com/journals/bmri/2017/3937254.pdf · 2019-07-30 · BioMedResearchInternational 3 5. Electromyography

10 BioMed Research International

International Journal of Medical Informatics vol 58-59 pp 59ndash69 2000

[41] G Morren S Van Huffel I Helon et al ldquoEffects of non-nutritive sucking on heart rate respiration and oxygenationa model-based signal processing approachrdquo Comparative Bio-chemistry and PhysiologymdashAMolecular and Integrative Physiol-ogy vol 132 no 1 pp 97ndash106 2002

[42] R Latif E Aassif M Laaboubi A Moudden and G MazeldquoDetermination of the cut-off frequency of the anti-symmetriccircumferential waves using the time-varying autoregressiveTVARrdquo in Proceedings of the 3rd International Symposium onCommunications Control and Signal Processing (ISCCSP rsquo08)pp 357ndash361 IEEE March 2008

[43] R Latif E Aassif M Laaboubi and G Maze ldquoDeterminationof the group velocity using the time-varying autoregressiveTVARrdquo in Proceedings of the 9th International Symposium onSignal Processing and Its Applications (ISSPA rsquo07) pp 1ndash4 IEEESharjah United Arab Emirates February 2007

[44] A Al Zaman X Luo M Ferdjallah and A Khamayseh ldquoA newTVARmodeling in cascaded form for nonstationary signalsrdquo inProceedings of the IEEE International Conference on AcousticsSpeech and Signal Processing (ICASSP rsquo06) pp 353ndash356 May2006

[45] A Al Zaman M A Ahad M Ferdjallah and J J Wertsch ldquoAnew approach for muscle fatigue analysis in young adults atdifferent MVC levelsrdquo in Proceedings of the IEEE International48th Midwest Symposium on Circuits and Systems (MWSCASrsquo05) pp 499ndash502 August 2005

[46] Z G Zhang H T Liu S C Chan K D K Luk and Y HuldquoTime-dependent power spectral density estimation of surfaceelectromyography during isometric muscle contraction meth-ods and comparisonsrdquo Journal of Electromyography and Kinesi-ology vol 20 no 1 pp 89ndash101 2010

[47] AAl ZamanM Ferdjallah andAKhamayseh ldquoMuscle fatigueanalysis for healthy adults using TVAR model with instanta-neous frequency estimationrdquo in Proceedings of the 38th South-eastern Symposium on SystemTheory pp 44ndash47 March 2006

[48] K Englehart B Hudgins P A Parker and M StevensonldquoClassification of the myoelectric signal using time-frequencybased representationsrdquoMedical Engineering and Physics vol 21no 6-7 pp 431ndash438 1999

[49] N Selvaraj J Lee and K H Chon ldquoTime-varying methods forcharac-terizing nonstationary dynamics of physiological sys-temsrdquo Methods of Information in Medicine vol 49 no 5 pp435ndash442 2010

[50] D Tkach H Huang and T A Kuiken ldquoStudy of stability oftime-domain features for electromyographic pattern recogni-tionrdquo Journal of NeuroEngineering and Rehabilitation vol 7article 21 13 pages 2010

[51] S Du and M Vuskovic ldquoTemporal vs spectral approach tofeature extraction from prehensile EMG signalsrdquo in Proceedingsof the IEEE International Conference on Information Reuseand Integration (IRI rsquo04) pp 344ndash350 Las Vegas Nev USANovember 2004

[52] R M Enoka and J Duchateau ldquoMuscle fatigue what why andhow it influences muscle functionrdquo Journal of Physiology vol586 no 1 pp 11ndash23 2008

[53] W J Williams and J Jeong ldquoKernel design for reduced interfer-ence distributionsrdquo IEEE Transactions on Signal Processing vol40 no 2 pp 402ndash412 1992

[54] M B I Reaz M S Hussain and F Mohd-Yasin ldquoTechniquesof EMG signal analysis detection processing classification and

applicationsrdquo Biological Procedures Online vol 8 no 1 pp 11ndash35 2006

[55] R A Malinzak S M Colby D T Kirkendall B Yu and W EGarrett ldquoA comparison of knee joint motion patterns betweenmen and women in selected athletic tasksrdquo Clinical Biomechan-ics vol 16 no 5 pp 438ndash445 2001

[56] T I Arabadzhiev V G Dimitrov N A Dimitrova and G VDimitrov ldquoInterpretation of EMG integral or RMS and esti-mates of lsquoneuromuscular efficiencyrsquo can bemisleading in fatigu-ing contractionrdquo Journal of Electromyography and Kinesiologyvol 20 no 2 pp 223ndash232 2010

