SPS 2011
Jachranka VillageJune 8-10, 2011
SPS 2011 Technical Committee
Program Committee Chair:
Krzysztof S. Kulpa Warsaw University of Technology, Warszawa, Poland
Symposium Coordinator:
Piotr Samczynski Warsaw University of Technology, Warszawa, Poland
Conference Assistant:
Anna Kurowska Warsaw University of Technology, Warszawa, Poland
TPC Members:
Christopher Baker The Australian National University, Canberra, Australia
Fabrizio Berizzi University of Pisa, Pisa, Italy
Bohdan Butkiewicz Warsaw University of Technology, Warszawa, Poland
Jaroslav Cechak University of Defence, Brno, Czech Republic
Zbigniew Czekala Radwar, Warszawa, Poland
Katarzyna Dabrowska-Zielinska Institute of Geodesy and Cartography, Warszawa, Poland
Adam Dabrowski Poznan University of Technology, Poznan, Poland
Andrzej Dabrowski Warsaw University of Technology, Warszawa, Poland
Enzo Dalle-Mese University of Pisa, Pisa, Italy
Anna Dzvonkovskaya Hamburg University of Technology, Hamburg, Germany
Andrzej Jakubiak Warsaw University of Technology, Warszawa, Poland
Hristo Kabakchiev Sofia University, Sofia, Bulgaria
Ryszard Katulski Gdansk University of Technology, Gdansk, Poland
Adam Kawalec Military University of Technology, Warszawa, Poland
Wieslaw Klembowski Przemyslowy Instytut Telekomunikacji S.A., Warszawa, Poland
Boris Lewitas Geozondas Ltd, Vilnius, Lithuania
Konstantin Lukin Institute of Radiophysics & Electronics, NASU, Kharkov, Ukraine
Mateusz Malanowski Warsaw University of Technology, Warszawa, Poland
Paulo Marques Instituto Superior de Engenharia de Lisboa, Lisboa, Portugal
Jacek Misiurewicz Warsaw University. of Technology, Warszawa, Poland
Dalius Navakauskas Vilnius Gediminas Technical University, Vilnius, Lithuania
Daniel O'Hagan Fraunhofer-FHR, Wachtberg, Germany
Piotr Orleanski Space Research Center of Polish Academy of Sciences, Warszawa, Poland
Jerzy Pietrasinski Military University of Technology, Warszawa, Poland
Joachim Schiller Fraunhofer-FHR, Wachtberg, Germany
Edward Sedek Przemyslowy Instytut Telekomunikacji S.A., Warszawa, Poland
Maciej Smolarczyk Przemyslowy Instytut Telekomunikacji S.A., Warszawa, Poland
Miroslaw Swiercz Bialystok University Of Technology, Bialystok, Poland
Krzysztof Wesolowski Poznan University of Technology, Poznan, Poland
Felix Yanovsky National Aviation University, Kiev, Ukraine
Editors:
Krzysztof Kulpa
Anna Kurowska
Piotr Samczyński
Janusz Kulpa
Wojciech Kaska
Institute of Electronic Systems
ul. Nowowiejska 15/19
00-665 Warsaw, Poland
phone: +48 22 2347478
fax: +48 22 8252300
e-mail: [email protected]
web: www.sps2011.irs.edu.pl
Instytut Systemów Elektronicznych, Politechnika Warszawska
Institute of Electronic Systems, Warsaw University of Technology
http://www.ise.pw.edu.pl
Radiolocation and Digital Signal Processing
Students' Research Group
Warsaw University of Technology
http://www.ise.pw.edu.pl/rdsp/
SIGNAL PROCESSING SYMPOSIUM, 8-10 JUNE 2011 JACHRANKA, POLAND
June 7th 2011 Tuesday
18:00 - 20:00 REGISTRATION
June 8th 2011 Wednesday
08:30 - 10:30 REGISTRATION
10:30 - 11:50 SESSION A - OPENNING SESSION Krzysztof Kulpa,
A_1 (273) Current Research in Polish Radar Industry - Zenon Szczepaniak,