[57] C Suetta P Aagaard A Rosted et al ldquoTraining-inducedchanges in muscle CSA muscle strength EMG and rate offorce development in elderly subjects after long-term unilateraldisuserdquo Journal of Applied Physiology vol 97 no 5 pp 1954ndash1961 2004

[58] M A Oskoei andHHu ldquoSupport vectormachine-based classi-fication scheme for myoelectric control applied to upper limbrdquoIEEE Transactions on Biomedical Engineering vol 55 no 8 pp1956ndash1965 2008

[59] J J Villarejo A Frizera T F Bastos and J F Sarmiento ldquoPatternrecognition of hand movements with low density sEMG forprosthesis control purposesrdquo in Proceedings of the IEEE 13thInternational Conference onRehabilitation Robotics (ICORR rsquo13)Seattle Wash USA June 2013

[60] A Phinyomark C Limsakul and P Phukpattaranont ldquoA novelfeature extraction for robust EMG pattern recognitionrdquo Journalof Computing vol 1 no 1 pp 71ndash80 2009

[61] M Zardoshti-Kermani B C Wheeler K Badie and R MHashemi ldquoEMG Feature Evaluation for Movement Control ofUpper Extremity Prosthesesrdquo IEEE Transactions on Rehabilita-tion Engineering vol 3 no 4 pp 324ndash333 1995

[62] D Tkach H Huang and T A Kuiken ldquoStudy of stability oftime-domain features for electromyographic pattern recogni-tionrdquo Journal of NeuroEngineering and Rehabilitation vol 7 no1 article no 21 2010

[63] K Kiguchi S Kariya K Watanabe K Izumi and T FukudaldquoAn exoskeletal robot for human elbowmotion supportmdashsensorfusion adaptation and controlrdquo IEEE Transactions on SystemsMan and Cybernetics Part B Cybernetics vol 31 no 3 pp 353ndash361 2001

[64] F Al Omari J Hui C Mei and G Liu ldquoPattern recognition ofeight hand motions using feature extraction of forearm EMGsignalrdquo Proceedings of the National Academy of Sciences IndiaSection AmdashPhysical Sciences vol 84 no 3 pp 473ndash480 2014

[65] A R Siddiqi S N Sidek andAKhorshidtalab ldquoSignal process-ing of EMG signal for continuous thumb-angle estimationrdquo inProceedings of the 41st Annual Conference of the IEEE IndustrialElectronics Society (IECON rsquo15) pp 374ndash379 Yokohama JapanNovember 2015

[66] A Kilbom G M Hagg and C Kall ldquoOne-handed loadcarryingmdashcardiovascular muscular and subjective indices ofendurance and fatiguerdquo European Journal of Applied Physiologyand Occupational Physiology vol 65 no 1 pp 52ndash58 1992

[67] F AlOmari and G Liu ldquoAnalysis of extracted forearm sEMGsignal using LDA QDA K-NN classification algorithmsrdquoOpenAutomation and Control Systems Journal vol 6 no 1 pp 108ndash116 2014

[68] A Phinyomark G Chujit P Phukpattaranont C Limsakuland H Hu ldquoA preliminary study assessing time-domain EMGfeatures of classifying exercises in preventing falls in the elderlyrdquoin Proceedings of the 9th International Conference on Electri-cal EngineeringElectronics Computer Telecommunications and

BioMed Research International 11

Information Technology (ECTI-CON rsquo12) pp 1ndash4 IEEE Phetch-aburi Thailand May 2012

[69] D R Rogers and D T MacIsaac ldquoEMG-based muscle fatigueassessment during dynamic contractions using principal com-ponent analysisrdquo Journal of Electromyography and Kinesiologyvol 21 no 5 pp 811ndash818 2011

[70] K Buchenrieder ldquoDimensionality reduction for the controlof powered upper limb prosthesesrdquo in Proceedings of the 14thAnnual IEEE International Conference and Workshops on theEngineering of Computer-Based Systems (ECBS rsquo07) pp 327ndash333IEEE Tucson Ariz USA March 2007

[71] G Venugopal M Navaneethakrishna and S RamakrishnanldquoExtraction and analysis of multiple time window featuresassociated with muscle fatigue conditions using sEMG signalsrdquoExpert Systems with Applications vol 41 no 6 pp 2652ndash26592014

[72] A Phinyomark C Limsakul and P Phukpattaranont ldquoEMGfeature extraction for tolerance of white Gaussian noiserdquo inProceedings of the International Workshop and Symposium onScience and Technology (I-SEEC rsquo08) pp 178ndash183 December2008

[73] M S Al-Quraishi A J Ishak S A Ahmad and M K HasanldquoImpact of feature extraction techniques on classification accu-racy for EMG based ankle joint movementsrdquo in Proceedings ofthe 10th Asian Control Conference (ASCC rsquo15) pp 1ndash5 SabahMalaysia June 2015