A_2 (275) Usage of Airborne SAR System for Rescue Operations during Natural Disasters - Maciej Smolarczyk,
A_3 (224) Historical Overview and Recent Results in Weather Radar Signal Processing - Felix Yanovsky,
11:50 - 12:10 COFFEE BREAK
12:10 - 13:30 SESSION A - PASSIVE RADARS - FOCUS SESSION
Mateusz Malanowski, Pavlo Vyplavin,
A_1 (91) Synthetic Aperture Radar using emitters of opportunity - Adam Kawalec, Jerzy Pietrasinski, Piotr Serafin, Czesław Leśnik, Pyc Michal,
A_2 (32) Real Time Processing of Networked Passive Coherent Location Radar System - Mathew John, Michael Inggs, Dario Petri,
A_3 (127) System Concept of WIFI Based Passive Radar - Stanisław Rzewuski, Krzysztof
Kulpa,
A_4 (262) Trial results on non-cooperative bistatic radar - Michal Wilkowski, Piotr
Samczynski, Krzysztof Kulpa,
A_5 (263) Target tracking using FM-based passive radar - Mateusz Malanowski,
SESSION B - IMAGE & VOICE PROCESSING
Ibrahim Abu-Isbeih, Antoni Grzanka,
B_1 (189) Objective video quality metric based on mutual information and Human Visual System - Michal Mardiak, Jaroslav Polec,
B_2 (94) GOP structure adaptable to position of shot cuts - Lenka Krulikovská, Jaroslav Polec,
B_3 (126) Recognition of human emotion from a speech signal based on Plutchik\'s model. - Dorota Kamińska, Adam Pelikant,
B_4 (190) Smoke detection in video - Rikhard Bogush, Nadezhda Brovko,
13:30 - 15:00 LUNCH
15:00 - 16:50 SESSION C - POSTER Mikheil Chikhradze, Piotr Serafin,
C_1 (30) Shape Independent Coding Using NURBS and Bézier Approximation and Interpolation - Sandra Ondrusova, Jaroslav Polec,
C_2 (33) A Zero-Text Watermarking Algorithm Using English Language Consonants -
BUSHRA JALIL, ZUNERA JALIL,
C_3 (124) Nonparametric filter for voice signals - Liliia Kolchenko, Rustem Sinitsyn,
C_4 (37) Predictive Maintenance Sensors Placement by Combinatorial Optimization - Daneila Borissova,
C_5 (45) Improving of dielectric resonator coupling with rectangular waveguide - Borys
Pratsiuk, Kostiantyn Savin, Pavlo Sergienko, Yuriy Prokopenko ,
C_6 (51) Survey of single-photon sources and their applicability to quantum key distribution - Makhamisa Senekane, Francesco Petruccione,
C_7 (61) Controlling Multi-agent System for Sensor Networks - Software Architecture Modeling and Diagnostics - Dimitar Kamenov, Vassil Sgurev, Lyubka Doukovska,
C_8 (70) Custom energy partition between diffracted waves employing surface plasmon-polariton resonance - Mykhailo Tymchenko, Ivan Spevak, Alexander Kats,
C_9 (71) Clustering Based Voiced/Unvoiced Decision for Speech Signals - Mojtaba
Radmard, Mahdi Hadavi, Shahrokh Ghaemmaghami, Muhammad Mahdi Nayebi,
C_10 (120) Markov chain error generator for wireless channels - Juraj Pavlovič, Jaroslav Polec, Ján Poctavek, Kvetoslava Kotuliaková,
C_11 (122) The universal environment for creation and translation of intelligent computer tutoring programs - Stanislav Pedan, Andriy Chukhray,
C_12 (123) FPGA Implementation of Antenna Selection for Maximum Ratio Combining MIMO System - Yousef Alzahrani, NAWAF ALMUTAIRI,
C_13 (187) Carrier tracking loop for deeply coupled GNSS receiver - Petr Bojda,
C_14 (192) Intelligibility Threshold for Cued Speech in H.264 Video - Jaroslav Polec, Petra
Heribanova, Ján Poctavek,