[74] G-C Chang W-J Kang L Jer-Junn et al ldquoReal-time imple-mentation of electromyogram pattern recognition as a controlcommand of man-machine interfacerdquoMedical Engineering andPhysics vol 18 no 7 pp 529ndash537 1996

[75] A Georgakis L K Stergioulas and G Giakas ldquoFatigue analysisof the surface EMG signal in isometric constant force con-tractions using the averaged instantaneous frequencyrdquo IEEETransactions on Biomedical Engineering vol 50 no 2 pp 262ndash265 2003

[76] R Merletti and L R Lo Conte ldquoSurface EMG signal processingduring isometric contractionsrdquo Journal of Electromyography andKinesiology vol 7 no 4 pp 241ndash250 1997

[77] C J De Luca ldquoUse of the surface EMG signal for performanceevaluation of backmusclesrdquoMuscle and Nerve vol 16 no 2 pp210ndash216 1993

[78] F Khanam and M Ahmad ldquoFrequency based EMG powerspectrum analysis of Salat associated muscle contractionrdquo inProceedings of the 1st International Conference on Electricaland Electronic Engineering (ICEEE rsquo15) Rajshahi BangladeshNovember 2015

[79] M M Altaf B M Elbagoury F Alraddady and M RoushdyldquoExtended case-based behavior control for multi-humanoidrobotsrdquo International Journal of Humanoid Robotics vol 13 no2 2016

[80] A Phinyomark A Nuidod P Phukpattaranont and C Lim-sakul ldquoFeature extraction and reduction of wavelet transformcoefficients for EMG pattern classificationrdquo Elektronika ir Elek-trotechnika vol 122 no 6 pp 27ndash32 2012

[81] M Gonzalez-Izal A Malanda I Navarro-Amezqueta et alldquoEMG spectral indices and muscle power fatigue duringdynamic contractionsrdquo Journal of Electromyography and Kine-siology vol 20 no 2 pp 233ndash240 2010

[82] G V Dimitrov T I Arabadzhiev K N Mileva J L Bowtell NCrichton andNADimitrova ldquoMuscle fatigue during dynamiccontractions assessed by new spectral indicesrdquo Medicine andScience in Sports and Exercise vol 38 no 11 pp 1971ndash1979 2006

[83] MVitor-Costa H Bortolotti T V Camata et al ldquoEMG spectralanalysis of incremental exercise in cyclists and non-cyclistsusing fourier and wavelet transformsrdquo Revista Brasileira deCineantropometria e Desempenho Humano vol 14 no 6 pp660ndash670 2012

[84] M Wacker and H Witte ldquoTime-frequency techniques in bio-medical signal analysis a tutorial review of similarities anddifferencesrdquoMethods of Information in Medicine vol 52 no 4pp 279ndash296 2013

[85] S Karlsson J Yu and M Akay ldquoTime-frequency analysis ofmyoelectric signals during dynamic contractions a compara-tive studyrdquo IEEE Transactions on Biomedical Engineering vol47 no 2 pp 228ndash238 2000

[86] A L Martinez-Herrera L M Ledesma-Carrillo M Lopez-Ramirez S Salazar-Colores E Cabal-Yepez and A Garcia-Perez ldquoGabor and theWigner-Ville transforms for broken rotorbars detection in induction motorsrdquo in Proceedings of the 24thInternational Conference on Electronics Communications andComputers (CONIELECOMP rsquo14) pp 83ndash87 IEEE CholulaMexico February 2014

[87] D MacIsaac P A Parker and R N Scott ldquoThe short-timeFourier transform and muscle fatigue assessment in dynamiccontractionsrdquo Journal of Electromyography and Kinesiology vol11 no 6 pp 439ndash449 2001

[88] E F Shair A R Abdullah T N S Tengku Zawawi S AAhmad and SMohamad Saleh ldquoAuto-segmentation analysis ofEMG signal for lifting muscle contraction activitiesrdquo Journal ofTelecommunication Electronic and Computer Engineering vol8 no 7 pp 17ndash22 2016

[89] MM Jankovic andD B Popovic ldquoAnEMGsystem for studyingmotor control strategies and fatiguerdquo in Proceedings of the 10thSymposium on Neural Network Applications in Electrical Engi-neering (NEUREL-rsquo10) pp 15ndash18 Belgrade Serbia September2010

[90] T N S T Zawawi A R Abdullah I Halim E F Shair andS M Salleh ldquoApplication of spectrogram in analysing elec-tromyography (EMG) signals of manual liftingrdquo ARPN Journalof Engineering andApplied Sciences vol 11 no 6 pp 3603ndash36092016

[91] M Yochum T Bakir R Lepers and S Binczak ldquoEstimationof muscular fatigue under electromyostimulation using CWTrdquoIEEE Transactions on Biomedical Engineering vol 59 no 12 pp3372ndash3378 2012

[92] J L Dantas T V Camata M A O C Brunetto A C MoraesT Abrao and L R Altimari ldquoFourier and Wavelet spectralanalysis of EMG signals in isometric and dynamic maximaleffort exerciserdquo in Proceedings of the 32nd Annual InternationalConference of the IEEEEngineering inMedicine andBiology Soci-ety (EMBC rsquo10) pp 5979ndash5982 IEEE Buenos Aires ArgentinaSeptember 2010

[93] L Ai J Nan and J Shen ldquoApplication of S-transform to drivingfatigue in EEG analysisrdquo in Proceedings of the IEEE InternationalConference on Intelligent Computing and Intelligent Systems(ICIS rsquo10) pp 814ndash817 IEEE Guilin China October 2010

[94] R Upadhyay P K Padhy and P K Kankar ldquoEEG artifactremoval and noise suppression by discrete orthonormal S-transform denoisingrdquo Computers and Electrical Engineeringvol 53 pp 125ndash142 2016

[95] C Dora and P K Biswal ldquoRobust ECG artifact removalfrom EEG using continuous wavelet transformation and linearregressionrdquo in 2016 International Conference on Signal Process-ing and Communications (SPCOM) pp 1ndash5 Bangalore IndiaJune 2016

12 BioMed Research International

[96] S Assous and B Boashash ldquoEvaluation of themodified S-trans-form for time-frequency synchrony analysis and source locali-sationrdquo Eurasip Journal on Advances in Signal Processing vol2012 article 49 2012

[97] A L Ricamato R G Absher M T Moffroid and J P Tra-nowski ldquoA time-frequency approach to evaluate electromyo-graphic recordingsrdquo in Proceedings of the 5th Annual IEEESymposium on Computer-Based Medical Systems pp 520ndash527IEEE Durham NC USA 1992

[98] M Khezri and M Jahed ldquoAn inventive quadratic time-freq-uency scheme based on Wigner-Ville distribution for classifi-cation of sEMG signalsrdquo in Proceedings of the 6th InternationalSpecial Topic Conference on Information TechnologyApplicationsin Biomedicine pp 261ndash264 IEEE Tokyo Japan November2007

[99] A Subasi and M K Kiymik ldquoMuscle fatigue detection in EMGusing time-frequency methods ICA and neural networksrdquoJournal of Medical Systems vol 34 no 4 pp 777ndash785 2010

[100] W Martin and P Flandrin ldquoWigner-Ville Spectral Analysisof Nonstationary Processesrdquo IEEE Transactions on AcousticsSpeech and Signal Processing vol 33 no 6 pp 1461ndash1470 1985

[101] H-I Choi and W J Williams ldquoImproved time-frequency rep-resentation of multicomponent signals using exponential ker-nelsrdquo IEEE Transactions on Acoustics Speech and Signal Pro-cessing vol 37 no 6 pp 862ndash871 1989

[102] G R Pereira L F de Oliveira and J Nadal ldquoReducing crossterms effects in the Choi-Williams transform of mioelectricsignalsrdquo Computer Methods and Programs in Biomedicine vol111 no 3 pp 685ndash692 2013

[103] M Alemu D K Kumar and A Bradley ldquoTime-frequency anal-ysis of SEMGmdashwith special consideration to the interelectrodespacingrdquo IEEE Transactions on Neural Systems and Rehabilita-tion Engineering vol 11 no 4 pp 341ndash345 2003

[104] P A Karthick G Venugopal and S Ramakrishnan ldquoAnalysis ofsurface EMG signals under fatigue and non-fatigue conditionsusing B-distribution based quadratic time frequency distribu-tionrdquo Journal of Mechanics in Medicine and Biology vol 15 no2 Article ID 1540028 2015

[105] V Sucic B Barkat and B Boashash ldquoPerformance evaluationof the B-distributionrdquo in Proceedings of the 5th InternationalSymposium on Signal Processing and Its Applications (ISSPA rsquo99)pp 267ndash270 Queensland Australia August 1999

[106] PAKarthick and S Ramakrishnan ldquoSurface electromyographybasedmuscle fatigue progression analysis usingmodified B dis-tribution time-frequency featuresrdquo Biomedical Signal Processingand Control vol 26 pp 42ndash51 2016

[107] B Boashash and V Sucic ldquoResolution measure criteria for theobjective assessment of the performance of quadratic time-frequency distributionsrdquo IEEE Transactions on Signal Process-ing vol 51 no 5 pp 1253ndash1263 2003