C_15 (197) Recurrent Neural Networks for Predictive Maintenance of Mill Fan Systems - Petia Koprinkova-Hristova, Mincho Hadjiski, Lyubka Doukovska, Simeon Beloreshki,
C_16 (199) Electromechanical Tuning of Microstrip Filters - Pavlo Sergienko, Kostiantyn Savin, Borys Pratsiuk, Yuriy Prokopenko ,
C_17 (208) Remote Sensing Unit for Monitoring and Control of Laser Welding in Industry - Simon Rerucha, Libor Mrna, Martin Sarbort, Petr Jedlicka,
C_18 (209) A New Criterion for Antenna Placement in MIMO Passive Coherent Location -
Mojtaba Radmard, Seyyed Muhammad Karbasi, Babak Hosein Khalaj, Muhammad Mahdi Nayebi,
C_19 (211) Application of the Reed-Solomon code for correcting errors in a text - Oksana
Volivach,
C_20 (225) Controlling Multi-agent System for Sensor Networks - Software Architecture Modeling and Diagnostics - Vassil Sgurev, Dimitar Kamenov, Lyubka Doukovska,
C_21 (234) Digital Vibration Sensor Constructed with MEMS Technology - Gustaw Mazurek,
C_22 (36) Simulation of electroencephalographic signals for depth of anesthesia assessment - Anton Popov, Oleg Panichev, Oleg Bodilovskyi,
C_23 (270) Biophysical parameters assessed from microwave and optical data - Maria Budzynska, Katarzyna Dabrowska-Zielinska, Wanda Kowalik, Iwona Malek, Konrad Turlej,
15:00 - 16:00 SESSION B - SPECIAL MEASUREMENT SESSION
B_1 (265) Generation and analysis of radar signals using Tektronix instruments - Tadeusz Asyngier, Lukasz Krainski,
16:50 - 18:00 SESSION A - FUZZY LOGIC -
FOCUS SESSION Bohdan Butkiewicz, Lyubka
Doukovska,
A_1 (25) Fuzzy Model of 16PSK and 16QAM Modulation - Bohdan Butkiewicz,
A_2 (35) Estimation of heart rate variability fluctuations by wavelet transform - Anton Popov, Yevgeniy Karplyuk,
A_3 (16) Learning system based on ontological approach and fuzzy logic method - Olga Morozova,
A_4 (195) Approaches for Diagnostic and Predictive Maintenance - Kosta Boshnakov, Venko Petkov, Lyubka Doukovska, Daneila Borissova, Stefan Kojnov,
SESSION B - RADAR I Theresa Haumtratz, Maciej
Smolarczyk,
B_1 (125) Doppler processing of UWB signals - Boris Levitas, Jonas Matuzas,
B_2 (193) Detection of the Radar Echo from the Helicopter Rotor - Lech Raczynski, Zbigniew Czekala,
B_3 (198) Challenges for Non-Cooperative Target Identification in a bistatic radar configuration - Theresa Haumtratz, Joachim Schiller, Stefan Lindenmeier,
B_4 (220) Multipath Effect in Multilateration Surveillance System - Inna Konchenko, Felix Yanovsky,
20:00 - SYMPOSIUM BARBECUE
June 9th 2011 Thursday
09:20 - 10:30 SESSION A - METEOROLOGICAL RADAR SIGNAL PROCESSING AND
MODELING - FOCUS SESSION Felix Yanovsky, Edward Sedek,
A_1 (31) Double frequency method for measurement of rain intensity based on contact measurements of microphysics rain parameters - Anna Linkova,
A_2 (44) Estimation of the meteorological formations parameters in pulsed Doppler weather radars with arbitrary staggering of pulse repetition intervals - Dmytro Rachkov,
Andrii Semeniaka, David Lekhovytskiy,
A_3 (194) Least square spline decomposition in time-frequency analysis of meteorological signals - Kseniia Shelevytska, Oksana Semenova, Felix Yanovsky,
A_4 (235) COMPARATIVE ANALYSIS OF TWO HAIL DETECTION ALGORITHMS - I.