[108] S Dong G Azemi and B Boashash ldquoImproved characteriza-tion of HRV signals based on instantaneous frequency featuresestimated from quadratic time-frequency distributions withdata-adapted kernelsrdquoBiomedical Signal Processing andControlvol 10 no 1 pp 153ndash165 2014

[109] B Boashash and T Ben-Jabeur ldquoDesign of a high-resolutionseparable-kernel quadratic TFD for improving newborn healthoutcomes using fetal movement detectionrdquo in Proceedings ofthe 11th International Conference on Information Science SignalProcessing and their Applications (ISSPA rsquo12) pp 354ndash359Quebec Canada July 2012

Submit your manuscripts athttpswwwhindawicom

Stem CellsInternational

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MEDIATORSINFLAMMATION

of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Behavioural Neurology

EndocrinologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Disease Markers

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

BioMed Research International

OncologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Oxidative Medicine and Cellular Longevity

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

PPAR Research

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Immunology ResearchHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

ObesityJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Computational and Mathematical Methods in Medicine

OphthalmologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Diabetes ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Research and TreatmentAIDS

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Gastroenterology Research and Practice

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Parkinsonrsquos Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttpwwwhindawicom

Page 11: EMG Processing Based Measures of Fatigue Assessment during …downloads.hindawi.com/journals/bmri/2017/3937254.pdf · 2019-07-30 · BioMedResearchInternational 3 5. Electromyography

BioMed Research International 11

Information Technology (ECTI-CON rsquo12) pp 1ndash4 IEEE Phetch-aburi Thailand May 2012

[69] D R Rogers and D T MacIsaac ldquoEMG-based muscle fatigueassessment during dynamic contractions using principal com-ponent analysisrdquo Journal of Electromyography and Kinesiologyvol 21 no 5 pp 811ndash818 2011

[70] K Buchenrieder ldquoDimensionality reduction for the controlof powered upper limb prosthesesrdquo in Proceedings of the 14thAnnual IEEE International Conference and Workshops on theEngineering of Computer-Based Systems (ECBS rsquo07) pp 327ndash333IEEE Tucson Ariz USA March 2007

[71] G Venugopal M Navaneethakrishna and S RamakrishnanldquoExtraction and analysis of multiple time window featuresassociated with muscle fatigue conditions using sEMG signalsrdquoExpert Systems with Applications vol 41 no 6 pp 2652ndash26592014

[72] A Phinyomark C Limsakul and P Phukpattaranont ldquoEMGfeature extraction for tolerance of white Gaussian noiserdquo inProceedings of the International Workshop and Symposium onScience and Technology (I-SEEC rsquo08) pp 178ndash183 December2008

[73] M S Al-Quraishi A J Ishak S A Ahmad and M K HasanldquoImpact of feature extraction techniques on classification accu-racy for EMG based ankle joint movementsrdquo in Proceedings ofthe 10th Asian Control Conference (ASCC rsquo15) pp 1ndash5 SabahMalaysia June 2015

[74] G-C Chang W-J Kang L Jer-Junn et al ldquoReal-time imple-mentation of electromyogram pattern recognition as a controlcommand of man-machine interfacerdquoMedical Engineering andPhysics vol 18 no 7 pp 529ndash537 1996

[75] A Georgakis L K Stergioulas and G Giakas ldquoFatigue analysisof the surface EMG signal in isometric constant force con-tractions using the averaged instantaneous frequencyrdquo IEEETransactions on Biomedical Engineering vol 50 no 2 pp 262ndash265 2003

[76] R Merletti and L R Lo Conte ldquoSurface EMG signal processingduring isometric contractionsrdquo Journal of Electromyography andKinesiology vol 7 no 4 pp 241ndash250 1997

[77] C J De Luca ldquoUse of the surface EMG signal for performanceevaluation of backmusclesrdquoMuscle and Nerve vol 16 no 2 pp210ndash216 1993

[78] F Khanam and M Ahmad ldquoFrequency based EMG powerspectrum analysis of Salat associated muscle contractionrdquo inProceedings of the 1st International Conference on Electricaland Electronic Engineering (ICEEE rsquo15) Rajshahi BangladeshNovember 2015

[79] M M Altaf B M Elbagoury F Alraddady and M RoushdyldquoExtended case-based behavior control for multi-humanoidrobotsrdquo International Journal of Humanoid Robotics vol 13 no2 2016

[80] A Phinyomark A Nuidod P Phukpattaranont and C Lim-sakul ldquoFeature extraction and reduction of wavelet transformcoefficients for EMG pattern classificationrdquo Elektronika ir Elek-trotechnika vol 122 no 6 pp 27ndash32 2012

[81] M Gonzalez-Izal A Malanda I Navarro-Amezqueta et alldquoEMG spectral indices and muscle power fatigue duringdynamic contractionsrdquo Journal of Electromyography and Kine-siology vol 20 no 2 pp 233ndash240 2010

[82] G V Dimitrov T I Arabadzhiev K N Mileva J L Bowtell NCrichton andNADimitrova ldquoMuscle fatigue during dynamiccontractions assessed by new spectral indicesrdquo Medicine andScience in Sports and Exercise vol 38 no 11 pp 1971ndash1979 2006

[83] MVitor-Costa H Bortolotti T V Camata et al ldquoEMG spectralanalysis of incremental exercise in cyclists and non-cyclistsusing fourier and wavelet transformsrdquo Revista Brasileira deCineantropometria e Desempenho Humano vol 14 no 6 pp660ndash670 2012

[84] M Wacker and H Witte ldquoTime-frequency techniques in bio-medical signal analysis a tutorial review of similarities anddifferencesrdquoMethods of Information in Medicine vol 52 no 4pp 279ndash296 2013

[85] S Karlsson J Yu and M Akay ldquoTime-frequency analysis ofmyoelectric signals during dynamic contractions a compara-tive studyrdquo IEEE Transactions on Biomedical Engineering vol47 no 2 pp 228ndash238 2000

[86] A L Martinez-Herrera L M Ledesma-Carrillo M Lopez-Ramirez S Salazar-Colores E Cabal-Yepez and A Garcia-Perez ldquoGabor and theWigner-Ville transforms for broken rotorbars detection in induction motorsrdquo in Proceedings of the 24thInternational Conference on Electronics Communications andComputers (CONIELECOMP rsquo14) pp 83ndash87 IEEE CholulaMexico February 2014

[87] D MacIsaac P A Parker and R N Scott ldquoThe short-timeFourier transform and muscle fatigue assessment in dynamiccontractionsrdquo Journal of Electromyography and Kinesiology vol11 no 6 pp 439ndash449 2001

[88] E F Shair A R Abdullah T N S Tengku Zawawi S AAhmad and SMohamad Saleh ldquoAuto-segmentation analysis ofEMG signal for lifting muscle contraction activitiesrdquo Journal ofTelecommunication Electronic and Computer Engineering vol8 no 7 pp 17ndash22 2016

[89] MM Jankovic andD B Popovic ldquoAnEMGsystem for studyingmotor control strategies and fatiguerdquo in Proceedings of the 10thSymposium on Neural Network Applications in Electrical Engi-neering (NEUREL-rsquo10) pp 15ndash18 Belgrade Serbia September2010

[90] T N S T Zawawi A R Abdullah I Halim E F Shair andS M Salleh ldquoApplication of spectrogram in analysing elec-tromyography (EMG) signals of manual liftingrdquo ARPN Journalof Engineering andApplied Sciences vol 11 no 6 pp 3603ndash36092016

[91] M Yochum T Bakir R Lepers and S Binczak ldquoEstimationof muscular fatigue under electromyostimulation using CWTrdquoIEEE Transactions on Biomedical Engineering vol 59 no 12 pp3372ndash3378 2012

[92] J L Dantas T V Camata M A O C Brunetto A C MoraesT Abrao and L R Altimari ldquoFourier and Wavelet spectralanalysis of EMG signals in isometric and dynamic maximaleffort exerciserdquo in Proceedings of the 32nd Annual InternationalConference of the IEEEEngineering inMedicine andBiology Soci-ety (EMBC rsquo10) pp 5979ndash5982 IEEE Buenos Aires ArgentinaSeptember 2010

[93] L Ai J Nan and J Shen ldquoApplication of S-transform to drivingfatigue in EEG analysisrdquo in Proceedings of the IEEE InternationalConference on Intelligent Computing and Intelligent Systems(ICIS rsquo10) pp 814ndash817 IEEE Guilin China October 2010

[94] R Upadhyay P K Padhy and P K Kankar ldquoEEG artifactremoval and noise suppression by discrete orthonormal S-transform denoisingrdquo Computers and Electrical Engineeringvol 53 pp 125ndash142 2016

[95] C Dora and P K Biswal ldquoRobust ECG artifact removalfrom EEG using continuous wavelet transformation and linearregressionrdquo in 2016 International Conference on Signal Process-ing and Communications (SPCOM) pp 1ndash5 Bangalore IndiaJune 2016

12 BioMed Research International

[96] S Assous and B Boashash ldquoEvaluation of themodified S-trans-form for time-frequency synchrony analysis and source locali-sationrdquo Eurasip Journal on Advances in Signal Processing vol2012 article 49 2012

[97] A L Ricamato R G Absher M T Moffroid and J P Tra-nowski ldquoA time-frequency approach to evaluate electromyo-graphic recordingsrdquo in Proceedings of the 5th Annual IEEESymposium on Computer-Based Medical Systems pp 520ndash527IEEE Durham NC USA 1992