Braun, Rustem Sinitsyn, Felix Yanovsky,
SESSION B - RADAR II / SIGNAL PROCESSING I
Zbigniew Czekala, Christo Kabakchiev,
B_1 (28) Design of Pareto-Optimal Radar Receive Filters - Marco Piezzo, Antonio De Maio, Salvatore Iommelli, Alfonso Farina,
B_2 (22) Phase discontinuity in MIMO radars - Pier Francesco Sammartino, Raimondo
Giuliani, Dario Tarchi,
B_3 (56) A nearly optimal fractional delay filter design with asymmetric window -
Maciej Sac, Marek Blok,
B_4 (233) SYMDAS - Raw Data Simulator for the Radar with Electronically Controlled Beam - Gustaw Mazurek,
10:30 - 11:00 COFFEE BREAK
11:00 - 13:00 SESSION A - NOISE RADARS - FOCUS SESSION Krzysztof Kulpa,
A_1 (34) Evaluation of IQ imbalance compensation algorithms in a noise radar with analog IQ direct downconverters - Michal Meller, Lukasz Cwikowski, Pawel Paprocki,
A_2 (205) Application of Arbitrary Waveform Generator for Noise Radar - Konstantin Lukin, Oleg Zemlyaniy, Pavlo Vyplavin, Volodymyr Palamarchuk,
SESSION B - SIGNAL PROCESSING II
Marek Blok, Abdel-Rahman Al-Qawasmi,
B_1 (18) Noise and Artifacts removal from Electrocardiograms utilizing Significant Features - BUSHRA JALIL, ERIC FAUVET,
B_2 (266) CAD of Nanocircuits on quantum cellular automata(corrected) - Volodymyr Ivakhnyuk, Oleksandr Melnik,
B_3 (226) Nonlinear Trend Analysis for
A_3 (206) Experimental investigation of factors affecting stability of interferometric measurements with ground based noise waveform SAR - Konstantin Lukin, Volodymyr
Palamarchuk, Pavlo Vyplavin, Volodymyr Kudriashov,
A_4 (210) High Resolution Noise Radar without fast ADC - Konstantin Lukin, Pavlo Vyplavin, Oleg Zemlyaniy, Volodymyr Palamarchuk, Sergii Lukin,
A_5 (218) Software Defined Noise Radar with Low Sampling Rate - Pavlo Vyplavin,
Sergii Lukin, Elena Savkovich, Konstantin Lukin,
A_6 (243) Pseudo-noise waveform design minimizing range and Doppler masking e ect - Janusz Kulpa, Jacek Misiurewicz,
Diagnostics and Predictive Maintenance - Mincho Hadjiski, Lyubka Doukovska, Stefan Kojnov, Dimitar Kamenov,
B_4 (228) Estimation of two sinusoids in a very short signal - Rafał Rytel-Andrianik,
13:00 - 14:30 LUNCH
14:30 - 16:00 SESSION A - RADAR SIGNAL
PROCESSING I Michal Meller, Fabrizio Berizzi,
A_1 (15) Adaptive Detection in Presence of Mutual Coupling and Interfering Signals - Silvio De Nicola, Antonio De Maio, Alfonso Farina, Michele FIORINI, Leo Infante, Marco Piezzo,
A_2 (29) Exploiting Knowledge for Transmit Signal and Receive Filter Design in Signal-Dependent Clutter - Augusto Aubry, Antonio De Maio, Alfonso Farina, Michael Wicks,
A_3 (97) CFAR BI Detector for Mariner Targets in Time Domain for Bistatic Forward Scattering Radar - Christo Kabakchiev, Ivan Garvanov, M. Cherniakov, M. Gashinova, A. Kabakchiev, Vladimir Kiovtorov, M. Vladimirova, P. Daskalov,
A_4 (221) RADAR Signal Parameters Estimation in the MTD Tasks - Igor
Prokopenko, I. P. Omelchuk, Y.D. Chirka,
SESSION B - RADIO-
COMMUNICATIONS TECHNOLOGY / TELECOMMUNICATIONS
Jaroslav Cechak, Rafał Rytel-Andrianik,
B_1 (21) Modulation-Mode Assignment in SVD-Aided Downlink Multiuser MIMO-OFDM Systems - Aust Sebastian, Andreas
Ahrens,
B_2 (42) Calculation of the areas most likely to appear suitable for communication meteor trails - Helen Kharchenko,
B_3 (60) DSP Algorithms in Analysis of Radio Channel Characteristics - Kamil
Bechta,
B_4 (222) The Effect of Cell Phones on Human Health - Ibrahim Abu-Isbeih, Dina Saad,
B_5 (267) New FEC encoding technique based Parity Selected Codes for 4-ary PAM Signal - Abdel-Rahman Al-Qawasmi, Aiman Al-
Lawama,
16:00 - 16:30 COFFEE BREAK
16:30 - 17:45 SESSION A - REMOTE SENSING Piotr Samczynski, Anton Popov,
A_1 (74) The ISAR Image Post-Processing for Multi-Point Target Identification - Maxim
Konovalyuk, Anastasia Gorbunova, Yury Kuznetsov, Andrey Baev,
A_2 (271) CHANGE DETECTION FOR SAR IMAGERY USING CONNECTED COMPONENTS ANALYSIS - Artur Gromek,
Małgorzata Jenerowicz,
A_3 (104) Improving the consistency and continuity of MODIS 8 days Leaf Area Index Product - Sivasathivel Kandasamy, Philippe Neveux, Aleixandre Verger, Samuel Buis, Marie Weiss, Frederic Baret,
A_4 (247) Preliminary results of ground reflectivity measurements using noise radar - Łukasz Maślikowski, Piotr Krysik, Katarzyna Dabrowska-Zielinska, Wanda Kowalik, Maciej Bartold,
A_5 (264) Linear landmark extraction in SAR images with application to augmented integrity aero-navigation: an overview to a novel processing chain - Luca Fabbrini, M.
Messina, Mario Greco, Pinelli Gianpaolo,
SESSION B - NAVIGATION TECHNOLOGY/SPACE
TECHNOLOGY Mikheil Chikhradze, Zenon
Szczepaniak,
B_1 (62) Wireless Device for Activation an Underground Shock Wave Absorber - Mikheil Chikhradze,
B_2 (261) SAR and InSAR georeferencing algorithms for Inertial Navigation Systems
- Mario Greco, Krzysztof Kulpa, Pinelli Gianpaolo, Piotr Samczynski,
B_3 (63) A time-domain model of GPR antenna radiation pattern - Alexei Popov, Fedor Edemsky,
B_4 (268) Information and navigation system for person with visual impairment - Jaroslav Cechak,
B_5 (207) GTAG: Architecture and Design of Miniature Transmitter with Position Logging for Radio Telemetry - Simon Rerucha, Tomas Bartonicka, Petr Jedlicka,
18:00 - 18:20 SESSION A - CLOSSING SESSION Krzysztof Kulpa,
A_1 (212) Historical overview and current research of the Noise Radar - Konstantin Lukin,
Ram Narayanan, Pavlo Vyplavin,
20:00 - SYMPOSIUM BARBECUE
June 10th 2011 Friday
08:00 - 10:00 CHECK OUT & DEPARTURE
Simulation of electroencephalographic signals for depth of anesthesia assessment
Oleg Panichev, Anton Popov*, Oleg Bodilovskyi, Valerii Tkachenko
off. 423, Politekhnichna Str. 16, 03056, Kyiv, Ukraine Department of Physical and Biomedical Electronics,
National Technical University of Ukraine “Kyiv Polytechnic Institute”
ABSTRACT The problem of simulating electroencephalographic (EEG) signals for different stages of anesthesia is considered. Review of existing techniques for EEG simulation is made and the new technique for simulating the EEG using only average magnitude of harmonic components for predefined frequency ranges is presented. During experimental part EEG signals for four stages of ether anesthesia were simulated.
Keywords: EEG, depth of anesthesia, EEG simulation, narcosis
1. INTRODUCTION Despite the high level and variety of diagnostic equipment in modern surgical operating rooms, there is still no single indicator that would adequately assess depth of anesthesia (DA) of a patient. The depth of anesthesia is currently measured by many indicators of the patient’s life simultaneousely, such as pupillary reflex, parameters of the cardiovascular system (heart rate, blood pressure, parameters of electrocardiogram etc.), analysis of expired gases. These parameters separately are not very reliable for evaluation of depth of narcosis because they depend on many factors, including age, sex, nature of human disease and nature of surgical intervention. The DA also depends of the type of anesthetic drugs used, type and scheme of intervention 1-3. All these make anesthesia still hardly controlled process with many influencing factors. Therefore, the invention of a universal criterion of depth of anesthesia is a priority. It is known that the human brain is responsible for consciousness and unconsciousness, and thus for the depth of anesthesia. Thus to find a more appropriate unified criterion of DA monitoring, the estimation of a central nervous system state is widely used, mostly by means of brain electrical activity analysis with electroencephalography.
Many techniques of electroencephalogram (EEG) analysis are used for creating the parameters for DA but there is still a need for more sophisticated measures of evaluating the DA. One of the tasks in developing such elaborated techniques is its testing with the EEG signals corresponding to the different brain states. Full sets of signals corresponding to the various states and conditions affecting EEG during anesthesia are rarely available for the analysis. The aim of this work is to develop the technique for simulation of EEG signal for different stages of brain activity during the transition from consciousness to deep narcosis.
2. SIMULATION OF EEG SIGNALS DURING NARCOSIS 2.1. SELECTION OF SIMULATION TECHNIQUE
Simulation of EEG signal starts from the decision about what parameters of real EEG should be reproduced in the simulated signal. Depending on this, various techniques could be applied: autoregression, physiological modeling of EEG as the activity of artificial neurons, statistical simulation of EEG as noise signals 4-7.
EEG signal could be simulated as an output signal of some linear system with white noise input. In the most common autoregressive approach each EEG sample is defined as a weighted sum of previous samples. The main task is to define the system parameters, i.e. the coefficients of difference equation. Several EEG channels could be simulated in the framework of multivariate autoregressive approach, where samples of each channel depend on the previous samples in all channels. The problem in using this approach consists of finding of appropriate linear system parameters for different stages of narcosis and in its on-line adjusting for reflecting the transitions between stages. Moreover this approach could be used for linear models only, while EEG during anesthesia might have nonlinear characteristics.
* [email protected]; phone +380 50 231-6448; www.phbme.kpi.ua
Nonlinear technique such as generalized autoregressive conditional heteroskedasticity approach could be promising for this case.
EEG signal can be also modeled as the random quasistationary signal, i.e. considered to be stationary and Gaussian only within short intervals. But there is no evidences about whether the EEG during anesthesia is stationary, what are probability density distributions and intervals of stationarity. It is believed now that during changes in fuctional state of the brain Gaussian assumption is rarely valid. Thus a lot of work is to be done for adequately simulate the EEG during transition between anesthesia stages as a random signal.
Physiological modeling of large sets of spiking neurons also could be used in the simulation of EEG. The basis of this approach is classical Hodgkin-Huxley model of action potential of single neurons by estimating the ion currents through voltage-driven channels. This model describes time change of membrane potential depending on the initial membrane potentials, characteristics of ion channels etc. More complex models could be constructed for the neurons interconnections and signal transfer through synapse and for the neuronal networks. The signals generated by these models could be used as a model of EEG signal while appropriate parameters and number of neurons are selected. Beside this kind of models are considered to be potentially the most adequate to the real brain electrical activity, its use is complicated because of the need of huge amount of parameters identification and resulting large-scale differential equation systems.
Now there are four more or less established electroencephalographic stages of narcosis: 1) analgesia; 2) excitation; 3) surgical (consisting of 3 sub-stages during which the surgery should be conducted); 4) agony (terminal stage). All these stages are achievable with administrating the appropriate amount of anesthetic drugs. There are clinical data about brain electrical activity during each stage. For example, the parameters of EEG signal for various stage of Ether anesthesia are given in Table 1, the same parameters for Ftorothan are given in Table 2.
The aim of this work is to represent the properties of EEG, which are important and useful for accessing the DA in the simulated signal. Thus the first task is to find what parameters are used for evaluating the DA and anesthetic drugs administration control.
Thus in the case considered here only the average magnitudes of oscillations in some frequency ranges are known, thus we have very limited information about brain electrical activity during anesthesia. The simulation technique for this case should use only the available parameters, that’s why no method mentioned above could be used without additional knowledge of EEG parameters.
Table 1 – Stages of anesthesia and corresponding EEG parameters for the case of ether administration.
2.2. EEG SIMULATION TECHNIQUE
The proposed approach to EEG simulation is based on the technique previously described in 8, where simulation was done by setting the average power spectral density for each EEG frequency band. In the case considered here only the average magnitudes for the frequency bands are given for all stages of anesthesia, thus the previous technique needs to be modified as follows. Oscillations in every EEG’s frequency range are constructed separately as the sum of harmonics with the frequencies prescribed by the corresponding range limits. For the sake of simplicity the values of spectral amplitudes r nS f in every rhythm are defined as constants over the whole range interval:
, ,
0,r n rs re
r n
a if f F FS f
otherwise
, (1)
where nf frequency of the harmonic component; ra – the average value of harmonic magnitude; rsF and reF are
starting and end frequency of every EEG range, Hz.
Total number of harmonics for the range is r re rsN F F T . The whole amplitude spectrum of simulated EEG is
constructed as the sum of separate r nS f for all ranges with indexes r , , ,r :
n r nr
S f S f
. (2)
To calculate the samples of EEG signal in time domain, the representation of the signal as the sum of harmonic oscillations with frequencies nf and amplitudes nS f (2) can be written as inverse Fourier transform:
cos 2re
n rs
F
r n n nr f F
EEG n S f f t
, (3)
where n initial phase angle. To additionally enhance the randomness inherent to EEG signals, the value of n is
selected to be random uniformly distributed numbers within the range from 0 to 2 . This will not affect the average value of the signal but will reflect the random nature of brain’s potential oscillations. Thus by using simple technique of adjusting the average magnitude of harmonics in required frequency range EEG signals with desired properties for every stage of anesthesia could be obtained. The main feature of such technique is a possibility to control the contribution of every harmonic component into resulting signal by setting appropriate amplitude. The model is
Table 2 – Stages of anesthesia and corresponding EEG parameters for the case of Ftorothan administration.
independent of sampling frequency and may have adjustable shape of harmonics magnitude’s distribution in every frequency band.
The technique described above is used for simulating only one EEG channel but it could be easily extended for constructing multichannel signals. Each EEG lead could be simulated independently from each other and may have different parameters, thus possible nonuniform surface distribution of signals with appropriate spectral components’ magnitudes is also taken into account in the proposed technique. This could be beneficial for simulation of signal from different brain regions and thus allows to more precise description of brain electric activity during anesthesia. This is useful for studying the DA assessment techniques with different number of registering electrodes and at various electrode locations. The issues of the difference in stages of EEG activity for sleep and for anesthesia are highlighted and considered in the model.
3. EXPERIMENTAL RESULTS The EEG signals for the case of ether anesthesia were simulated using the abovementioned technique. The samples of one channel of EEG for each stage from Table 1 were calculated for sampling rate 256 Hz. Normal EEG signal of conditionally healthy conscious person is presented on Fig. 1 for reference. Obtained signals for different stages of anesthesia and their power spectral densities’ estimates by Welch method are shown on Fig. 2-5. Signal corresponding to same stage (excitation and analgesia) as on Fig. 2 but for the case of Ftorothan administration is presented on the Fig. 6.
0 5 10 15-80
-60
-40
-20
0
20
40
60
80Simulated EEG signal, consciousness
Time, s
Mag
nitu
de,
uV
2 4 8 14 400
10
20
30
40
50
60PSD using Welch's method
F, Hz
PS
D,
V2 /H
z
Figure 1 – Simulated EEG signal for healthy conscious person and it’s power spectral density
0 5 10 15-80
-60
-40
-20
0
20
40
60
80Simulated EEG signal, Ether anesthesia
Time, s
Mag
nitu
de,
uV
2 4 8 14 400
10
20
30
40
50
60PSD using Welch's method
F, Hz
PS
D,
V2 /H
z
Figure 2 – Simulated EEG signal for ether anesthesia (analgesia and excitation stage) and it’s power spectral density
0 5 10 15-80
-60
-40
-20
0
20
40
60
80Simulated EEG signal, Ether anesthesia
Time, s
Mag
nitu
de,
uV
2 4 8 14 400
10
20
30
40
50
60PSD using Welch's method
F, Hz
PS
D,
V2 /H
z
Figure 3 – Simulated EEG signal for ether anesthesia (1st surgical stage) and it’s power spectral density
0 5 10 15-80
-60
-40
-20
0
20
40
60
80Simulated EEG signal, Ether anesthesia
Time, s
Mag
nitu
de,
uV
2 4 8 14 400
10
20
30
40
50
60PSD using Welch's method
F, Hz
PS
D,
V2 /H
z
Figure 4 – Simulated EEG signal for ether anesthesia (2nd surgical stage) and it’s power spectral density
0 5 10 15-80
-60
-40
-20
0
20
40
60
80Simulated EEG signal, Ether anesthesia
Time, s
Mag
nitu
de,
uV
2 4 8 14 400
10
20
30
40
50
60PSD using Welch's method
F, Hz
PS
D,
V2 /H
z
Figure 5 – Simulated EEG signal for ether anesthesia (3rd and 4th surgical stages) and it’s power spectral density
0 5 10 15-80
-60
-40
-20
0
20
40
60
80Simulated EEG signal, Ftorothan anesthesia
Time, s
Mag
nitu
de,
uV
2 4 8 14 400
10
20
30
40
50
60PSD using Welch's method
F, Hz
PS
D,
V2/H
z
Figure 6 – Simulated EEG signal for Ftorothan anesthesia (analgesia and excitation stage) and it’s power spectral density
4. CONCLUSIONS Technique for constructing simulated EEG for different brain states basing on required average magnitude of spectral components in predefined frequency ranges corresponding to different stages of anesthesia is proposed. Representation of obtained artificial EEG signal as the sum of harmonic oscillations with random initial phases is proposed. The main feature of such representation is a possibility to control the contribution of every harmonic component into resulting signal by setting appropriate amplitude. The model is independent of sampling frequency and allows adjustable shape of harmonics magnitude distribution in every frequency band.
After reviewing obtained simulated signals it can be concluded that EEG signals for different stages of anesthesia as well as for different anesthetic drugs could be simulated by application of proposed technique using only the average magnitude of harmonic components for various frequency ranges. These signals cold be used for preliminary testing of the algorithms of depth of anesthesia assessment.
5. ACKNOWLEDGMENTS This work was partially supported by Ukrainian Ministry of Education, Science, Youth and Sports under grant № 2337-п, and by UTAS Co., Kyiv, Ukraine.
6. REFERENCES 1. Buniatian A. A. et al. Anesthesiology and resuscitation, 432 p., Meditsina, Moskva, 1977. (in Russian) 2. Prior P. Monitoring cerebral function, 328 p., Meditsina, Moskva, 1983. (in Russian) 3. Popov A. O. et al. “Methods and devices for evaluation of the depth of anaesthesia”, Electronics and
communications, 1(2007), pp. 59 – 64, 2007. (in Ukrainian) 4. Sanei S., Chambers J. A. EEG Signal Processing, 313 p., John Wiley & Sons Ltd., Chichester, England, 2007. 5. Marple-Jr. S. L. Digital Spectral Analysis with Applications, 584 p., Mir, Moskva, 1990. (in Russian) 6. Al-Nashash H. A. et al. “A neural networks approach to EEG signlas modeling”, Proceedings of 25th annual
international conference of IEEE EMBS, pp. 2451 – 2454, 2003. 7. Rankin L. et al. “A nonstationary model of newborn EEG”, IEEE Transactions on biomedical engineering, 54(1),
pp. 19 – 28, 2007. 8. Panichev O., Popov A., Zhukov M. “Simulation of Electroencephalographic Signals”, MIKON-2010, Signal
Processing Workshop SPW-2010, Vilnius (Lithuania), 2010.