[98] M Khezri and M Jahed ldquoAn inventive quadratic time-freq-uency scheme based on Wigner-Ville distribution for classifi-cation of sEMG signalsrdquo in Proceedings of the 6th InternationalSpecial Topic Conference on Information TechnologyApplicationsin Biomedicine pp 261ndash264 IEEE Tokyo Japan November2007

[99] A Subasi and M K Kiymik ldquoMuscle fatigue detection in EMGusing time-frequency methods ICA and neural networksrdquoJournal of Medical Systems vol 34 no 4 pp 777ndash785 2010

[100] W Martin and P Flandrin ldquoWigner-Ville Spectral Analysisof Nonstationary Processesrdquo IEEE Transactions on AcousticsSpeech and Signal Processing vol 33 no 6 pp 1461ndash1470 1985

[101] H-I Choi and W J Williams ldquoImproved time-frequency rep-resentation of multicomponent signals using exponential ker-nelsrdquo IEEE Transactions on Acoustics Speech and Signal Pro-cessing vol 37 no 6 pp 862ndash871 1989

[102] G R Pereira L F de Oliveira and J Nadal ldquoReducing crossterms effects in the Choi-Williams transform of mioelectricsignalsrdquo Computer Methods and Programs in Biomedicine vol111 no 3 pp 685ndash692 2013

[103] M Alemu D K Kumar and A Bradley ldquoTime-frequency anal-ysis of SEMGmdashwith special consideration to the interelectrodespacingrdquo IEEE Transactions on Neural Systems and Rehabilita-tion Engineering vol 11 no 4 pp 341ndash345 2003

[104] P A Karthick G Venugopal and S Ramakrishnan ldquoAnalysis ofsurface EMG signals under fatigue and non-fatigue conditionsusing B-distribution based quadratic time frequency distribu-tionrdquo Journal of Mechanics in Medicine and Biology vol 15 no2 Article ID 1540028 2015

[105] V Sucic B Barkat and B Boashash ldquoPerformance evaluationof the B-distributionrdquo in Proceedings of the 5th InternationalSymposium on Signal Processing and Its Applications (ISSPA rsquo99)pp 267ndash270 Queensland Australia August 1999

[106] PAKarthick and S Ramakrishnan ldquoSurface electromyographybasedmuscle fatigue progression analysis usingmodified B dis-tribution time-frequency featuresrdquo Biomedical Signal Processingand Control vol 26 pp 42ndash51 2016

[107] B Boashash and V Sucic ldquoResolution measure criteria for theobjective assessment of the performance of quadratic time-frequency distributionsrdquo IEEE Transactions on Signal Process-ing vol 51 no 5 pp 1253ndash1263 2003

[108] S Dong G Azemi and B Boashash ldquoImproved characteriza-tion of HRV signals based on instantaneous frequency featuresestimated from quadratic time-frequency distributions withdata-adapted kernelsrdquoBiomedical Signal Processing andControlvol 10 no 1 pp 153ndash165 2014

[109] B Boashash and T Ben-Jabeur ldquoDesign of a high-resolutionseparable-kernel quadratic TFD for improving newborn healthoutcomes using fetal movement detectionrdquo in Proceedings ofthe 11th International Conference on Information Science SignalProcessing and their Applications (ISSPA rsquo12) pp 354ndash359Quebec Canada July 2012

Submit your manuscripts athttpswwwhindawicom

Stem CellsInternational

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MEDIATORSINFLAMMATION

of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Behavioural Neurology

EndocrinologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Disease Markers

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

BioMed Research International

OncologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Oxidative Medicine and Cellular Longevity

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

PPAR Research

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Immunology ResearchHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

ObesityJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Computational and Mathematical Methods in Medicine

OphthalmologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Diabetes ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Research and TreatmentAIDS

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Gastroenterology Research and Practice

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Parkinsonrsquos Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttpwwwhindawicom

Page 12: EMG Processing Based Measures of Fatigue Assessment during …downloads.hindawi.com/journals/bmri/2017/3937254.pdf · 2019-07-30 · BioMedResearchInternational 3 5. Electromyography

12 BioMed Research International

[96] S Assous and B Boashash ldquoEvaluation of themodified S-trans-form for time-frequency synchrony analysis and source locali-sationrdquo Eurasip Journal on Advances in Signal Processing vol2012 article 49 2012

[97] A L Ricamato R G Absher M T Moffroid and J P Tra-nowski ldquoA time-frequency approach to evaluate electromyo-graphic recordingsrdquo in Proceedings of the 5th Annual IEEESymposium on Computer-Based Medical Systems pp 520ndash527IEEE Durham NC USA 1992

[98] M Khezri and M Jahed ldquoAn inventive quadratic time-freq-uency scheme based on Wigner-Ville distribution for classifi-cation of sEMG signalsrdquo in Proceedings of the 6th InternationalSpecial Topic Conference on Information TechnologyApplicationsin Biomedicine pp 261ndash264 IEEE Tokyo Japan November2007

[99] A Subasi and M K Kiymik ldquoMuscle fatigue detection in EMGusing time-frequency methods ICA and neural networksrdquoJournal of Medical Systems vol 34 no 4 pp 777ndash785 2010

[100] W Martin and P Flandrin ldquoWigner-Ville Spectral Analysisof Nonstationary Processesrdquo IEEE Transactions on AcousticsSpeech and Signal Processing vol 33 no 6 pp 1461ndash1470 1985

[101] H-I Choi and W J Williams ldquoImproved time-frequency rep-resentation of multicomponent signals using exponential ker-nelsrdquo IEEE Transactions on Acoustics Speech and Signal Pro-cessing vol 37 no 6 pp 862ndash871 1989

[102] G R Pereira L F de Oliveira and J Nadal ldquoReducing crossterms effects in the Choi-Williams transform of mioelectricsignalsrdquo Computer Methods and Programs in Biomedicine vol111 no 3 pp 685ndash692 2013

[103] M Alemu D K Kumar and A Bradley ldquoTime-frequency anal-ysis of SEMGmdashwith special consideration to the interelectrodespacingrdquo IEEE Transactions on Neural Systems and Rehabilita-tion Engineering vol 11 no 4 pp 341ndash345 2003

[104] P A Karthick G Venugopal and S Ramakrishnan ldquoAnalysis ofsurface EMG signals under fatigue and non-fatigue conditionsusing B-distribution based quadratic time frequency distribu-tionrdquo Journal of Mechanics in Medicine and Biology vol 15 no2 Article ID 1540028 2015

[105] V Sucic B Barkat and B Boashash ldquoPerformance evaluationof the B-distributionrdquo in Proceedings of the 5th InternationalSymposium on Signal Processing and Its Applications (ISSPA rsquo99)pp 267ndash270 Queensland Australia August 1999

[106] PAKarthick and S Ramakrishnan ldquoSurface electromyographybasedmuscle fatigue progression analysis usingmodified B dis-tribution time-frequency featuresrdquo Biomedical Signal Processingand Control vol 26 pp 42ndash51 2016

[107] B Boashash and V Sucic ldquoResolution measure criteria for theobjective assessment of the performance of quadratic time-frequency distributionsrdquo IEEE Transactions on Signal Process-ing vol 51 no 5 pp 1253ndash1263 2003

[108] S Dong G Azemi and B Boashash ldquoImproved characteriza-tion of HRV signals based on instantaneous frequency featuresestimated from quadratic time-frequency distributions withdata-adapted kernelsrdquoBiomedical Signal Processing andControlvol 10 no 1 pp 153ndash165 2014

[109] B Boashash and T Ben-Jabeur ldquoDesign of a high-resolutionseparable-kernel quadratic TFD for improving newborn healthoutcomes using fetal movement detectionrdquo in Proceedings ofthe 11th International Conference on Information Science SignalProcessing and their Applications (ISSPA rsquo12) pp 354ndash359Quebec Canada July 2012

Submit your manuscripts athttpswwwhindawicom

Stem CellsInternational

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MEDIATORSINFLAMMATION

of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Behavioural Neurology

EndocrinologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Disease Markers

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

BioMed Research International

OncologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Oxidative Medicine and Cellular Longevity

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

PPAR Research

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Immunology ResearchHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

ObesityJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Computational and Mathematical Methods in Medicine

OphthalmologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Diabetes ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Research and TreatmentAIDS

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Gastroenterology Research and Practice

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Parkinsonrsquos Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttpwwwhindawicom

Page 13: EMG Processing Based Measures of Fatigue Assessment during …downloads.hindawi.com/journals/bmri/2017/3937254.pdf · 2019-07-30 · BioMedResearchInternational 3 5. Electromyography

Submit your manuscripts athttpswwwhindawicom

Stem CellsInternational

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MEDIATORSINFLAMMATION

of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Behavioural Neurology

EndocrinologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Disease Markers

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

BioMed Research International

OncologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Oxidative Medicine and Cellular Longevity

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

PPAR Research

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Immunology ResearchHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

ObesityJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Computational and Mathematical Methods in Medicine

OphthalmologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Diabetes ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Research and TreatmentAIDS

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Gastroenterology Research and Practice

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Parkinsonrsquos Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttpwwwhindawicom