tutorial auditory steady-state responses · response, as well as information regarding how...

Delivered by Ingenta to: University of North CarolinaIP : 152.19.15.202 On: Mon, 09 Sep 2013 18:40:51

Tutorial

Auditory Steady-State ResponsesDOI: 10.3766/jaaa.23.3.3

Peggy Korczak*

Jennifer Smart*Rafael Delgado†

Theresa M. Strobel*

Christina Bradford*

Abstract

Background: The auditory steady state response (ASSR) is an auditory evoked potential (AEP) that canbe used to objectively estimate hearing sensitivity in individuals with normal hearing sensitivity and with

various degrees and configurations of sensorineural hearing loss (SNHL). For this reason, many audiol-ogists want to learn more about the stimulus and recording parameters used to successfully acquire this

response, as well as information regarding how accurately this response predicts behavioral thresholdsacross various clinical populations.

Purpose: The scientific goal is to create a tutorial on the ASSR for doctor of audiology (Au.D.) stu-dents and audiologists with limited (1–5 yr) clinical experience with AEPs. This tutorial is needed

because the ASSR is unique when compared to other AEPs with regard to the type of terminologyused to describe this response, the types of stimuli used to record this response, how these stimuli

are delivered, the methods of objectively analyzing the response, and techniques used to calibratethe stimuli. A second goal is to provide audiologists with an understanding of the accuracy with

which the ASSR is able to estimate pure tone thresholds in a variety of adult and pediatric clinicalpopulations.

Design: This tutorial has been organized into various sections including the history of the ASSR,unique terminology associated with this response, the types of stimuli used to elicit the response,

two common stimulation methods, methods of objectively analyzing the response, technical param-eters for recording the ASSR, and the accuracy of ASSR threshold prediction in the adult and pediatric

populations. In each section of the manuscript, key terminology/concepts associated with the ASSRare bolded in the text and are also briefly defined in a glossary found in the appendix. The tutorial

contains numerous figures that are designed to walk the reader through the key concepts associatedwith this response. In addition, several summary tables have been included that discuss various topics

such as the effects of single versusmultifrequency stimulation techniques on the accuracy of estimat-ing behavioral thresholds via the ASSR; differences, if any, in monaural versus binaural ASSR thresh-

olds; the influence of degree and configuration of SNHL on ASSR thresholds; test-retest reliability of

the ASSR; the influence of neuro-maturation on ASSR thresholds; and the influence of various tech-nical factors (i.e., oscillator placement, coupling force, and the number of recording channels) that

affect bone conducted ASSRs.

Conclusion:Most researchers agree that, in the future, ASSR testing will play an important role in clinical

audiology. Therefore, it is important for clinical audiologists and Au.D. students to have a good basicunderstanding of the technical concepts associated with the ASSR, a knowledge of optimal stimulus

and recording parameters used to accurately record this response, and an appreciation of the currentrole and/or limitations of using the ASSR to estimate behavioral thresholds in infants with various degrees

and configurations of hearing loss.

*Audiology, Speech-Language Pathology, Towson University; †Intelligent Hearing Systems Corp., Miami, FL

Peggy Korczak, Towson University, Audiology, Speech-Language Pathology, 8000 York Road, Towson, MD 21252; Phone: 410-704-5903; Fax:410-704-4131; E-mail: [email protected]

J Am Acad Audiol 23:146–170 (2012)

146


Key Words: Auditory evoked potentials, carrier frequency, diagnostic techniques, pediatric audiology,

modulation frequency, screening issues

Abbreviations: ABR5 auditory brainstem response; AEP5 auditory evoked potential; AM5 amplitudemodulated; ASSR 5 auditory steady state response; BC-ASSR 5 bone conduction ASSR; BHT 5

behavioral hearing threshold; CF 5 carrier frequency; EASSR 5 electrically evoked auditory steady

state response; EEG 5 electroencephalography; ERP 5 event-related potential; FFT 5 fast Fouriertransform; FM 5 frequency modulated; fMRI 5 functional magnetic resonance imaging; GA 5

gestational age; MDS 5 mean difference score; MEG 5 magnetoencephalography; MM 5 mixedmodulation; MF 5 modulation frequency; PC2 5 phase coherence squared; PCA 5 post conceptual

age; RE 5 relative efficiency; RN 5 residual noise; RSG 5 repeating sequence gated; SAM 5

sinusoidally amplitude modulated; SNHL 5 sensorineural hearing loss

Auditory evoked potentials (AEPs) are often used

in clinical audiology to estimate behavioral pure

tone thresholds in certain populations includ-

ing infants, young children, and individuals with intel-

lectual disabilities. An AEP is a response that can be

recorded from the brain following presentation of audi-

tory stimuli, such as clicks, tone bursts, and/or speech.

In many audiology centers, the auditory brainstemresponse (ABR) is the AEP of choice for estimating

behavioral pure tone thresholds, due to the high test-

retest reliability of this response (Lauter and Loomis,

1986, 1988; Lauter and Karzon, 1990a, b, c; Hood, 1998;

Hall, 2007) and the well-established normative data-

base for using this response to estimate behavioral

thresholds in infants, young children, and adults (e.g.,

Stapells et al, 1990, 1995; Munnerley et al, 1991; Beattieand Kennedy, 1992; Stapells, 2000, 2011). However,

more recent developments in the field suggest that a

relatively new method of recording AEPs, known as

the auditory steady state response (ASSR), is comparable

to the ABR with respect to the accuracy of estimating

pure tone thresholds and the potential to reduce testing

time. (For a general review of AEPs and their clinical

applications see Burkard et al [2007], Hall [2007], andPicton [2011]. Additionally, any term that appears in

bold below can be found in the glossary in Appendix A.)

HISTORY OF THE ASSR

Occasional reports of steady state responses to audi-

tory stimuli recorded from the human scalp have

appeared in the AEP literature in the 1960s (Geisler,

1960) and in the 1970s (Campbell et al, 1977). However,

the ASSR was first described in detail in the literature

by Galambos et al (1981). In this study, Galambosand colleagues (1981) recorded auditory brainstem re-

sponses and middle latency responses to 500 Hz tonal

stimuli presented at stimulus rates ranging from 3.3

to 55/sec in adults with normal-hearing sensitivity.

These investigators demonstrated that when the stim-

uli were presented at a rate of 40/sec, an overlap in the

positive and negative peaks of the response occurred at

approximately 25 msec intervals within the 100 msecpoststimulus analysis window (see Fig. 1A). Galambos

and colleagues plotted the amplitude of this ASSR as

a function of stimulus rate and demonstrated that for

adults the largest amplitude of this response occurred

at 40 Hz (see Fig. 1B). Therefore, these investigators

named this response the 40 Hz event-related potential

(ERP); however, this response has also been referred to

as the steady state evoked potential (Stapells et al,1984; Linden et al, 1985; Cohen et al, 1991; Rickards

et al, 1994; Rance and Rickards, 2002).

Figure 1. (A) The positive (Pa, Pb, Pc) peaks and negative (Na, Nb, Nc) troughs of the ABR and MLR overlap at about 25 msec intervalswithin the 100 msec poststimulus analysis window (modified from Galambos et al, 1981). (B) Mean response amplitudes of the ASSR as afunction of stimulus rate. Responses for adults and for children are shown (modified from Stapells et al, 1988).

ASSRs/Korczak et al

147


Galambos et al’s (1981) data revealed several useful

characteristics of the 40 Hz response. First, this response

was present at intensity levels near behavioral thresholds

and thus could be used to predict hearing sensitivity forthese adult subjects. Second, the 40Hz responsewas easy

to identify. Third, the amplitude of the 40 Hz response

remained relatively large even close to threshold.

Subsequent research in the mid- to late 1980s, how-

ever, identified two critical limitations of the 40 Hz

ERP. One limitation was that the 40 Hz response could

not be reliably recorded in infants and young children

as the peak amplitude of their ASSR occurred at ratesof approximately 20 Hz as shown in Figure 1B (Suzuki

and Kobayashi, 1984; Stapells et al, 1988). Secondly,

the presence of the 40 Hz response was dependent upon

subject state and could only be reliably recorded in

awake subjects (Linden et al, 1985; Jerger et al, 1987;

Kuwada et al, 1986; Cohen et al, 1991). These limita-

tions posed a problem for the clinical feasibility of record-

ing this response especially in the pediatric populationwho are often tested while asleep or sedated.

Restored interest in the 40 Hz ERP for adults tran-

spired years later when Cohen et al (1991) proved that

the ASSR could be reliably recorded in adults during var-

ious states of arousal when testing at higher stimulation/

modulation rates ($70 Hz). Several pediatric studies

have also demonstrated that the ASSR can be suc-

cessfully recorded in either awake or sleeping infantsand young children using fast ($70 Hz) stimulation

rates (Aoyagi et al, 1993; Rickards et al, 1994; Rance

et al, 1995). As a result of these discoveries, reference

to the 40 Hz ERP was dismissed for infants and young

children, and this AEP was now generally referred to as

the ASSR. Considerable interest, however, remained in

the clinical application of the 40 Hz response in adults.

In 2004, Pethe and colleagues sought to determinewhat modulation frequency (40 or 80 Hz) would yield

the best signal-to-noise ratios (SNR) for recording

ASSRs in young children, aged 2 mo to 14 yr. Specifi-

cally these investigators recorded responses to 1000 Hz

carrier frequency (CF) tones with modulation frequen-

cies (MFs) of 40 and 80 Hz presented at stimulus inten-

sities ranging from 10 to 50 dB nHL. Pethe et al (2004)

reported that for infants below 1 yr of age, the ampli-

tude of the 40 Hz response was approximately the same

as the amplitude of the 80 Hz response. However, by 13yr of age, the amplitude of the 40Hz responsewas almost

twice as large (i.e., at 50 dB nHL, the amplitudes of the

responseswerez150nVandz80nV for the 40 and80Hz

responses, respectively). Since the amplitude of the

residual background electroencephalography (EEG)

noise is significantly higher at 40 Hz than at 80 Hz

(van der Reijden et al, 2001), then the SNR for ASSRs

in younger children is considerable better for the higher(80 Hz) versus lower (40 Hz) MFs. Based on this data,

Pethe and colleagues (2004) concluded it appears that

13 yr of age is a critical time when the optimal MF

changes from a high- to a low-frequency range.

Given the substantial differences in the response prop-

erties of the ASSRs generated at lower (i.e., 40Hz) versus

higher (i.e.,$70Hz) stimulation rates, researchers began

to speculate as to why these differences occurred. Oneleading explanation for these rate-sensitive differences

was that the ASSR was receiving contributions from dif-

ferent underlying neural generators in the peripheral

and/or central auditory nervous system when elicited

at lower rather than higher stimulus rates.

NEURAL GENERATORS OF THE ASSR

The underlying neural generators of the ASSR

have been investigated using various types of neuro-

imaging techniques including Brain Electrical Source

Analysis, or BESA (Herdman et al, 2002);magnetoen-

cephalography, or MEG (Johnson et al, 1988; Hari

et al, 1989; Ross et al, 2000), and functional magnetic

resonance imaging, or fMRI (Giraud et al, 2000). The

neural generators of the ASSR have also been investi-gated using patients with known lesions in the auditory

cortex and/ormidbrain regions of theCANS (Spydell et al,

1985) and by conducting animal studies (Makela et al,

1990; Kiren et al, 1994; Kuwada et al, 2002).

Collectively, the results of these neural generator

studies suggest that when ASSRs are elicited by stimuli

Figure 2. 500 Hz carrier frequency tone moves through the outer and middle ear into the cochlea. The point on the basilar membranethat is best tuned to 500 Hz is then activated.

Journal of the American Academy of Audiology/Volume 23, Number 3, 2012

148


presented at rates less than 20 Hz, these responses are

mainly generated by activity in the primary auditory

cortex (Hari et al, 1989; Makela et al, 1990; Herdman

et al, 2002). When ASSRs are elicited by stimuli pre-

sented at rates between 20 and 60 Hz, the underlying

neural generators are mainly located in the primary

auditory cortex, auditory midbrain, and thalamus(Spydell et al, 1985; Johnson et al, 1988; Hari et al,

1989; Makela et al, 1990; Kiren et al, 1994; Herdman

et al, 2002). Lastly, when ASSRs are elicited by stimuli

presented at rates greater than 60 Hz, these responses

are generated primarily by contributions from the supe-

rior olivary complex, inferior colliculus, and cochlear

nucleus (Hari et al, 1989; Makela et al, 1990; Kiren

et al, 1994; Cone-Wesson, Dowell, et al, 2002; Herdmanet al, 2002; Picton et al, 2003). The results of these neu-

ral generator studies also demonstrated that ASSRs

recorded at any of these stimulation/modulation rates

receive contributions from multiple generators. How-

ever, recording parameters such as stimulus rate and

EEG bandpass filter settings can suppress contribu-

tions from certain underlying neural generators to

the final averaged response.Knowledge of the changes in theunderlyingneural gen-

erators of the ASSR as a function of stimulus/modulation

rate helps to explain the two primary limitations that

were discovered in the early research conducted on the

40 Hz response. Galambos and his colleagues (1981) were

able to successfully record robust 40 Hz responses in

awake adults with normal hearing sensitivity, as their

auditory cortices were fully mature and intact. In con-trast, the 40 Hz response was not observable in awake

infants and young children because their auditory cortices

are not fully mature. When ASSRs are elicited using high

(i.e., $70 Hz) stimulus/modulation rates, the primary

neural generators occur within the auditory brainstem

region, similar to the ABR, and thus are not affected by

subject state and age.

TERMINOLOGY ASSOCIATED WITH THE ASSR

I naddition to understanding the neural generators of

the ASSR, it is also important for audiologists to

have a working knowledge of the terminology associ-

ated with this response. Two primary terms associated

with the ASSR are carrier frequency (CF) and mod-

ulation frequency (MF). The CF of the tonal stimulus

is the test frequency of interest. The CF is associatedwith the region in the cochlea where the hair cells

are activated in response to the presentation of a stim-

ulus (Hall, 2007). For example, if a 500 Hz CF tone is

used to elicit the ASSR, the portion of the basilar mem-

brane that is activated is the one best tuned to 500 Hz

(see Fig. 2). The extent of basilar membrane excitation

that occurs in this area is dependent on stimulus inten-

sity, such that higher intensity stimuli produce a largerarea of cochlear excitation. Typical CF tones used to

record the ASSR are 500, 1000, 2000, and 4000 Hz.

The MF, in contrast, is the frequency at which the

EEG activity is synchronized to fire. This can be derived

by calculating the period of the MF. For example, if a

2000 Hz CF tone is presented with a 100 Hz MF, then

the response follows the MF at 100 Hz resulting in a

peak every 10 msec (see Fig. 3). This 10 msec intervalcorresponds to the period of the MF that can be deter-

mined by calculating the period (T) of the modulation

frequency ðT5 1f 5

1 secMF 5 1000msec

100Hz 5 10 msecÞ: Audiolo-gists can think of MF as similar to stimulus rate.

Several other terms are used with the ASSR to describe

the type of stimuli, the stimulation techniques, and the

way the response is analyzed. Themajority of these terms

are fairlyunique to thisAEPand canbe found inAppendixA. Some of the common terms used to describe the types of

stimuli are frequency modulated tones, amplitude

modulated tones, and mixed modulated tones and

are discussed in the section labeled “Types of Stimuli.”

Terms typically associated with the stimulation techni-

ques used to elicit the ASSR are the single frequency

stimulation technique and themultifrequency stim-

ulation technique,and these arediscussed in the sectionlabeled “Stimulation Techniques.” Lastly, the terminology

associated with analyzing the response includes terms

such as phase-coherence, fast Fourier transform

(FFT) analysis, and F-test, and these are discussed in

the section labeled “Methods of Analyzing Responses.”

TYPES OF STIMULI

There are several types of stimuli used to record

ASSRs. These stimuli can be generalized into two

categories: broadband (i.e., non–frequency specific) stimuli

and frequency-specific stimuli. Broadband stimuli

Figure 3. ASSR response to a 2000 Hz CF tone with a 100 HzMF. The response follows the MF at 100 Hz resulting in a peakevery 10 msec (modified from Grason-Stadler Inc, 2001).

ASSRs/Korczak et al

149


encompass a range of frequencies and include clicks,

noises, and chirps. In contrast, frequency-specific stimuli

include filtered clicks, tone bursts, pure tones, and

band-limited chirps (Beck et al, 2007). Themost commontypes of stimuli used in clinically recording the ASSR are

sinusoidally amplitude modulated tonal stimuli, fre-

quency modulated tonal stimuli, mixed modulated tonal

stimuli, and repeating sequence gated tonal stimuli. The

following is a discussion of the temporal and frequency

characteristics of these four types of stimuli.

Amplitude modulated (AM) tones are tones that

change in amplitude over a period of time, and theyare the most common type of stimuli used to evoke the

ASSR (Picton et al, 2003). Amplitude-modulated tonal

stimuli are created when a sinusoidal function is used

to modulate the primary tone. Generally, the higher fre-

quency signal is the carrier frequency (CF) tone, and the

lower frequency signal serves as the MF (Lins and

Picton, 1995). The degree of change in the amplitude

of the signal is referred to as the depth of modulationand is reported as a percentage, with a larger number

(90–100%) indicating a greater change in the amplitude

of the response in comparison to a smaller number

(30–40%). For example, if the carrier frequency is

4000 Hz, the MF is 100 Hz, and the tone is amplitude

modulated by 100%, then the amplitude of the signal will

change over time within each cycle, as seen in the tem-

poral waveform (see Fig. 4A). In the frequency domain,this AM signal has its primary energy at theCF (4000Hz)

and has two sidebands of energy, one at the CF 2 MF

(3900 Hz) and the other at the CF 1 MF (4100 Hz).

A frequencymodulated (FM) tone is a stimulus in

which only the frequency content of the stimulus

changes over the duration of the tone (see Fig. 4B). Fre-

quency modulated tonal stimuli are formed bymodulat-ing both the frequency and the phase of the CF tone.

Frequencymodulation looks at the maximum andmini-

mum frequencies present and how they relate to the CF

(John et al, 2001). For example, if the CF is 4000Hz and

it is frequencymodulated by 20%, then themaximumand

minimum frequency values will differ by 620% from the

CF, and thus the frequencies will vary from 3200 (CF –

800 Hz) to 4800 (CF 1 800 Hz) (as seen in the temporalwaveform). In the frequency domain, an FFT analysis

conducted on the FM stimulus shows that the primary

energy is at the carrier frequency (4000 Hz) and extends

to 800 Hz above and below the CF.

A third way to modulate the stimuli used in ASSR is

mixed modulation (MM) tone is a stimulus that in-

volves a combination of amplitude and frequency mod-

ulation. For example, if the CF is 4000 Hz, the MF is100 Hz, and there is 100% AM and 20% FM (see Fig.

4C), then one would expect to see changes in both the

amplitude and frequency of the tonal stimulus within

each cycle, as shown in the temporal waveform. For this

example, in the first cycle of the stimulus, the amplitude

increases from baseline to a maximum value at approx-

imately 5 msec, and it is evident that the frequency

changes from a lower frequency signal at approximately1 msec to a higher frequency signal in the range of 4

to 6 msec. In the frequency domain, there is a spread

of energy from approximately 3200 to 4800 Hz; thus,

Figure 4. Most common types of stimuli used to elicit an ASSR response as seen in the temporal and frequency domains. Figure conceptmodified from John and Purcell (2008).


150


the MM stimulus is less frequency specific than the AM

tonal stimulus.

The final stimulus commonly used to elicit the ASSR

is a repeating sequence gated (RSG) tone. RSGtones can include various types of tonal stimuli such

as linear-gated tones, cosine squared gated tones, and

Blackman-gated tones. As the name implies, these

RSG tones have a regular repeating pattern (as seen

in Fig. 4D). This pattern can also be seenmathematically

by calculating the period of theMF. In the example shown

in Figure 4D, the CF 5 4000 Hz and the MF 5 82 Hz;

therefore, the period of the modulation frequency 5 12msec ðT5 1

f 51 secMF 5 1000msec

82Hz 5 12msecÞ: In Figure 4D,

we see that the time difference between the maximum

positive peaks of the repeating stimuli is 12 msec. In

the frequency domain, the primary peak of energy is

located at the CF, with side lobes of energy extending

from approximately 3500 to 4500 Hz.

At least one commercial ASSR system (i.e., the Intelli-

gent Hearing System [IHS] SmartEP-ASSR) uses brieftonal stimuli, such as Blackman-gated tones, presented

at durations ranging from 4 to 8 msec, as their default

stimuli. Recently, Mo and Stapells (2008) investigated

the effect of stimulus duration on single frequency and

multifrequency ASSRs elicited to 500 and 2000 Hz CF

Blackman-gated tones. These tonal stimuli were pre-

sented at 75 dB SPL and had stimulus durations ranging

from 0.5 to 12 msec. These investigators reported thatfor the single frequency technique, ASSR amplitudes

increased as stimulus duration decreased for both 500

and 2000 Hz; however, the duration of the stimuli needed

to be quite brief (2 msec for 2000 Hz and 6 msec for

500 Hz). In contrast, for the multifrequency technique,

response interference tended to reduce the ASSR ampli-

tudes, and at 500 Hz there was no change in amplitude of

the ASSR as stimulus duration decreased. Based on thesefindings,Mo and Stapells (2008) concluded that brief-tone

stimuli may not be optimal for ASSR threshold estima-

tion, due to the compromise in frequency specificity that

accompanies the use of very brief tonal stimuli.

Overall, there are some advantages and disadvan-

tages for using each type of stimuli. The AM tone is

the most frequency specific stimuli of these four types

of stimuli. In contrast, theMMtone is the least frequencyspecific tone of these four types of stimuli; however, large

response amplitudes are elicited with this type of stim-

ulus (John et al, 2002, 2003). A unique aspect of the MM

stimulus is that it is affected by the phases of theAMand

FMcomponents, and this can alter the frequency spectra

of the tone (Dimitrijevic et al, 2002). When the AM and

FM components are out-of-phase by 180�, the peak of the

spectra will skew to the lower frequencies and potentiallydecrease the amplitude of the response (Dimitrijevic

et al, 2002). In contrast, when the AM and FM compo-

nents are in-phase, reaching their maximum amplitude

at the same time, the peak of theMMspectrawill skew to

the higher frequencies while increasing the amplitude of

the response (Dimitrijevic et al, 2002; John and Purcell,

2008).

Recently, John and Purcell (2008) reported that theamplitudes of the ASSRs recorded to MM tones, with

in-phase AM and FM components, are approximately

20% larger than those recorded to either AM tones or

FM tones and the responses still remain fairly frequency

specific. Therefore these investigators suggested that

the use of a MM tone to elicit the ASSR may provide

audiologists with the most easily detected responses

(John and Purcell, 2008).

STIMULATION TECHNIQUES

There are two primary stimulation techniques used

to record the ASSR, a single frequency stimulation

technique and a multifrequency stimulation technique

(Regan, 1982). The single frequency stimulation techni-

que presents one carrier frequency tone to one ear usingone MF. For example, a 2000 Hz CF tone presented at a

MF of 95 Hz is delivered to the client’s right ear.

In contrast, themultifrequency stimulation technique is

unique in its ability to test many carrier frequency tones

presented simultaneously in either one or both ears. The

typical carrier frequencies used in the multifrequency

technique are 500, 1000, 2000, and 4000 Hz. In the multi-

frequency stimulation technique, the ASSR softwareassigns a unique MF between 75 and 110 Hz to each of

the carrier frequency tones. Figure 5 displays an example

of a monaural multifrequency stimulation technique.

In this example, four CF tones (500, 1000, 2000, and

4000Hz) are delivered simultaneously to one of the sub-

ject’s ears. The compound stimulus being delivered to

the ear contains energy at each one of these carrier fre-

quencies (as shown in bottom left side of this figure). Thecorresponding modulation frequencies assigned to these

CF tones are 76Hz (500), 82Hz (1000), 95Hz (2000), and

101 Hz (4000). These unique modulation frequencies are

necessary for the processing of the stimuli to remain inde-

pendent through the auditory system and up to the brain.

The four CF tones in turn activate the four regions of the

basilarmembrane that are best tuned to these specific fre-

quencies, as shown on the right side of this figure. Thebrain’s response to these four unique MFs is seen in the

FFT results (as shown in the panel on the right side of this

figure). The strategies for analyzing the ASSR will be dis-

cussed in the next section.

With the multifrequency stimulation technique, it is

also possible to record the ASSR binaurally. In this bin-

auralmode, eight CF tones are presented simultaneously

(four per ear). Each CF tone is assigned a unique MF,which can range from approximately 75 to 110 Hz.

The possible advantage of using binaural stimulation

with the multifrequency technique is that hearing

sensitivity could be assessed at 500–4000 Hz in both ears

ASSRs/Korczak et al

151


in approximately the same amount of time it would take

to test one stimulus frequency in one ear using the single

frequency stimulation technique (Lins et al, 1996).

An important issue that needs to be considered whenemploying the multifrequency stimulation technique with

either normal hearing or hearing-impaired listeners is

the potential of interactions that occur in the cochlea

and/or brain among these stimuli at each of the carrier fre-

quencies. When tonal stimuli occur together, several types

of interactions can occur including masking effects, sup-

pression, and/or facilitation (see Picton, 2011, for a more

in-depth discussion of this issue). Despite these concerns,several investigators have shown that ASSR amplitudes

in normal hearing adults to the simultaneous presentation

of four AM tones withMFs ranging between 70 and 110Hz

to one and/or both ears at stimulus intensities#60 dB SPL

are not significantly different fromASSR amplitudes when

each AM tone is presented alone (Lins and Picton, 1995;

John et al, 1998; Herdman and Stapells, 2001; Mo and

Stapells, 2008). In addition, Herdman and Stapells(2001) reported that there were no significant differences

in ASSR thresholds for normal hearing adults when single

AM toneswere presented to one ear orwhenmultiple (four)

AM tones were presented unilaterally or bilaterally.

A few investigators have raised concern whether the

inclusion of lower frequency stimuli (e.g., 500 or 1000

Hz tones) in the multifrequency stimulation technique

would causemasking of ASSRs to higher frequency stim-uli (e.g., 2000 or 4000 Hz) for individuals with moderate

to severe SNHLs (Picton et al, 1998; Dimitrijevic et al,

2002). Specifically, Dimitrijevic and colleagues (2002)

reported that a few (n5 5) of their hearing-impaired sub-

jects had more accurate ASSR threshold estimations for

2000 and 4000 Hz using the single frequency versus the

multifrequency stimulation methods, thus suggesting

that a possible masking effect was occurring in the

MF test condition. In a more recent study, Herdmanand Stapells (2003) addressed this issue by comparing

ASSR thresholds for 2000 and 4000 Hz obtained using

the single frequency versus the multifrequency stimula-

tion techniques in ten adults with severe SNHLs. These

investigators reported there were no significant differen-

ces in the mean ASSR thresholds as a function of stim-

ulation technique (single frequency 5 63 6 9 dB nHL;

multifrequency 5 64 6 14 dB nHL) for these higherCFs. Therefore, Herdman and Stapells (2003) concluded

that there is no masking of high-frequency ASSRs by

concomitant presentation of lower frequency stimuli in

the multifrequency ASSR technique.

John et al (1998) provided several recommendations to

avoid significant interactions effects in adults when

using the multifrequency stimulation technique. These

recommendations include (1) MFs for the CF tonesshould be between 70 and 110 Hz, (2) CF tones need

to be at least one octave apart in order to simultane-

ously present up to four tonal stimuli to one ear without

significant loss in the amplitude of the ASSR, and (3)

stimulus intensities of the CF tones need to be 60 dB

SPL or less.

Recently, Hatton and Stapells (2011) addressed the

issue of possible interaction effects in the cochlea and/orbrain to the simultaneous presentation of multifrequency

stimuli at 60 dBSPL inASSRs recorded in normal hearing

infants. In this study, the response amplitudes of ASSRs

recorded to four CF tones (500–4000 Hz) in 15 normal-

hearing infants, agesz6-38 weeks, were compared across

Figure 5. Displays how the four carrier tones are presented simultaneously and thus stimulate the frequency regions of the basilarmembrane best tuned to these frequencies. The energy present at the MF can be seen in the FFT results. (Modified and adapted fromJohn and Purcell, 2008).


152


three different stimulus conditions: monaural single fre-

quency, monaural multifrequency, and binaural multifre-

quency. The stimuli were presented at 60 dB SPL for all

test conditions. All infants had passed DPOAE screeningsbilaterally on the day of testing. Hatton and Stapells

(2011) reported that the mean ASSR amplitudes for

themonaural single frequency test conditionwere signif-

icantly larger than the response amplitudes for the two

multifrequency test conditions. The infants’ mean res-

ponse amplitudes decreased as the number of simultane-

ous stimuli increased. These findings suggest that

interactions in the cochlea and/or brain do occur inresponse to the presentation of multiple stimuli at

60 dB SPL in the infants’ ears. These results differ sub-

stantially from those seen in adults, where no significant

interactions to the presentation of multifrequency stimuli

were seen at stimulus intensities#60 dB SPL (John et al,

1998; Herdman and Stapells, 2001). Hatton and Stapells

(2011) suggest that the reductions in amplitude seen in

the infants’ multifrequency test conditions are likelythe result of the immaturity of neural developmentwithin

the auditory brainstem region as well as possible imma-

turity in more peripheral structures, such as the ear

canal, middle ear, and/or cochlea.

METHODS OF ANALYZING RESPONSES

TheASSR is unlike many other AEPs in the way thatthe responses are analyzed. Traditionally, for most

AEPs, latency and amplitude measurements are taken

on the various components present in the response. This

peak-picking task requires some degree of subjective in-

terpretation on the part of the examiner. In contrast, the

analysis of the ASSR is objective and relies on statistical

methods, such as the F test, to predict the presence or

absence of a response with a certain degree of statisticalaccuracy (p, 0.05). Two primary techniques are used to

analyze the ASSR, and both methods initially require

that the temporal waveform of the ASSR be converted

into the frequency domain using FFT analysis.

One techniqueused toanalyze theASSRrelies onphase-

coherence values. “Phase coherence (PC) is related to the

signal (response)-to-noise (backgroundEEGandmyogenic)

ratio” (Cone and Dimitrijevic, 2009, p. 333). Phase coher-ence uses a measure called the phase coherence squared

(PC2) value. These PC2 values range from 0.0 to 1.0 and

are measured on a normalized scale. The closer the value

is to 1.0, the higher the coherence value is, indicating that

the amplitude of the response is significant and is distin-

guishable from the amplitude of the background noise. In

this technique, the amplitude and phase information, pro-

vided by the FFT results, is used to forma plot displayed inpolar coordinates, commonly called a polar plot (see Fig. 6).

Themagnitude or amplitude of the response corresponds to

the length of the vector, whereas the phase or time delay is

indicated by vector angle (see Fig. 6A).

If the vectors in the resultant polar plot are located pri-

marily in one quadrant (see Fig. 6B), they form a cluster of

responses. The pattern is referred to as phase locked, and

the phase coherence value is close to 1.0. This situationonly occurs when the brain is accurately responding or fir-

ing in response to the temporal information present in the

stimulus. Figure 6B shows an example of a phase-locked

response. The vectors are clustered in one quadrant indi-

cating synchronous firing to the stimulus presentation.

The PC2 value is 0.9, indicating a response is present, dis-

tinguishable from theEEGnoise, and the brain is synchro-

nously firing to the stimulus. In contrast, if the polar plotcontains vectors at random phase angles (see Fig. 6C) this

means that the pattern is not phase locked and the phase

coherence value would be closer to 0.0. Figure 6C displays

an example of a random response. In this polar plot, the

vectors are present in all four quadrants indicating dys-

synchrony in the neural firing pattern. The PC2 value in

this example is 0.1, indicating a responsewas not detected.

TheASSR is judged to be a randomresponse andnot a trueneural response to the stimulus being presented.

Another method of analyzing the ASSR uses a combina-

tionof theFFTresultsandtheF-test, to statistically evaluate

the presence or absence of a response to a certain CF tone

presentedat one stimulus intensity. TheFFTresults provide

a spectral viewof the energy occurringat themodulation fre-

quency(ies) in comparison to the energy present at the

Figure 6. (A)An example of anEEGsample shownas a vector on apolar plot where the angle of the vector indicates phase informationand the length of the vector indicates the magnitude of the response.(B) An example of a polar plot showing a response that is phase-locked with all vectors in the same quadrant and a PC2 value of0.9 indicating a high coherence value. (C) An example of an ASSRpolar plot showing a response that is random due to the spread ofthe vectors across all four quadrants with a PC2 value of 0.1 indicat-ing a low coherence value. (Modified fromGrason-Stadler Inc, 2001).

ASSRs/Korczak et al

153


surrounding frequencies. If the amplitude of the response at

the MF is significantly larger than the EEG energy at fre-

quencies above and below theMF, then a response has been

detected for theCFtonepresentedat that stimulus intensity.Figure 7 displays examples of this type of response

analysis being applied to anASSR recorded using a single

frequency stimulation technique (top panel) and using a

multifrequency stimulation technique (lower panel). The

FFT results in the top panel clearly demonstrate that

energy present at the MF (85 Hz) is considerably larger

than the amplitude of the EEG activity in the neighbor-

ing bins. The F-test is then used to objectively determinethe strength of the energy present at the MF relative to

the energy present in the surrounding bins (usually 60

bins both above and below theMF). Clinically, eachASSR

system sets an alpha level (typically p , 0.05) as the cri-

teria for determining if the energy present at that MF is

significantly greater in amplitude in comparison to the

ongoing EEG energy present in the surrounding bins.

In this example, the ASSR would be judged to be presentfor the 1000 CF tone at this stimulus intensity.

In contrast, the multifrequency analysis begins with a

MM stimulus, which consists of four CF tones (500, 1000,

2000, and 4000 Hz) being presented at four unique MFs

(77, 85, 93, and 101 Hz) to the right ear (seen in lower

panel of Fig. 7). The results of the FFT show that the

energy present at the fourMFs is significantly larger than

that of the surrounding EEG activity. Therefore, in thisexample, the ASSR would be judged to be present at

500, 1000, 2000, and 4000 Hz for this stimulus intensity.

Regardless ofwhich analysis technique is used to deter-

mine if a response is present in an ASSR recording, the

results are usually very similar (Dobie andWilson, 1996).

While the two methods above are the most commonly

used methods of analysis, there are also variations and

combinations of these methods used (Picton et al, 2003).A concern with the use of the F-test statistic for ASSR

response detection arises with repeated measures. As

sweeps are collected, the F-test is applied to each sub-

sequent ensemble average. The repeated use of the

F-test results in diminishing the statistical strength

of the test. The probability of multiple repeated mea-

sures being significant is higher than a single measure.

Several methods have been proposed to overcome this

issue. One approach is to compensate for the repeated

measure by increasing the response statistical criteriafor each subsequent measure. This may be accomplished

using a Bonferroni correction (Benjamini and Hochberg,

1995). Another approach is to monitor the response level

over multiple measures in order to assure that the res-

ponse remains at a statistically significant level over a

specific time period and therefore reduce the probability

of false response detection (Luts et al, 2008). Still another

approach is to modify the response criteria depending onother response quality measures (Sankoh et al, 1997;

John and Purcell, 2008).

TECHNICAL PARAMETERS

Recording the ASSR requires the use of specific tech-

nical parameters in order tomaximize the amplitude

of the response and to reduce the background noise. Fourtechnical factors that play an important role in recording

theASSRare the analogEEGbandpass filter settings, the

electrodemontage, the number of recording channels, and

automatic stopping rules and residual noise criteria that

can be used to determine the number of sweeps needed to

successfully meet that criteria for each test condition.

Analog EEG Band Pass Filter Setting

Theappropriate analogEEGbandpassfilter settings for

recording any AEP are determined by knowledge of the

spectral energy that is present in the response. The energy

present in the ASSR is determined by the modulation fre-quencies used to record the response. TheseMFs generally

range from 77 to 101 Hz (Small and Stapells, 2008b). In

order to successfully capture the energy present at these

modulation frequencies and to help prevent electrical arti-

fact at the rate of modulation, a commonly used and rec-

ommended analog EEG band pass filter for recording

either the single frequency and/or multifrequency ASSR

to air and/or bone conducted stimuli is 30–300 Hz (Linset al, 1996; Bohorquez and Ozdamar, 2008).

Figure 7. This figure shows a comparison of the single frequency and monaural multifrequency stimulation techniques in the temporaland frequency domains. (Modified and adapted from Lins et al, 1996.)


154


Electrode Montage

The electrode montage used in recording an ASSR is

similar to that used in recording anABR. The ground elec-trode is typically placed on the forehead at Fpz. The non-

inverting electrode is placed at the vertex or Cz location.

The inverting electrodes are placed on the earlobes of both

the test ear and the non–test ear, referred to as A1 and A2

locations. This configuration of electrodes allows for a two

channel (ipsilateral and contralateral) recording, and it

also permits the ability to switch test ears without chang-

ing thepositions of the electrodes.A single channelmidlinerecording technique is alsopossible; this electrodemontage

is useful for newborn hearing screening applications.

Number of Recording Channels

To date, only three research groups have investigated

whether ipsilateral and/or contralateral recording channels

would yield larger ASSR amplitudes, lower ASSR thresh-

olds, and higher SNRs for air-conducted or bone-conducted

stimuli in adults and infants (van der Reijden et al,

2001, 2005; Small and Stapells, 2008b; Van Maanen

and Stapells, 2009). Small and Stapells (2008b) reported

that in normal-hearing adults, there were no large differ-ences (within 1 dB) in themeanASSR thresholds recorded

in the ipsilateral versus contralateral channels for both

air- and bone-conduction testing. These investigators also

reported the largest difference in ASSR thresholds bet-

ween recording channels was 10 dB, and this occurred

in only 28% of their adult subjects.

In contrast, all three research groups reported that air-

conducted ASSRs recorded in infants using the multifre-quency stimulation technique were either significantly

smaller in amplitude or often absent in the contralateral

channel when the responses were judged to be present in

the ipsilateral channel. Specifically, Van Maanen and

Stapells (2009) recorded multifrequency air-conduction

ASSRs in 54 normal hearing infants (aged 0.7 to 66 mo)

and reported that the amplitudes of the contralateral

ASSRs to 500, 1000, and 2000 Hz stimuli were less than50% of the amplitudes of the ipsilateral responses for all

stimulus intensities. These authors also reported that the

contralateral ASSRs were only present for 31% of their

infants across all test frequencies. Based on these find-

ings, Van Maanen and Stapells (2009) concluded that

when considering infants’ ASSR to AC stimuli, absent

contralateral ASSRs cannot be used to indicate an ele-

vated AC threshold, and only the results from the ipsilat-eral recording channel should be considered.

Small and Stapells (2008a) reported a similar pattern

of findings for bone-conducted ASSRs in 14 normal

hearing infants (aged 8–44 wk). Specifically, infants

had significantly smaller mean ASSR amplitudes in

the contralateral versus ipsilateral channel, and the

mean bone-conduction ASSR thresholds, collapsed across

test frequencies, were 13 to 15 dB poorer for ASSRs

recorded in the contralateral versus ipsilateral channels.

These authors also reported that 34% of the infants did

not have significant bone conduction ASSRs (BC-ASSRs)in the contralateral channel when the responses were

judged to be present in the ipsilateral channel.

Lastly, van der Reijden et al (2005) investigated

whether certainEEGrecording channelswould yield high

SNRs in ASSRs recorded in infants aged 0 to 5 mo. These

investigators simultaneously recorded theASSRon10dif-

ferent channels, with various electrodes located on the

frontal lobe, parietal lobe, right/left mastoids/earlobes,the inion, and the nape of the neck, all referenced to Cz.

van der Reijden and colleagues (2005) reported that the

highest SNR was obtained when the inverting electrode

was ipsilateral to the stimulated ear, with the noninvert-

ing electrode located at the vertex (Cz).

In general, the use of two recording channels (ipsilateral

and contralateral) is recommended for ASSR testing, espe-

ciallywith infants. If differences inASSR threshold estima-tions between channels are seen during testing and testing

errors have been eliminated, then audiologists should rely

on the responses obtained in the ipsilateral channel.

Automatic Stopping Rules and Criteria

Stopping rules include rules or algorithms that will ter-

minate a testwhena response is detectedandwhen there isno reasonable possibility of detecting a response. Typically,

a relatively large number of sweeps is needed to obtain high

SNRs and accurate estimates of behavioral thresholds

(John andPicton, 2000; Tlumak et al, 2007). As the number

of sweeps increases, so does the testing time, which can be

clinically undesirable. In order to shorten test duration, but

still maintain an adequate SNR, automatic stopping rules

have been developed. Most ASSR response detection algo-rithms rely on the SNR of the response to determine if a

true neural response has occurred (Cone and Dimitrijevic,

2009). As discussed previously, an F-test is used to deter-

mine the statistical strength of the SNR measure. If the

SNR is found to be statistically significant and stable over

multiple sweeps, the ASSR system will determine that a

response has been detected and will automatically stop

the test. In cases where multifrequency ASSRs are beingacquired, the system may continue to test all frequencies

until all frequencies have met the required SNR criteria

or may stop testing specific frequencies as each one meets

the SNR criteria. If the SNR criterion is not met, most sys-

temswill continue to acquire data until a prespecified num-

ber of maximum sweeps is reached.

Response detection is an important component in ASSR

testing and automated data acquisition; however, stoppingrules when “no response” is present are also important.

The use of a residual noise (RN) measure has been previ-

ously used in the acquisition of auditory evoked potentials

(AEPs). The RN measure can be used both to determine

ASSRs/Korczak et al

155


the quality of a response and to stop recording when no

response is present (Ozdamar and Delgado, 1996). A com-

monly used method for evaluating RN is to use a split-

sweep buffer technique, in which the even and odd sweeppresentations are averaged into different acquisition buf-

fers. The sum of the two buffers provides the signal esti-

mate while the difference provides the noise estimate.

The RN can be calculated from the noise estimate in both

in the time and frequency domains. The RN measure pro-

vides an efficient method for automatically stopping the

acquisition of ASSR recordings when these have reached

an adequate noise level. This technique is currently usedin some commercial ASSR systems.

SUBJECT VARIABLES

ASSRs may be affected by various subject factors

including age, subject state (awake versus asleep),

and the listener’s attention to the task. The impact of

each of these factors will be briefly discussed.

Age

The effect of age on the ASSR was one of the original

limitations for the 40 Hz response. Specifically, a

robust 40 Hz response could be recorded from adults

with normal hearing sensitivity; however, this response

was absent in infants and young children. Possible ex-

planations for the discrepancy in the presence/absence

of this response between these clinical populations in-

cluded these: (1) children have immature auditory corticescompared to adults, making it potentially more difficult

for them to process fast stimulation rates, and (2) the

absent 40 Hz responses in infants was likely due to

these young children being asleep during the recordings

(Picton et al, 2003). It was later discovered that the

ASSR could be reliably recorded in infants and children

but at significantly higher modulation rates of 80 Hz or

higher (Aoyagi et al, 1993; Levi et al, 1993; Rickardset al, 1994).

Sleep

The patient’s sleep state, both natural and sedated, can

affect the amplitude and detectability of the response dueto the changes that sleep causes in the physical-electrical

activity that occurs in the brain. Several investigators

have reported that the response amplitudes of the ASSRs

were considerably smaller when patients were in natural

and/or sedated sleep compared to when they were in an

alert state (Galambos et al, 1981; Linden et al, 1985;

Plourde and Picton, 1990; Dobie andWilson, 1998; Picton

et al, 2003). For example, Plourde and Picton (1990)reported that the mean amplitude of the ASSR was

0.41 mV for their patients prior to undergoing general

anesthesia and then significantly decreased to 0.17 mV

during late induction of the anesthetic agent.

Attention

A subject’s attention to the listening task often plays an

important role in the responses obtained to various audi-tory evoked testing; however, the relationship between

attention and the ASSR is less defined. Early studies

showed that the role of attention was negligible on the

40 Hz response, but more recent studies have shown that

the general amplitude of the ASSR increases when the

subjects were attending to the stimulus (Linden et al,

1987; Ross et al, 2004). Linden et al (1987) reported no sig-

nificant difference in the amplitude of the ASSR in eightadults, regardless of whether the subject was attending to

or ignoring the stimuli. In contrast, Ross et al (2004) re-

ported that the grand-mean amplitude of ASSR increased

by 60%when subjects (n5 12)were attending to the stim-

uli compared to when they were not. Onemain reason for

the discrepancy in the results of these two studies is likely

due to the type of attention task used. Linden and col-

leagues (1987) used intensity and frequency discrimina-tion tasks in which the subjects counted the number of

intensity or frequency changes they heard in the “Attend”

condition. In contrast, Ross and colleagues (2004) used an

AM discrimination task in which the listeners were

required to discriminate changes in the rhythm of the

stimulus in the “Attend” condition. Picton et al (2003)

commented that the relationship between attention

and the ASSR is still unclear and further investigationof this topic is needed.

ACCURACY OF BEHAVIORAL

THRESHOLD PREDICTION

One of the primary roles of the ASSR is to estimatethe pure tone audiogram in difficult to test popu-

lations. Two issues that are integral in determining the

accuracy of behavioral threshold predictions are the fre-

quency specificity of the ASSR as well as the cochlear

place specificity of this response. Each of these two con-

cepts will be briefly defined below.

The frequency specificity of the ASSR is clearly depend-

ent upon the type of stimuli used to record the responseand the frequency or acoustic specificity of these stimuli.

As previously mentioned, the stimuli commonly used to

clinically record the ASSR (i.e., AM, FM, MM, and RSG

tones) all have good acoustic specificity, as their main

energy is located at the CF with small side lobes of energy

present both above and below the CF (CF1MF and CF2

MF) as shown in Figure 4. The frequency specificity of

the response is a termgenerally applied to threshold esti-mations, and it “refers to how independent a threshold at

one stimulus frequency is of contributions from surround-

ing frequencies” (Oates andStapells, 1998, p. 61). Cochlear

place specificity, in contrast, refers to the specific point

along the basilarmembrane that has beenmaximally acti-

vated by the stimulus (Herdman et al, 2002).


156


Herdman and colleagues (2002) investigated the coch-

lear place specificity of the ASSR using a noise masking

technique known as the high pass noise derived response

(HP/DR) technique. Results from this study demonstra-ted that the maximum amplitude of the derived bands

occurredwithin a half octave of the CF tone at each stim-

ulus frequency and that there were no significant differ-

ences in the place specificity of the ASSR for the single

frequency versus multifrequency technique. Thus, these

investigators concluded that “ASSRs to moderately

intense tonal stimuli (60 dB SPL) reflects activation of

a reasonably narrow cochlear region surrounding theCF tone, regardless of whether the AM tones are pre-

sented simultaneously or separately” (Herdman et al,

2002, p. 1569). They also concluded that the place specif-

icity of the ASSR is as good or slightly better as that

obtainedwith theABRand/orMLRelicited by brief tones.

One way to assess the frequency specificity of the

ASSR is to see how well the ASSR thresholds estimate

behavioral pure tone thresholds, especially in individu-als with sensorineural hearing loss (SNHL). The next

sections of this manuscript will review how well the

ASSR estimates behavioral thresholds in certain clinical

populations (i.e., normal hearing sensitivity and SNHL).

In these sections of themanuscript, the findings from the

air conduction (AC) ASSR studies will be presented first

followed by the findings from the bone conduction (BC)

ASSR studies. Within the AC and BC sections, the re-sults of adult ASSR studies will precede the results of

ASSR studies conducted on the pediatric population.

AIR CONDUCTION

Adults with Normal Hearing Sensitivity

Several studies have investigated the accuracy ofusing AC ASSR thresholds to predict behavioral pure

tone thresholds in adults with normal hearing sensitiv-

ity by calculating mean difference scores (MDSs)

(Lins et al, 1996; Cone-Wesson, Dowell, et al, 2002;

Dimitrijevic et al, 2002; Herdman and Stapells, 2003).

These MDSs are calculated by subtracting the ASSRthreshold from the behavioral pure tone threshold at

the CF of interest, typically 500–4000 Hz.

As shown in Table 1, the MDS across studies ranged

from 23.72 to 14 dB for the single frequency stimulation

technique (see row A) and from 4 to 17 dB for the multifre-

quency stimulation technique (see row B) across the four

CFs. The variability present in these studies, reflected in

the SD values, was similar for both the single frequencyand multifrequency stimulation techniques. In general,

there wereminimal differences in theseMDS as a function

of carrier frequency. Collectively, the results of these stud-

ies suggest that both the single frequency and themultifre-

quency ASSR techniques can be used to reliably estimate

AC behavioral pure tone thresholds from 500 through

4000 Hz in adults with normal hearing sensitivity.

Recently, D’Haenens et al (2008) demonstrated thatASSR thresholds have excellent test-retest reliability

for all four CF tones as evidenced by little or no changes

inMDS from trial 1 to trial 2 (as seen in Table 1, row C).

Lastly, Herdman and Stapells (2003) demonstrated

minimal differences (1–3 dB) in the MDS across CFs

for the monaural versus dichotic/binaural ASSR test

conditions (see row D). Therefore these investigators

concluded that the use of the binaural multifrequencytechnique has the potential to significantly reduce test-

ing time for individuals with normal hearing, without

sacrificing the accuracy of threshold prediction.

Adults with SNHL

Much research has been conducted on the accuracy

of the ASSR in estimating AC behavioral thresholdsin adults with SNHL. The issues addressed in the lit-

erature include the overall accuracy of threshold

Table 1. Summary of the Mean Difference Scores (and their SD values) for the Four Carrier Frequency Tones Reportedacross Studies for Adults with Normal Hearing Sensitivity

Description

Mean Difference Scores (MDS)

500 Hz 1000 Hz 2000 Hz 4000 Hz

A Single Frequency Cone-Wesson, Dowell,

et al (2002)

SF, Monotic 23.72 (15.0) 20.45 (14.7) 1.67 (13.7) 20.96 (15.0)

Herdman and Stapells

(2003)

SF, Monotic 7 (13) 10 (12) 12 (10) 14 (6)

B Multifrequency Lins et al (1996) MF, Monotic 14 (11) 12 (11) 11 (8) 13 (11)

Dimitrijevic et al (2002) MF, Monotic 17 (10) 4 (11) 4 (8) 11 (7)

Herdman and Stapells

(2003)

MF, Monotic 11 (11) 10 (11) 11 (10) 14 (10)

C Test-Retest Reliability D’Haenens et al (2008) MF, Monotic Trial 1 19 (13) 14 (10) 10 (9) 13 (10)

MF, Monotic Trial 2 19 (11) 13 (10) 12 (9) 14 (9)

D Monotic vs. Dicotic Herdman and Stapells

(2001)

MF, Monotic 11 (11) 10 (11) 11 (10) 14 (10)

MF, Dichotic/ Binaural 14 (10) 8 (7) 8 (9) 15 (9)

Note: SF 5 single frequency stimulation technique; MF 5 multifrequency stimulation technique.

ASSRs/Korczak et al

157


prediction, the influence of degree and configuration of

the SNHL on test accuracy, and the efficacy of the single

frequency versus multifrequency techniques in this

clinical population.The overall accuracy of the ASSR in estimating behav-

ioral thresholds has been studied using adolescents and

adults with varying degrees of SNHL (Lins et al, 1996;

Dimitrijevic et al, 2002). The range of MDS in these two

studieswas from5 to 13 dBacross the fourCF tones (Table

2, row A). The variability in these measures, reflected in

the SD values, was essentially similar across the four

CF tones. Dimitrijevic and colleagues (2002) reported thatthe accuracy of threshold prediction at 500Hzwas slightly

poorer than at the other CFs tested, as reflected in the

higher MDS at this test frequency.

Several investigators have looked at the influence that

degree and/or configuration of the SNHL has on the ability

of the ASSR to accurately predict behavioral thresholds in

adults (Rance et al, 1995;HerdmanandStapells, 2003). Ini-

tially, Rance et al (1995) indicated that the ASSR is moreaccuratewhen thedegree of theSNHL is greater (i.e., 60 dB

HLor greater) as shownby the smallerMDS in this sample

(see Table 2, rowB).More recently, however,Herdman and

Stapells (2003) reported that the multifrequency ASSR

method provides a good estimate of both degree and config-

uration of SNHL in adults with losses ranging frommild to

profound. Specifically, these investigators reported signifi-

cant correlations (r5 0.75 to 0.89) exist between pure tonebehavioral thresholds and ASSR thresholds for all four CF

tones (i.e., 500–4000 Hz). They reported that the ASSR

thresholds obtained using the multifrequency technique

were within 20 dB of their behavioral thresholds in 81,

93, 93, and 100% for the hearing-impaired subjects

at CFs of 500, 1000, 2000, and 4000 Hz, respectively.

HerdmanandStapells (2003) also demonstrated that the

configuration of the SNHL (steeply sloping versus flat/shallow) had little or no effect on the accuracy of the

ASSR threshold prediction as shown by the similar

MDS for these two clinical groups (see Table 2, row C).

Studies have compared the accuracy of ASSR threshold

estimations using the single frequency versus multifre-

quency techniques in individuals with SNHL (e.g., Luts

and Wouters, 2005). These investigators reported therewere no significant differences between ASSR threshold

predictionsobtainedusing the single frequencyversusmul-

tifrequency stimulation techniqueas shownbysimilarMDS

for these two stimulation techniques (see Table 2, row D).

Collectively, the results of these studies suggest that

the accuracy of AC threshold prediction in adults is not

influenced by either the degree of the SNHL or by the

configuration of the hearing loss. Both single frequencyand multifrequency ASSR techniques are accurate pre-

dictors of AC behavioral thresholds (within z8–13 dB)

in adults with SNHL.

INFANTS AND YOUNG CHILDREN WITH

NORMAL HEARING SENSITIVITY

I n the pediatric population, the early detection andtreatment of hearing loss are critical steps in prevent-

ing the delay of speech and language development in an

infant and/or young child (Yoshinaga-Itano et al, 1998;

Moeller, 2000). To achieve these goals, many countries

worldwide have mandated newborn hearing screening

programs, which rely on objective tests, such as click-

evoked ABRs and/or OAEs, as the screening instruments

(Joint Committee on Infant Hearing [JCIH], 2007). Forinfants ,6 mo of age, electrophysiological evaluation pro-

cedures have been recommended to quantify the degree

and configuration of the hearing loss. Currently, recording

the ABR to AC and/or BC tonal stimuli is the clinical gold

standard for estimating behavioral thresholds in this neo-

nate population (JCIH, 2007). More recently, the ASSR

has emerged as a possible alternative electrophysiological

technique to theABR for estimating behavioral thresholdsin this clinical population. The next section of this manu-

script will discuss how factors, such as the type of test

environment, the length of the recording time per stimulus

Table 2. Summary of the Mean Difference Scores (and their SD values) between Behavioral and ASSR ThresholdsReported across Studies for Individuals with SNHL

Description

Mean Difference Scores (MDS)

500 Hz 1000 Hz 2000 Hz 4000 Hz

A Accuracy Lins et al (1996) MF, Moderate SNHL 9 (9) 13 (12) 11 (10) 12 (13)

Dimitrijevic et al (2002) MF, Mild-severe SNHL 13 (11) 5 (8) 5 (9) 8 (11)

B Degree Rance et al (1995) SF, Degree of loss: 0–55 dB HL 9.6 8.6 6.3 4.7

SF, Degree of loss: 601 dB HL 7.9 5.6 3.8 5.0

C Configuration Herdman and Stapells

(2003)

MF, Group A:Steeply sloping

($30 dB/octave)

13 (13) 8 (10) 12 (10) 1 (10)

MF, Group B:Flat/shallow

(#30 dB/octave)

15 (13) 7 (8) 7 (11) 5 (9)

D SF vs. MF Luts and Wouters (2005) SF, AUDERA 20 (8) 14 (7) 13 (7) 14 (13)

MF, MASTER 17 (12) 12 (8) 17 (8) 19 (12)

Note: SF 5 single frequency stimulation technique; MF 5 multifrequency stimulation technique.


158


intensity, the SNR of the response, and the gestational age

of the infant can influence ASSR thresholds. Lastly, there

is a brief review of the effects of SNHL on ASSRs recordedin infants and young children.

The data in Table 3 clearly demonstrate that the type of

test environment and the level of the ambient noise

present in this environment can have a negative effect

on ASSR thresholds in infants. In a study by Lins and col-

leagues (1996) the ASSR infant thresholds were reported

fromtwodifferent sites (OttawaandHavana). TheOttawa

data were recorded in a sound-attenuated room, and theambient noise levels measured ranged from 26 to 47 dB

SPL, while the data collected in Havana was recorded

in a room with no sound attenuation and the ambient

noise levels ranged from 36 to 53 dB SPL (Lins et al,

1996). As can be seen in Table 3, the ASSR thresholds

fromOttawa are significantly lower (better) for all four

CF tones in comparison to the ASSR thresholds recorded

in Havana. These researchers hypothesized that thisdifference in ASSR thresholds between these two sites

is most likely due to the difference in ambient back-

ground noise present in each of these test environments

(Lins et al, 1996). The impact of high ambient noise lev-

els is also evident in the Savio et al (2001) data. In this

study the ambient noise levels ranged from 62 to 65 dB

SPL. The ASSR thresholds in the Savio et al study were

approximately 5 to 20 dB higher (poorer) than the ASSRthresholds recorded in several infant ASSR studies,

which utilized sound-attenuated chambers for their

recordings (e.g., Levi et al, 1995; Lins et al, 1996 [Ottawa

data]; Swanepoel and Steyn, 2005; Stroebel et al, 2007;

Van Maanen and Stapells, 2009). Savio and colleagues

(2001) hypothesized that their ASSR thresholds may

have been elevated due to the level and spectral compo-

sition of the ambient noise present during their testing.

A second factor that may influence the ASSR thresh-

olds in infants is the length of the recording time. In sev-

eral of the single frequency studies (e.g., Levi et al,1995; Rance et al, 2006), the recording time per inten-

sity is fairly brief, ranging from z20 to 100 sec. In con-

trast, the mean recording times per intensity for the

multifrequency studies were considerably longer (e.g.,

3–13 min for the Ottawa data in Lins et al [1996]

and 6.3 6 3.1 min for the Van Maanen and Stapells

[2009] data). Picton and colleagues (2005) have stated

that when the duration of the recording is longer, theresidual noise in the EEG is less, making small ampli-

tude responses close to threshold easier to recognize.

A related concept that influences the detectability of

small amplitude responses in infants at near threshold lev-

els is the SNR of the response. SNR is calculated by divid-

ing the response amplitude of the ASSR by the amplitude

of the background EEG noise. Both of these amplitude

units are typically expressed in nanovolts (nV). Severalinvestigators have demonstrated that infants have con-

siderably smaller ASSR response amplitudes in compar-

ison to adults (e.g., John et al, 2004; Luts et al, 2006).

Specifically, John et al (2004) reported that the mean

ASSR amplitude to a 50 dB SPL exponentially modulated

AM tone (AM2) tone isz35 nV in normal hearing adults,

while the mean amplitude to the same tone in a newborn

is only 17 nV. Therefore, one way to enhance the SNR ofthe ASSR in infants is to employ a stringent noise crite-

rion (the lower the nV value, the more stringent the cri-

teria). For example, Luts et al (2006) reported that the

mean ASSR amplitude for a 2000 Hz CF tone presented

at 30 dB SPL in infants was 6 nV. If a strict 5 nV noise

criterion is employed, then the SNR5 1.20 (6/5), which is

double the SNR value that would occur if a less strict noise

criterion, such as 10 nV, were employed (SNR5 0.6 [6/10]).

Table 3. Summary of the Mean Threshold Levels in dB HL (and their SD values) for the Four Carrier Frequency TonesReported across Studies for Infants with Normal Hearing Sensitivity

Age Participants

Test Environment/Avg.

Ambient Noise Level

Recording

Time

dB HL

500 Hz 1000 Hz 2000 Hz 4000 Hz

A Single

Frequency

Rickards et al, 1994 1–7 day 245 Quiet Room/z30 dBA 0.5–3.5 min 41.36 34.51

Levi et al, 1995 1 mo 55 SA 100 sec 36* 42* 34*

Rance and Tomlin,

2006

0 wk 20 NSA/z46.1 dB SPL 22.4–89.6/sec 46.0 (11.4) 38.5 (11.0)

6 wk 20 NSA/z45.6–46.1

dB SPL

22.4–89.6/sec 39.7 (7.3) 32.2 (7.9)

B Multifrequency Lins et al, 1996 1–10 mo (Ottawa) 21 SA/26–47 dB SPL 3–13 min 33† 22† 17† 21†

3–11 mo (Havana) 30 NSA/36–53 dB SPL 3–13 min 46† 36† 30† 32†

Savio et al, 2001 0–1 mo 25 NSA/62–65 dBA 3–8 min* 57* 55* 51* 48*

7–12 mo 13 NSA/62–65 dBA 46* 44* 37* 33*

Swanepoel and Steyn,

2005

3–8 wk 5 SA 37 (8) 34 (10) 34 (11) 30 (11)

Van Maanen and

Stapells, 2009

Younger (#6 mo) 10 SA 7.03 min (3.53) 39.0 (7.4) 33.1 (4.6) 29.3 (7.3) 24.3 (10.1)

Older (.6 mo) 19 SA 5.80 min (2.71) 41.3 (7.4) 36.9 (10.5) 31.4 (7.8) 22.2 (10.3)

All 29 SA 6.3 min. (3.10) 40.5 (7.4) 35.8 (9.1) 30.8 (7.6) 22.8 (10.1)

Note: SA 5 sound attenuated chamber; NSA 5 no sound attenuation.

*Data taken from table 1 in John et al (2004).†All data from Lins et al (1996) study was reported in dB SPL values. Ottawa data subsequently converted into dB HL values and reported in

John et al (2004). Current authors converted Havana data into dB HL values using the same conversion factors.

ASSRs/Korczak et al

159


In theVanMaanen and Stapells (2009) study, these inves-

tigators continued recording at each stimulus intensity

until the mean noise level in the side bins surrounding

the modulation frequencies was#5 nV. They demonstra-ted that the detectability of the ASSR in infants was sub-

stantially enhanced for all four CF tones when the noise

criterion was reduced from 10 to 5 nV due to improve-

ments in the SNRs (Van Maanen and Stapells, 2009).

A fourth factor that influences ASSR thresholds in

infants is the gestational age (GA) of the infant. Two

primary approaches to studying the effects of age on

infant ASSRs have been (1) to compare ASSR thresh-olds in infants versus adults and (2) to compare ASSR

thresholds in newborns versus older infants. Lins et al

(1996) measured ASSR thresholds to 500–4000 Hz CF

tones in a group of normal hearing infants (1–10 mo)

and in a group of normal hearing adults and reported

that ASSR thresholds were z10–15 dB higher (poorer)

in the infants across all frequencies. Similarly, Rance

andRickards (2002) comparedASSR thresholds of adultsversus older infants (mean age of 3mo) and reported that

older children and adults had approximately 10 dB lower

(better) thresholds in comparison to infants. Lastly, Van

Maanen and Stapells (2009) reported that the ASSR

thresholds for their normal-hearing infants were consid-

erably higher (poorer) at 500–2000 Hz in comparison to

their ASSR data reported for normal hearing adults in

Herdman and Stapells (2001).Several investigators were interested in comparing

ASSRs in younger babies versus older babies (Savio

et al, 2001; Cone-Wesson, Parker, et al, 2002; John et al,

2004; Luts et al, 2006; Rance and Tomlin, 2006; Ribeiro

et al, 2010). In general, the collective results of these stud-

ies demonstrate that ASSR thresholds improve as the

infant matures. For example, Savio et al (2001) compared

results for agroupofneonates (0–1mo)andagroupof olderbabies (7–12mo) and found that the older babies hadASSR

thresholds that were z10–15 dB lower (better) than their

younger counterparts (see Table 3). Similarly, John et al

(2004) reported that their group of older babies (tested

between 3 and 15wk following birth) hadASSR thresholds

z10 dB lower (better) than their group of younger babies

(GA 5 37–42 wk, tested within 2–3 days of birth).

Based on the age-related effects seen in the ASSRthresholds, Rance and Tomlin (2006) conducted a longitu-

dinal study on 20 full-term infants (post conceptual age

[PCA] 5 39–41 wk) to systematically study the matura-

tional changes that occur in ASSR thresholds in normal

developing infants. Normal hearing status was suggested

by ABR and OAE results. Single frequency ASSR data

were recorded to 500 and 4000 Hz CF tones and was col-

lected at 4 discrete points in time (i.e., week 0 (3–6 daysfollowing birth) andweeks 2, 4 and 6). Their results dem-

onstrated that ASSR thresholds improved byz5–6 dB at

both test frequencies during the first 6 weeks of life (as

seen in Table 3). These investigators hypothesized that

the ASSR threshold changes seen in the neonatal/early

infancy period are the result of neural development in

the auditory brainstem. Rance and Tomlin (2006) also

reported that the variability in these threshold meas-ures, reflected in the SD values, tended to decrease as

the infants matured, and thus concluded that clinical

ASSR assessment may be better left until normal full-

term infants are at least 2 weeks of age.

EFFECTS OF PREMATURITY ON THE ASSR

Given the relatively high incidence of prematurebirths, a couple of important questions arise regard-

ing the potential clinical application of ASSR testing in

premature infants. These questions are as follows: (1)

Can ASSR thresholds be reliably recorded in premature

infants? and (2) Is there an optimal age at which ASSR

testing should be used with these infants? Several re-

searchers have addressed these concerns by comparing

response properties of ASSRs recorded in full-term neo-

nates versus those recorded in premature neonates

(Cone-Wesson, Parker, et al, 2002; John et al, 2004; Lutset al, 2006; Ribeiro et al, 2010). John et al (2004) compared

ASSRs recorded in 23 premature infants (GAs 5 37–42

weeks) to ASSRs recorded in 20 full-term infants (tested

between 3 and 15 wk following birth). ASSRs in both age

groups were recorded to AM, MM, and AM2 stimuli using

a monaural multifrequency stimulation technique. John

and colleagues (2004) reported that full-term infants

had significantly larger response amplitudes at 1000,

2000, and 4000 Hz in comparison to premature infants.

This finding was true regardless of stimulus type. Johnet al (2004) also reported that infants in both age groups

had significantly larger response amplitudes for the MM

and AM2 stimuli versus the AM stimuli and thus recom-

mended that these types of stimuli should be the stimuli of

choice in this clinical population.

Luts et al (2006) was interested in determining if there

was difference in the SNR of the binaural multifrequency

ASSR in premature infants (n 5 14 ears with PCA ,41

wk) versus full-term infants (n 5 16 ears with PCA $41wk). These investigators reported that the SNRof theASSR

increases significantlywith age. For example, at 50 dBSPL,

the SNR of the full-term infants was 41% larger/better than

the SNR of the premature infants. Based on this finding,

Lutsandcolleaguesconcludedthat theoptimalage forASSR

testing (corrected for prematurity) is between 1wkand3mo

due to the better SNRs obtained at this later age.

Ribeiro et al (2010) also compared ASSR thresholds

obtained in 27 full-term infants (GA .38 wk) versus

21 premature infants (GA ,37 wk). These investigatorsreported that the full-term infants had larger response

amplitudes in comparison to the premature infants, sim-

ilar to the findings of John et al (2004). However, in the

current study, these age-related threshold differences

only reached statistical significance at 500 and 4000.


160


Ribiero et al also reported that there were no significant

differences in the ASSR amplitudes, background EEG

noise levels, and SNR for these two age groups. Based

on these findings, Ribeiro and colleagues (2010) con-

cluded that a significantmaturational effect occurs during

gestation at the level of structures involved in the forma-

tion of the ASSR at 500 and 4000 Hz. They caution that

further studies in this area areneededand “should include

themeasurement of in situ thresholds during the preterm

period to ensure that the maturational component of the

threshold can be properly quantified and controlled to

obtain an accurate diagnosis and appropriate therapy”

(Ribeiro et al, 2010, p. 109.).

EFFECTS OF SNHL ON INFANT ASSRs

Both the single frequency and the multifrequency

stimulation techniques have been used to study

the relationship between ASSR thresholds and behav-

ioral thresholds in infants and young children with

SNHLs. A series of studies from the University of Mel-bourne, in Australia, have used the monaural single fre-

quency stimulation technique to record the ASSR and

have relied on phase coherence analysis techniques to

determine the presence/absence of a response at a partic-

ular CF and stimulus intensity (Rance et al, 1995; Rance

and Briggs, 2002; Cone-Wesson, Dowell, et al, 2002). In

general, these studies have reported that a strong positive

relationship exists between ASSR thresholds and behav-ioral hearing thresholds (BHTs) in hearing-impaired

infants and young children. For example, Rance et al

(1995) recorded ASSRs in 60 subjects, 25 children (mean

age 5 29 mo) with moderate to severe SNHLs and 35

adults (mean age5 55.7 yr) with BHTs ranging from nor-

mal to profound and reported that the correlations

between ASSR thresholds and BHTs were $0.96 at

250–4000Hz. Rance and Briggs (2002) extended these ini-tial findings to include 184 infants (aged 1–8 mo), with

moderate to profound SNHLs and also reported that

strong positive correlations (r 5 0.8120.93) exist across

frequencies between ASSR thresholds and BHTs in this

clinical population. Lastly, Cone-Wesson, Dowell, et al

(2002) investigated the relationship between ASSR and

BHTs in 51 cases with near normal/mild to severe/

profound hearing losses. Similarly, these investigatorsreported high correlation values (r5 0.7720.88) for these

two measures at 500–4000 Hz. Collectively, these three

research groups concluded that the single frequency

ASSR technique can be used to make frequency specific

estimates of behavioral audiometric thresholds in young

children with SNHLs, and are accurate enough to form a

basis for hearing aid fittings and early intervention.

In contrast, several research groups have focused theirinvestigations on the potential use of the multifrequency

ASSR stimulation technique to estimate frequency spe-

cific behavioral thresholds in hearing impaired infants

(Luts et al, 2004; Han et al, 2006; Rodrigues and Lewis,

2010; Van Maanen and Stapells, 2010). Luts et al (2004)

recorded binaural multifrequency ASSRs in 10 infants

(mean age5 7mo)whowere suspected of having a hearingloss. They compared ASSR thresholds to both click-evoked

ABR thresholds and to BHTs. Luts et al reported that a

significant positive correlation (r 5 0.77) exists between

click-evoked ABR thresholds and ASSR thresholds at

2kHz.Similarly, theyalso reported that significantpositive

correlations (r 5 0.9120.93) exist between ASSR thresh-

olds and BHTs at all four CFs. In 2006, Han et al also com-

pared the relationship between binaural multifrequencyASSR thresholds and BHTs in 40 young children, aged

6 mo to 5 yr, with varying degrees of SNHL. These investi-

gators reported these two measures were highly correlated

(r 5 0.7920.89) at 500 through 4000 Hz (Han et al 2006).

Recently two studies have looked at the relationship be-

tween ASSR thresholds and tone-evoked ABR thresholds

in this pediatric hearing impaired population (Rodrigues

and Lewis, 2010; Van Maanen and Stapells, 2010). Theseinvestigators were interested in using tone-evoked ABR

thresholds as opposed to click-evoked ABR thresholds,

as the former are currently the “clinical gold standard”

for predicting the pure tone audiogram in the pediatric

population. Rodrigues and Lewis (2010) recorded ASSR

in 17 infants, aged 2–36 mo, using the binaural multifre-

quency stimulation technique. In this study, the ASSR

and ABR testing were not conducted on the same day.Rodrigues and Lewis (2010) reported that the correlation

values between these two measures were all $0.76, indi-

cating that a strong relationship exists between tone-

evoked ABR thresholds and ASSR thresholds at 500–

4000 Hz. These investigators also compared the infants’

BHTs to both their tone-evoked ABR thresholds and their

ASSR thresholds. Rodrigues and Lewis (2010) reported

that strong positive correlations exist between bothtone-evoked ABR thresholds and BHTs (r 5 0.8120.94)

and between ASSR thresholds and BHTs (r 5 0.9420.97)

at 500–4000 Hz. In 2010, VanMaanen and Stapells also

compared binaural multifrequency ASSR thresholds to

tone-evoked ABRs in 141 ears with SNHL, 29 ears with

conductive hearing loss, and four ears with loss of

unknown origin as well as in a group (n 5 34) of young

children with normal hearing sensitivity (Van MaanenandStapells, 2010). A unique aspect of this studywas that

the tone-evokedABR thresholds and theASSR thresholds

were measured in the same test session for .93% of the

participants, to ensure that the infants’ hearing status

had not changed between test sessions. Van Maanen

and Stapells (2010) reported that the multifrequency

ASSR thresholds were strongly correlated (r5 0.97) with

the tone-evokedABR thresholds for all four CFs, similar tothat reported by Rodrigues and Lewis (2010). Van Maanen

and Stapells also reported that the multifrequency ASSR

thresholds (indBnHL)werez6–11dBhigher than the tone

ABR thresholds (in dB nHL).

ASSRs/Korczak et al

161


Collectively, the results of these four studies suggest that

the binaural multifrequency ASSR technique can be used

to reliably estimate frequency-specific hearing thresholds

in young infants with SNHL. However, both Rodriguesand Lewis (2010) and Van Maanen and Stapells (2010)

caution audiologists that although the ASSR looks like a

promising clinical tool for the assessment of hearing status

in infants and young children, the data on the use of the

binauralmultifrequency ASSR techniquewith infants and

young children who have SNHLs is still very limited, and

further studies with a larger number of infants and more

homogenous samples are needed.

SCREENING FOR HEARING LOSS IN THE

PEDIATRIC POPULATION

I n two recent studies, VanMaanen and Stapells (2009,

2010) have addressed the question of whether themul-

tifrequency ASSR technique can be used to distinguish

“normal” versus “elevated” thresholds in the pediatricpopulation. In their 2009 study, these investigators devel-

oped recommendedACscreening levels based on their own

data in conjunction with data from other infant ASSR

studies. Van Maanen and Stapells (2009) recommended

the use of 50 dB HL at 500 Hz, 45 dB HL at 1000 Hz,

and 40 dB HL at 2000 and 4000 Hz as “normal” screening

levels. In their preliminary data on a small (n5 23) group

of children with diverse hearing losses, the recommendedASSR screening levels did not pass any infantwith a hear-

ing loss. In 91.2% of the tests, the ASSR and tone ABR

results classified the infant’s hearing status in the same

category (“normal” versus “elevated”). There was only

1/80 tests (i.e., 1.3%) where the ASSR results were catego-

rized as “normal” when the tone ABR results were classi-

fied as “elevated.” In 2010, Van Maanen and Stapells

investigated whether the normal screening ASSR lev-els they had proposed in 2009 could now be accurately

applied to hearing impaired infants. These investiga-

tors reported that in 94% of the comparisons, the multi-

frequency ASSR thresholds categorized hearing in a

similar fashion to the tone ABR thresholds for this hear-

ing-impaired population (Van Maanen and Stapells,

2010). Although these data are preliminary, it lends

support to the use of the multifrequency ASSR toquickly confirm whether the infant has normal or ele-

vated AC thresholds for 500–4000 Hz in both ears.

BONE CONDUCTION

Accurate estimation of pure tone bone conduction

thresholds is important to distinguish the type of

hearing loss that may exist, and this issue is particularlyrelevant for infants and young children with unilateral

or bilateral otitis media or malformations of the outer or

middle ear (Small and Stapells, 2006, 2008a; Brooke

et al, 2009). One of the primary sources of information

on BC-ASSRs in the literature is a series of studies con-

ducted by Small and Stapells (Small and Stapells, 2005,

2006, 2008a, b; Small et al, 2007). Several clinically rel-

evant questions were addressed in these studies in-cluding (1) Do BC-ASSR thresholds accurately predict

behavioral thresholds in normal hearing adults? (2) Does

the accuracy of the ASSR threshold prediction vary as a

function of age? (3) Does the coupling method and place-

ment of a bone oscillator affect accuracy of threshold

prediction? and (4) Does the recording channel (ipsi-

lateral versus contralateral) employed affect the accu-

racy of ASSR threshold prediction? These issues will bebriefly addressed in this section of the manuscript.

Small and Stapells (2006) investigated the accuracy of

multifrequency BC-ASSR thresholds in predicting behavio-

ral thresholds in normal hearing adults and reported that

theirmean thresholds ranged from 16 to 25 dB nHL at CFs

of 500–4000 Hz (as shown in lower portion of Table 4, row

A). The BC-ASSR thresholds reported in this study were

approximately 6 dB lower at the higher versus lower fre-quencies. The variability in these measures, reflected

in the SD values, was similar across all four CFs.

Small and Stapells (2006) demonstrated that age

related differences occurred when BC-ASSR thresholds

in infants with normal hearing were compared to adults

with normal hearing, such that infants’ mean BC-ASSR

thresholds were considerably better (lower) at the lower

frequencies (500 and 1000Hz) in comparison to the adults’thresholds (see Table 4, row A). Based on these findings,

Small and Stapells (2008a) were interested in determin-

ing the time course of maturation of BC hearing sensitiv-

ity in these infants. In this subsequent study, BC-ASSRs

were recorded in three clinical groups: young infants

(0–11 mo), older infants (12–24 mo), and adults (19–48

yr). These investigators reported that low frequency

BC-ASSR thresholds increase with age/maturation, whilehigh frequency ASSR thresholds are essentially unaf-

fected by age. Specifically, the young infants had mean

ASSR thresholds that were approximately 15 to 20 dB

lower at 500 and 1000 Hz in comparison to the adults’

thresholds (Table 4, row B). These threshold differences

persist until at least 2 yr of age. In light of these findings,

Small and Stapells stressed the importance of establish-

ing normal BC hearing levels for a range of test frequen-cies when conducting ASSRs on infants of different ages.

Small et al (2007) looked at the role of both the coupling

method (elastic head band versus handheld) and the lo-

cation of the bone oscillator on the skull (temporal bone

versus mastoid versus forehead) on the accuracy of BC

behavioral threshold predictions via BC-ASSR. These

investigators reported that mean ASSR thresholds ob-

tained using an elastic head band and the handheld cou-pling methods were similar at 500–4000 Hz in normal

hearing adults (as shown in lower portion of Table 4,

row C). Overall, the mean elastic band minus handheld

threshold difference, collapsed across frequency, was only


162


21.4 dB. A similar pattern of findings was seen for the

infants. There were no significant differences in themeanASSR thresholds obtained at 500–4000 Hz using these

two coupling methods for the infants (as shown in the

upper portion of Table 4, row C). There was a trend for

the BC-ASSR threshold at 4000 Hz to be slightly higher

(i.e., 9 dB) for the handheld versus elastic band coupling

method; however, this difference did not reach statistical

significance. Small and colleagues (2007) also reported

the mean BC-ASSR thresholds recorded from the tempo-ral bone and the mastoid placement in infants were sim-

ilar at 500–4000 Hz (Table 4, row D), and these two

locations yielded considerably better (lower) thresholds

than did the forehead placement.

Lastly, Small and Stapells (2008b) investigated possible

asymmetries betweenmeanBC-ASSR thresholds obtained

from ipsilateral versus contralateral recording channels in

infants and adults with normal hearing. Their resultsrevealed similarmean thresholds obtained from ipsilateral

and contralateral channel recordings in adults, as shown in

the lower portion of Table 4, rowE. The samepattern, how-

ever, was not true for infants, who had substantially better

(lower) BC-ASSR thresholds for the ipsilateral versus con-

tralateral recordings.Thisfindingwas true for allCF tones.

One potential limitation that can occur when recording

BC multifrequency ASSRs to sinusoidally amplitudemodulated (SAM) stimuli presented at intensities $40

dBHL are electromagnetic artifacts, which can contaminate

the ASSR. Recently, Picton and John (2004) and Small and

Stapells (2004) reported that BC stimuli presented at mod-

erate to high stimulus intensities elicit electromagnetic arti-

facts, which can produce spurious ASSRs, especially for 500

and 1000 Hz CF tones. These investigators demonstratedthat large amplitude stimulus artifacts result in energy that

is aliased to exactly the ASSRmodulation rate and can mis-

takenly be interpreted as a “present” response.

Both Picton and John (2004) as well as Small and

Stapells (2004) reported that there are several ways to suc-

cessfully prevent these spurious ASSRs. One is to use anti-

aliasingfilterswith steepfilter slopes and/orprocedures that

deflect theartifact to regionsof the spectrumthataredistantfrom the frequency of the steady state response. A second

way is to ensure that the analog to digital (AD) conversion

rate is not an integer submultiple of the CF tones. Thirdly,

audiologists/hearing scientists can use stimuli with fre-

quency spectra that do not alias back to the response fre-

quencies, such as alternating SAM tones or beats. Lastly,

alternating the polarity of the stimulus can help to reduce

artifactual responses arising from stimulus artifact. PictonandJohn (2004) andSmall andStapells (2004) reported that

the artifactual BC responses that were seen for the low CF

tones (i.e., 500Hz)may have been influenced by physiologic,

nonauditory (likely vestibular responses) activity that was

not related to the stimulus artifact.

SUMMARY OF THE ACCURACY OF AIR AND

BONE CONDUCTION ASSR TESTING INESTIMATING BEHAVIORAL PURE

TONE THRESHOLDS

Collectively the results of the ASSR threshold stud-

ies, discussed above, reveal several patterns of

Table 4. Summary of the Mean ASSR BC Thresholds (and their SD values) Reported across Studies forInfants and Adults

Subject Age

DescriptionAverage Thresholds

CF (Hz) 500 1000 2000 4000

A Adults vs. Infants Small and Stapells

(2006)

Infants (0.5–27 wk) Postterm 13.6 (13.4) 2.1 (7.0) 26.4 (6.3) 22.1 (8.0)

Adults (20–48 yr) Adults 21.0 (9.9) 25.0 (12.7) 18.0 (7.9) 16.0 (10.8)

B Maturation Effects Small and Stapells

(2008a)

Young infants

(0–11 mo)

14 (12.9) 4.9 (7.8) 25.7 (10.1) 13.7 (10.6)

Older infants

(12–24 mo)

22.3 (10.9) 13.1 (6.3) 26.2 (8.7) 13.1 (9.5)

Adults 30.6 (15.1) 23.9 (13.8) 20.0 (7.7) 16.1 (10.9)

C Coupling Method Small et al (2007) Infants (0.5–38 wk) Elastic head

band

14.0 (14.3) 6.0 (8.43) 6.0 (8.43) 13.0 (11.6)

Handheld 14.0 (14.3) 1.0 (7.4) 25.0 (7.1) 22.0 (7.9)

Adults (20–39 yr) Elastic head

band

1.7 (3.7) 20.3 (6.6) 11.5 (7.5) 21.7 (5.5)

Handheld 1.5 (4.9) 2.3 (4.1) 8.7 (5.4) 0.1 (5.0)

D BC Oscillator

Placement

Small et al (2007) Infants (32–43 wk

PCA)

Temporal bone 16.0 (11.8) 16.7 (9.0) 34.6 (15.1) 33.3 (15.0)

Mastoid 17.3 (13.3) 14.0 (9.1) 32.3 (19.6) 26.0 (12.9)

Forehead 30.7 (16.2) 26.7 (13.5) 51.1 (13.6) 44.0 (9.7)

E Recording

Channels

Small and Stapells

(2008b)

Infants (2–11 mo) Ipsilateral 12.5 (12.9) 5.0 (5.2) 20.0 (12.8) 9.2 (7.9)

Contralateral 24.6 (12.1) 15.0 (15.7) 33.6 (12.1) 22.5 (12.2)

Adults (18–40 yr) Ipsilateral 31.3 (6.4) 17.5 (12.8) 20.0 (7.6) 10.0 (10.7)

Contralateral 31.3 (6.4) 18.8 (11.3) 18.8 (8.4) 11.3 (8.4)

ASSRs/Korczak et al

163


interest for estimating air and bone conduction pure

tone thresholds in individuals with normal hearing sen-

sitivity as well as those with SNHL. These patterns are

summarized below:

Air Conduction

� The overall accuracy of ASSR thresholds in estimat-ing behavioral thresholds, as reflected in the MDS,

was withinz0–17 dB for adults with normal hearing

sensitivity and within z5–13 dB for adults with

SNHL ranging from mild to severe.

� Degree and configuration of the SNHL does not

impact the accuracy of behavioral threshold predic-

tion for adults.

� Accuracy of threshold prediction is similar acrossCFs.� ASSR thresholds have good test-retest reliability in

normal hearing adults.

� Infants have considerably smaller ASSR response

amplitudes in comparison to adults and thus have

ASSR thresholds that are approximately 10–15 dB

higher (poorer) than adults.

� There are considerable differences in the response

amplitudes, ASSR thresholds, and SNR of theresponses for premature infants versus older infants

due to the maturational changes that occur in the

auditory system in the first few months of life.

� If audiologists are using the results of the multifre-

quency ASSR to distinguish between normal versus

elevated thresholds in infants and young children, it

is recommended that they use 50 dB HL at 500 Hz,

45 dB HL at 1000 Hz, and 40 dB HL at 2000 and4000 Hz as the normal screening levels (Van Maanen

and Stapells, 2009, 2010).

Bone Conduction

� BC-ASSR thresholds are fairly good predictors of

behavioral BC thresholds at 1000 through 4000 Hzin normal hearing adults.

� Low frequency BC-ASSR thresholds increase with

age, and thus different normal BC hearing levels

must be established for each frequency when con-

ducting infant ASSRs.

� There is no significant effect of BC coupling method

(handheld versus elastic head band) onASSR thresh-

olds in adults or infants.� Placement of the BC oscillator on either the temporal

bone or mastoid yields the best (lowest) BC-ASSR

threshold in infants.

� The ipsilateral recording channel yielded signifi-

cantly better (lower) BC-ASSR thresholds in infants.

� In conclusion of this threshold estimation section, it is

crucial for audiologists to realize that “it is premature

to rely on theASSR as the primary electrophysiological

measure of thresholds in infants, except possibly as the

first step in a diagnostic protocol to quickly differenti-

ate normal versus elevated thresholds” (Van Maanenand Stapells, 2010, p. 544). These investigators recom-

mend that audiologists switch to toneABRwhenASSR

thresholds are elevated to confirm the degree and type

of hearing loss. This is necessary due to the lack of suf-

ficient air- and bone-conduction ASSR data in infants

with various types and degrees of hearing loss. Recent

data by bothRodrigues andLewis (2010) andVanMaa-

nen and Stapells (2010) suggest that ASSR thresholdswould provide an excellent cross-check of the AC tone-

evoked ABR thresholds.

EFFECTS OF TEST DURATION ON THE

ACCURACY OF ASSR THRESHOLDS AND THE

RELATIVE EFFICIENCY OF RECORDING

SINGLE FREQUENCY VERSUS

MULTIFREQUENCY ASSRS

An important concern that audiologists need to con-

sider when establishing ASSR clinical test protocols

is the influence of test duration on the accuracy of theASSR

thresholds. Luts and Wouters (2004) explain that both the

accurate estimation of ASSR thresholds aswell as establish-

ing a reasonable testing time are crucial elements for the

clinical applicability of an ASSR technique. These in-vestigators reported that several factors can influence test

duration. These factors include the length of the individual

recording sweeps, the number of sweeps recorded per in-

tensity level, and the number of intensity steps that are

needed to define threshold. In a series of studies, Stapells

and colleagues have looked at the time efficiency of

the single frequency versus the multifrequency ASSR

techniques for both adults and infants, and their resultsare outlined below.

In 2001, Herdman and Stapells compared mean record-

ing times (RTs) needed to establish ASSR thresholds for

the single frequencyversus themultifrequency stimulation

techniques in adults with normal hearing sensitivity. In

this study, ASSRs were recorded using the MASTER soft-

ware, and a “no response” could only be determined after a

total of 48 sweeps had been collected at each stimulusintensity for each CF tone. These same recording param-

eters were applied in their subsequent studies. These

investigators reported that the mean RT needed to obtain

ASSR thresholds for the four CF tones tested in the single

frequency condition was 1646 22min. This mean RT was

almost twice as long as that required to obtain thresholds

for the four CF tones tested in the two multifrequency test

conditions. Therefore, Herdman and Stapells (2001) con-cluded that the RTs for either the monaural or binaural

multifrequency test conditions have the potential of being

up to four to eight times more time efficient, respectively,

than themonaural single frequency test condition, without


164


sacrificing the accuracy of behavioral threshold estima-

tions.

In a subsequent study, Herdman and Stapells (2003)

were interested in determining what mean RTs wereneeded to obtain accurate ASSR thresholds using the

monotic multifrequency technique in adults with vary-

ing degrees of SNHL. They reported that accurate ASSR

thresholds for the four CF tones could be obtainedwithin

60 min for 78% (18/23) of their subjects. The mean RTs

needed to determine accurate ASSR thresholds in adults

with steeply sloping SNHLs were 496 13 min, and 496

14 min for adults with flat or mildly sloping SNHLs.HerdmanandStapells (2003) concluded that themonotic

multifrequency ASSR technique is useful in accurately

estimating behavioral thresholds in individuals with

SNHL in a reasonable amount of time.

Recently, Hatton and Stapells (2011) were interested

in determining whether the single frequency technique

or the multifrequency technique was more time efficient

for recording ASSRs in normal hearing infants. Theyemployed a measure known as “relative efficiency” to

address this issue. “Relative efficiency is ameasurewhich

considers the increase in information relative to the

decrease in response amplitudes that occurs when going

from the single frequency to multifrequency ASSR test

conditions” (Hatton and Stapells. 2011, p. 352). These

investigators calculated relative efficiency (RE) values

using the following formula: RE 5 (AMPi/AMPg) 3

OK. In this formula, AMPi 5 individual subject’s

response amplitude for a specific CF tone and test condi-

tion; AMPg 5 mean group amplitude for a specific CF

tone in the monaural single frequency test condition;

and K 5 number of simultaneous stimuli being pre-

sented. Hatton and Stapells (2011) explained that in

practical terms, if a binaural multifrequency test condi-

tion (i.e., eight stimuli are being presented simultane-ously, four to each ear) had an RE value of 2, this

would predict that audiologists could obtain results for

these eight stimuli in this binaural multifrequency con-

dition in half the time it would take to achieve the same

SNR for the eight stimuli presented separately in the sin-

gle frequency test condition.

Hatton and Stapells (2011) demonstrated that the two

multifrequency test conditions (monaural and binaural)had significantly higher RE values in comparison to the

monaural single frequency test condition in these normal

hearing infants. The mean RE values (collapsed across

carrier frequency) were 1.0, 1.7, and 2.0 for the monaural

single frequency, monaural multifrequency, and binaural

multifrequency conditions, respectively. Given these find-

ings, Hatton and Stapells (2011) concluded that despite

the finding that the ASSR response amplitudes decreasedwhen going from single frequency to multifrequency stim-

uli (as discussed earlier in the section on interaction

effects), the multifrequency technique is still more time

efficient than the single frequency techniquewhen testing

at a supra-threshold level (e.g., 60 dB SPL) in normal-

hearing infants. Additional research is needed to

determine whether these RE findings hold true for

normal-hearing infants at near-threshold levels and/orfor infants with varying degrees of SNHL.

Calibration

An often overlooked yet important area that can

directly affect the accuracy of behavioral threshold

estimations is the calibration of the ASSR stimuli.

At present there is some variation in the physical unitsof measurement that investigators use to calibrate

ASSR stimuli. Researchers tend to calibrate the stim-

uli in either dB HL (e.g., Rance et al, 2006) or in dB

SPL (e.g., Herdman and Stapells, 2003) units. The rea-

son for the discrepancy is in the nature of the stimuli

used for ASSR. The AM and MM stimuli used in ASSR

are similar to the long duration pure tone stimuli

audiologists use in behavioral audiometry and, there-fore, the reference equivalent threshold sound pres-

sure level (RETSPL) used for pure tones should be

the same (Gorga et al, 2004; Stapells et al, 2005). This

has led many researchers and ASSR system manufac-

turers to calibrate ASSR stimuli in dB HL according to

various national or international standards such as

American National Standards Institute (ANSI)

(1996) standards (Gorga et al, 2004; Stapells et al,2005). Other researchers and ASSR system manu-

facturers calibrate ASSR stimuli in the same meas-

urement units (dB peak SPL, dB peak-to-peak

equivalent SPL, and dB nHL) in which ABR stimuli

are calibrated (Stapells et al, 2005). Recently, the

International Electrotechnical Commission (IEC) pub-

lished a new standard (IEC 60645-7) for measuring dB

peak SPL and dB peak-to-peak SPL for these AEPstimuli (IEC, 2009). Stapells and colleagues (2005)

reported that when ASSR thresholds obtained with

stimuli calibrated in dB HL were compared to pure

tone behavioral thresholds, they were found to be ele-

vated. In contrast, when ASSR thresholds were

obtained with stimuli calibrated in dB peak-to-peak

equivalent SPL units they were similar to ABR thresh-

olds evokedwith tones in infants (Stapells et al, 2005). Sta-pells and colleagues (2005) theorized that the response

seen in ASSR thresholds is likely a result of a brief por-

tion of the stimulus, unlike responses found in ABR

testing. The issue of calibration has not been widely dis-

cussed in the literature and is an area in need of future

research.

Potential Clinical Applications of the ASSR

As can be appreciated, ASSR testing has evolved greatly

since itwasfirst describedbyGalambos et al in 1981. There

has been awide range of ASSR stimuli proposed, including

ASSRs/Korczak et al

165


AM, FM, MM, and RSG tones. Each of these has contrib-

uted to our understanding of ASSR generation and pro-

vides different testing advantages. The introduction of

multifrequency stimulation further expanded the complex-

ity and possibilities of ASSR testing by allowing the eval-uation of several test frequencies simultaneously.Objective

response detection techniques provide ASSR testing with

the capability to offer unbiased estimates of behavioral

hearing thresholds. Unlike the ABR, which requires

human expert interpretation of time domain data; ASSR

response detection is by nature objective due to its fre-

quency domain statistical analysis approach. A multi-

tude of studies described earlier have documented the

accuracy of these behavioral hearing estimates. All

these capabilities and improvements have made theASSR into a valuable clinical tool with a wide range

of applications. In this tutorial, the authors have chosen

to concentrate on three clinically relevant applications

of ASSR; these are the use of the ASSR to assess the

functional benefit that individuals with SNHL derive

from their amplification; use of ASSR with cochlear

implant (CI) patients; and use of ASSR with difficult

to test patients, such as infants with perinatal brain

injury and individuals with auditory neuropathy spec-

trum disorder (ANSD).Picton et al (1998) conducted a preliminary investiga-

tion to determine whether the multifrequency ASSR

technique could be used to objectively estimate aided

sound field behavioral thresholds. Thirty-five children

(mean age 5 15 yr) with moderate SNHLs participated

in the study. Picton et al reported that the average differ-

ences between the physiological and behavioral thresh-

oldswere 17, 13, 13, and 16 dB for CFs of 500, 1000, 2000,

and 4000 Hz, respectively. More recently, Stroebel et al(2007) compared aided versus unaided ASSR thresholds

and subsequent behavioral thresholds in six infants with

moderate to profound SNHLs. Single frequency ASSRs

were recorded at 500–4000 Hz. Stroebel and colleagues

reported that aidedASSR thresholdswere obtained for 83%

of the frequencies where aided behavioral thresholds were

subsequently measured. The average difference between

the aided ASSR threshold and the behavioral threshold

was 13 dB (613). Collectively the results of these studies

suggest that theASSRshows promise in objectively assess-ing aided thresholds in subjects who cannot be reliably

tested with behavioral techniques.

During the last decade several investigators have also

looked at the efficacy of recording electrically evoked audi-

tory steady state responses (EASSRs) in CI recipients.

Some of these have been animal studies (Jeng et al,

2007, 2008), while others have been human studies

(Menard et al, 2004; Yang et al, 2008; Hofmann and

Wouters, 2010). One consistent problem that has occurredin recordingEASSRs across studies has been electrical arti-

fact contamination produced by stimulus pulses and radio

frequency (RF) transmission, especially at high stimulus

intensities. Jeng et al (2007) demonstrated that EASSRs

could be successfully recorded from adult guinea pigs by

separating the stimulus artifact from the evoked neural

response using the sum of alternate polarity waveformsand spectral analysis techniques. Similarly, Hofmann

andWouters (2010) reported they were able to successfully

recordand interpretEASSRs to lowpulse trains in six adult

users of the Cochlear Nucleus cochlear implant. These

investigators also employed a variety of artifact rejection

methods to compensate for the electrical artifacts. Overall

these investigators suggest that additional research is

needed in this area to close the gap between exploratorystudies of these issues and clinical practice.

Santiago-Rodrıguez et al (2005) investigated the

accuracy of the ASSR in correctly identifying hearing

loss in 53 infants with confirmed perinatal brain inju-

ries in comparison to their click-evoked ABR results.

For 63% of the infants, ABR results were consistent

with normal hearing; however, ASSR results revealed

only 32% of those same infants had normal hearing.Santiago-Rodrıguez et al (2005) reported that the mul-

tifrequency ASSR had a 100% sensitivity rate but only

a 48.5% specificity rate. Moreno-Aguirre et al (2010)

also evaluated the utility of the ASSR compared to

the ABR for detecting hearing loss in 299 infants with

perinatal brain injury. They reported that the ASSR

had a high sensitivity (92%) and moderate specificity

(68%) for identifying hearing loss in this population.Collectively these findings suggest the ASSR can be

used in conjunction with the ABR to diagnose hearing

loss in infants with perinatal brain injury.

Attias et al (2006) investigated how well the multi-

frequency ASSR predicted BHTs in individuals with

moderate SNHLs, ANSD, and/or CI candidates. They

reported that the ASSR and BHTs were similar in the

SNHL group. In contrast, the ANSD group had signifi-cantly higher ASSR thresholds (1000–4000 Hz) com-

pared to their BHTs while the CI candidates had

exactly the opposite pattern. Attias et al (2006) con-

cluded that the multifrequency ASSR technique

should be used in conjunction with other subjective

and objective measures to ensure the accuracy of

threshold prediction for patients who are CI candi-

dates or have ANSD.

Acknowledgments. The authors wish to thank our three

reviewers for their very helpful and insightful comments on

our manuscript. This is especially true for Dr. David Stapells

whose extensive comments and suggestions lead to a much

improved manuscript.

REFERENCES

American National Standards Institute (ANSI). (1996) AmericanNational Standard Specification for Audiometers. (ANSIS36.1996). New York: ANSI.


166


Aoyagi M, Kiren T, KimY, Suzuki Y, Fuse T, Koike Y. (1993) Opti-mal modulation frequency for amplitude-modulation followingresponse in young children during sleep. Hear Res 65:253–261.

Attias J, Buller N, Rubel Y, Raveh E. (2006) Multiple auditorysteady-state responses in children and adults with normal hear-ing, sensorineural hearing loss, or auditory neuropathy. Ann OtolRhinol Laryngol 115:268–276.

Beattie RC, Kennedy KM. (1992) Auditory brainstem response totone bursts in quiet, notch noise, highpass noise, and broadbandnoise. J Am Acad Audiol 3:349–360.

Beck DL, Speidel DP, Petrak M. (2007) Auditory steady-stateresponse (ASSR): a beginner’s guide. Hear Rev 1412:34–37.

Benjamini Y, Hochberg Y. (1995) Controlling the false discoveryrate: a practical and powerful approach to multiple testing. J RStat Soc Series B Stat Methodol 57:289–300.

Bohorquez J, Ozdamar O. (2008) Generation of the 40-Hz auditorysteady-state response (ASSR) explained using convolution. ClinNeurophysiol 119:2598–2607.

Brooke RE, Brennan SK, Stevens JC. (2009) Bone conductionauditory steady state response: investigations into reducing arti-fact. Ear Hear 30:23–30.

Burkard RF, Don M, Eggermont JJ. (2007) Auditory EvokedPotentials: Basic Principles and Clinical Application. Baltimore:Lippincott, Williams and Wilkins.

Campbell F, Atkinson J, Francis M, Green D. (1977) Estimation ofauditory thresholds using evoked potentials: a clinical screeningtest. Progress in Clin Neurophysiol 2:8–78.

Cohen LT, Rickards FW, Clark GM. (1991) A comparison ofsteady-state evoked potentials to modulated tones in awakeand sleeping humans. J Acoust Soc Am 90:2467–2479.

ConeB, Dimitrijevic A. (2009) The auditory steady-state response.In: Katz J, ed.Handbook of Clinical Audiology. 6th ed. Baltimore:Lippincott Williams and Wilkins, 322–350.

Cone-Wesson B, Dowell RC, Tomlin D, Rance G, Ming WJ. (2002)The auditory steady-state response: comparisons with auditorybrainstem response. J Am Acad Audiol 13:173–187.

Cone-Wesson B, Parker J, Swiderski N, Rickards F. (2002) Theauditory steady-state response: full-term and premature neo-nates. J Am Acad Audiol 13:260–269.

D’Haenens W, Vinck BM, De Vel E, et al. (2008) Auditory steady-state responses in normal hearing adults: a test-retest reliabilitystudy. Int J Audiol 47:489–498.

Dimitrijevic A, John MS, Van Roon P, et al. (2002) Estimating theaudiogram using multiple auditory steady-state responses. J AmAcad Audiol 13:205–224.

Dobie RA, Wilson MJ. (1996) A comparison of t test, F test, andcoherence methods of detecting steady-state auditory-evokedpotentials, distortion-product otoacoustic emissions, or othersinusoids. J Acoust Soc Am 100:2236–2246.

Dobie RA, Wilson MJ. (1998) Low-level steady-state auditoryevoked potentials: effects of rate and sedation on detectability.J Acoust Soc Am 104:3482–3488.

Galambos R, Makeig S, Talmachoff PJ. (1981) A 40 Hz auditorypotential recorded from the human scalp. Proc Natl Acad SciUSA 78:2643–2647.

Geisler CD. (1960) Average responses to clicks in man recordedby scalp electrodes. Technical Report 380. Cambridge, MA: Mas-sachusetts Institute of Technology Research Laboratory of Elec-tronics.

Giraud AL, Lorenzi C, Ashburner J, et al. (2000) Representation ofthe temporal envelope of sounds in the human brain. J Neurophy-siol 84:1588–1598.

Gorga MP, Neely ST, Hoover BM, Dierking DM, Beauchaine K,Manning C. (2004) Determining the upper limits of stimulationfor auditory steady-state response measurements. Ear Hear 25:302–307.

Grason-Stadler Inc. (2001) Auditory Steady-State Evoked Response:A New Tool for Frequency-Specific Hearing Assessment in Infantsand Children [Brochure]. Madison, WI: Grason-Stadler Inc.

Hall JW. (2007) New Handbook of Auditory Evoked Responses.New York: Pearson.

HanD,Mo L, Liu H, Chen J, Huang L. (2006) Threshold estimationin children using auditory steady-state responses tomultiple simul-taneous stimuli. ORL J Otorhinolaryngol Relat Spec 68:64–68.

HariR,HamalainenM,JoutsıniemiSL. (1989)Neuromagnetic steady-state responses to auditory stimuli. J Acoust Soc Am 86:1033–1039.

Hatton J, Stapells DR. (2011) The efficiency of the single- versusmultiple-stimulus auditory steady state responses in infants. EarHear 32:349–357.

HerdmanAT,PictonTW,StapellsDR. (2002)Place specificity ofmulti-ple auditory steady-state responses. J Acoust Soc Am 112:1569–1582.

Herdman AT, Stapells DR. (2001) Thresholds determined usingthe monotic and dichotic multiple auditory steady-state responsetechnique in normal-hearing subjects. Scand Audiol 30:41–49.

Herdman AT, Stapells DR. (2003) Auditory steady-state responsethresholds of adults with sensorineural hearing impairments. IntJ Audiol 42:237–248.

HofmannM,Wouters J. (2010) Electrically evoked auditory steadystate responses in cochlear implant users. J Assoc Res Otolaryngol11:267–282.

Hood LJ. (1998) Clinical Applications of the Auditory BrainstemResponse. San Diego: Singular Publishing Group.

International Electrotechnical Commission (IEC). (2009) IEC60645-7 Electroacoustics-Audiometric Equipment. Part 7: Instru-ments for the Measurement of Auditory Brainstem Response. Gen-eva, Switzerland: International Electrotechnical Commission.

Jeng FC, Abbas PJ, Brown CJ, Miller CA, Nourski KV,Robinson BK. (2007) Electrically evoked auditory steady-stateresponses in guinea pigs. Audiol Neurootol 12:101–112.

Jeng FC, Abbas PJ, Brown CJ, Miller CA, Nourski KV,Robinson BK. (2008) Electrically evoked auditory steady-stateresponses in a guinea pig model: latency estimates and effectsof stimulus parameters. Audiol Neurootol 13:161–171.

Jerger J, Chmiel R, Frost D, Frost JD. (1987) Rate and filter depend-ence of the middle-latency response in infants. Audiology 26:269–283.

John MS, Brown DK, Muir DK, Picton TW. (2004) Recording audi-tory steady-state responses in young infants. Ear Hear 25:539–553.

John MS, Dimitrijevic A, Picton TW. (2003) Efficient stimuli forevoking auditory steady-state responses. Ear Hear 24:406–423.

ASSRs/Korczak et al

167


John MS, Dimitrijevic A, VanRoon P, Picton TW. (2001) Multipleauditory steady-state responses to AM and FM stimuli. AudiolNeurootol 6:12–27.

John MS, Lins OG, Boucher BL, Picton TW. (1998) Multiple audi-tory steady-state responses (MASTER): stimulus and recordingparameters. Audiology 37:59–82.

John MS, Picton TW. (2000) MASTER: a windows program forrecordingmultiple auditory steady-state responses.ComputMeth-ods Programs Biomed 61:125–150.

John MS, Purcell DW. (2008) Introduction to technical principlesof auditory steady-state response testing. In: Rance G, ed. TheAuditory Steady-State Response: Generation, Recording, andClin-ical Application. San Diego, CA: Plural, 11–53.

John MS, Purcell DW, Dimitrijevic A, Picton TW. (2002) Advan-tages and caveats when recording steady-state responses to multi-ple simultaneous stimuli. J Am Acad Audiol 13:246–259.

Johnson BW, Weinberg H, Ribary U, Cheyne DO, Ancill R. (1988)Topographic distribution of the 40 Hz auditory evoked-relatedpotential in normal and aged subjects. Brain Topogr 1:117–121.

Joint Committee on Infant Hearing (JCIH). (2007) Position state-ment: principles and guidelines for early hearing detectionand intervention programs. Pediatrics 120:898–921.

Kiren T, Aoyagi M, Furuse H, Koike Y. (1994) An experimentalstudy on the generator of amplitude-modulation followingresponse. Acta Otolaryngol Suppl 511:28–33.

Kuwada S, Anderson JS, Batra R, Fitzpatrick DC, Teissier N,D’Angelo WR. (2002) Sources of the scalp-recorded amplitude-modulation following response. J Am Acad Audiol 13:188–204.

Kuwada S, Batra R, Maher VL. (1986) Scalp potentials of normaland hearing-impaired subjects in response to sinusoidally am-plitude-modulated tones. Hear Res 21:179–192.

Lauter JL, Karzon RG. (1990a) Individual differences in auditoryelectric responses: comparisons of between-subject and within-subject variability. III. A replication, and observations on individ-uals vs. group characteristics. Scand Audiol 19:67–72.

Lauter JL, Karzon RG. (1990b) Individual differences in auditoryelectric responses: comparisons of between-subject and within-subject variability. IV. Latency-variability comparison in early,middle, and late responses. Scand Audiol 19:175–182.

Lauter JL, Karzon RG. (1990c) Individual differences in auditoryelectric responses: comparisons of between-subject and within-subject variability. V. Amplitude-variability comparisons in early,middle, and late responses. Scand Audiol 19:201–206.

Lauter JL, Loomis RL. (1986) Individual differences in auditoryelectric responses: comparisons of between-subject and within-subject variability. I. Absolute latencies of brainstemvertex-positivepeaks. Scand Audiol 15:167–172.

Lauter JL, Loomis RL. (1988) Individual differences in auditoryelectric responses: comparisons of between-subject and within-subject variability. II. Amplitudes of brainstem vertex-positivepeaks. Scand Audiol 17:87–92.

Levi EC, Folsom RC, Dobie RA. (1993) Amplitude-modulationfollowing response (AMFR): effects of modulation rate, carrier fre-quency, age and state. Hear Res 68:42–52.

Levi EC, Folsom RC, Dobie RA. (1995) Coherence analysis ofenvelope-following responses (ERFs) and frequency-followingresponses (FFRs) in infants and adults. Hear Res 89:21–27.

Linden RD, Campbell KB, Hamel G, Picton TW. (1985) Human audi-tory steady state evoked potentials during sleep.EarHear 6:167–174.

Linden RD, Picton TW, Hamel G, Campbell KB. (1987) Humanauditory steady-state evoked potentials during selective attention.Electroencephalogr Clin Neurophysiol 66:145–159.

Lins OG, Picton TW. (1995) Auditory steady-state responses tomultiple simultaneous stimuli. Electroencephalogr Clin Neuro-physiol 96:420–432.

Lins OG, Picton TW, Boucher BL, et al. (1996) Frequency-specificaudiometry using steady-state responses. Ear Hear 17:81–96.

Luts H, Desloovere C, Kumar A, Vandermeersch E, Wouters J.(2004) Objective assessment of frequency-specific hearing thresh-olds in babies. Int J Pediatr Otorhinolaryngol 68:915–926.

Luts H, Desloovere C, Wouters J. (2006) Clinical application ofdichotic multiple-stimulus auditory steady-state responses in high-risk newborns and young children. Audiol Neurootol 11:24–37.

Luts H, VanDunB, Alaerts J,Wouters J. (2008) The influence of thedetection paradigm in recording auditory steady-state responses.Ear Hear 29:638–650.

Luts H,Wouters J. (2004) Hearing assessment by recordingmulti-ple auditory steady-state responses: the influence of test duration.Int J Audiol 43:471–478.

Luts H, Wouters J. (2005) Comparison of MASTER and AUDERAfor measurement of auditory steady-state responses. Int J Audiol44:244–253.

Makela JP, KarmosG,MolnarM, Csepe V,Winkler I. (1990) Steady-state responses from the cat auditory cortex. Hear Res 45:41–50.

MenardM,GallegoS,TruyE,Berger-VachonC,Durrant JD,Collet L.(2004) Auditory steady-state response evaluation of auditory thresh-olds in cochlear implant patients. Int J Audiol 43:S39–S43.

MoL,StapellsDR. (2008)Theeffect of brief-tone stimulus duration onthe brain stem auditory steady-state response.Ear Hear 29:121–133.

Moeller MP. (2000) Early intervention and language developmentin children who are deaf and hard of hearing. Pediatrics 106:E43.

Moreno-Aguirre AJ, Santiago-Rodrıguez E, Harmony T, Fernandez-Bouzas A, Porras-Kattz E. (2010) Analysis of auditory function usingbrainstem auditory evoked potentials and auditory steady stateresponses in infants with perinatal brain injury. Int J Audiol 49:110–115.

Munnerley GN, Greville KA, Purdy SC, Keith WJ. (1991)Frequency-specific auditory brainstem responses relationship tobehavioural thresholds in cochlear-impaired adults. Audiology30:25–32.

OatesP, StapellsDR. (1998)Auditory brainstem response estimatesof the pure-tone audiogram: current status. Semin Hear 19:61–85.

Ozdamar O, Delgado RE. (1996) Measurement of signal and noisecharacteristics in ongoing auditory brainstem response averaging.Ann Biomed Eng 24:702–715.

Pethe J, Muhler R, Siewert K, von Specht H. (2004) Near-threshold recordings of amplitude modulation followingresponses (AMFR) in children of different ages. Int J Audiol43:339–345.

Picton TW. (2011) HumanAuditory Evoked Potentials. SanDiego:Plural Publishing.


168


Picton TW, Dimitrijevic A, Perez-Abalo M-C, Van Roon P. (2005)Estimating audiometric thresholds using auditory steady-stateresponses. J Am Acad Audiol 16:140–156.

Picton TW, Durieux-Smith A, Champagne SC, et al. (1998) Objec-tive evaluation of aided thresholds using auditory steady-stateresponses. J Am Acad Audiol 9:315–331.

Picton TW, John MS. (2004) Avoiding electromagnetic artifactswhen recording auditory steady-state responses. J Am Acad Audiol15:541–554.

Picton TW, John MS, Dimitrijevic A, Purcell D. (2003) Humanauditory steady-state responses. Int J Audiol 42:177–219.

Plourde G, Picton TW. (1990) Human auditory steady-stateresponse during general anesthesia. Anesth Analg 71:460–468.

Rance G, Briggs RJ. (2002) Assessment of hearing in infants withmoderate to profound impairment: theMelbourne experience withauditory steady-state evoked potential testing. Ann Otol RhinolLaryngol Suppl 189:22–28.

Rance G, Rickards F. (2002) Prediction of hearing threshold ininfants using auditory steady-state evoked potentials. J Am AcadAudiol 13:236–245.

Rance G, Rickards FW, Cohen LT, DeVidi S, Clarke GM. (1995)The automated prediction of hearing thresholds in sleeping sub-jects using auditory steady-state evoked potentials. Ear Hear16:499–507.

Rance G, Tomlin D. (2006) Maturation of auditory steady-stateresponses in normal babies. Ear Hear 27:20–29.

Rance G, Tomlin D, Rickards FW. (2006) Comparison of auditorysteady-state responses and tone-burst auditory brainstem re-sponses in normal babies. Ear Hear 27:751–762.

Regan D. (1982) Comparison of transient and steady-state meth-ods. Ann N Y Acad Sci 388:45–71.

Ribeiro FM, Carvallo RM, Marcoux AM. (2010) Auditory steady-state evoked responses for preterm and term neonates. AudiolNeurootol 15:97–110.

Rickards FW, Tan LE, Cohen LT, Wilson OJ, Drew JH, Clark GM.(1994) Auditory steady-state evoked potential in newborns. Br JAudiol 28:327–337.

Rodrigues GR, Lewis DR. (2010) Threshold prediction in childrenwith sensorineural hearing loss using the auditory steady-stateresponses and tone-evoked auditory brain stem response. IntJ Pediatr Otorhinolaryngol 74:540–546.

Ross B, Borgmann C, Draganova R, Roberts LE, Pantev C. (2000)A high-precision magnetoencephalographic study of human audi-tory steady-state responses to amplitude-modulated tones.J Acoust Soc Am 108:679–691.

Ross B, Picton TW, Herdman AT, Hilyard SA, Pantev C. (2004)The effects of attention on the auditory steady-state response.Neurol Clin Neurophysiol 22:1–4.

Sankoh AJ, Huque MF, Dubey SD. (1997) Some comments onfrequently used multiple endpoint adjustment methods in clinicaltrials. Stat Med 16:2529–2542.

Santiago-Rodrıguez E, Harmony T, Bernardino M, Porras-Kattz E,Fernandez-Bouzas A, Fernandez T, Ricardo-Garcell J. (2005) Audi-tory steady-state responses in infants with perinatal brain injury.Pediatr Neurol 32:236–240.

Savio G, Perez-AbaloMI, Gonzalez A, Valdes J. (2001) The low andhigh frequency steady state responses mature at different rates.Audiol Neurootol 6:279–287.

Small SA, Hatton JL, Stapells DR. (2007) Effects of bone oscillatorcoupling method, placement location, and occlusion on bone-conduction auditory steady-state responses in infants. Ear Hear28:83–98.

Small SA, Stapells DR. (2004) Artifactual responses when record-ing auditory steady-state responses. Ear Hear 25:611–623.

Small SA, Stapells DR. (2005) Multiple auditory steady-stateresponse thresholds to bone-conduction stimuli in adults with nor-mal hearing. J Am Acad Audiol 16:172–183.

Small SA, Stapells DR. (2006) Multiple auditory steady-stateresponse thresholds to bone-conduction stimuli in young infantswith normal hearing. Ear Hear 27:219–228.

Small SA, Stapells DR. (2008a) Maturation of bone conductionmultiple auditory steady-state responses. Int J Audiol 47:476–488.

Small SA, Stapells DR. (2008b) Normal ipsilateral/contralateralasymmetries in infant multiple auditory steady-state responsesto air- and bone-conduction stimuli. Ear Hear 29:185–198.

Spydell JD, PatteeG,GoldieWD. (1985) The 40Hz auditory event-related potential: normal values and effects of lesions. Electroen-cephalogr Clin Neurophysiol 62:193–202.

Stapells DR. (2000) Threshold estimation by the tone-evoked audi-tory brainstem response: a literature meta-analysis. J SpeechLang Pathol Audiol 24:74–83.

Stapells DR. (2011) Frequency-specific threshold assessmentin young infants using the transient ABR and the brainstemASSR.In: Seewald RC, Tharpe AM, eds. Comprehensive Handbook ofPediatric Audiology. San Diego: Plural Publishing, 409–448.

Stapells DR, Galambos R, Costello JA, Makeig S. (1988) Inconsis-tency of auditory middle latency and steady-state responses ininfants. Electroencephalogr Clin Neurophysiol 71:289–295.

Stapells DR, Gravel JS,Martin BA. (1995) Thresholds for auditorybrain stem responses to tones in notched noise from infants andyoung childrenwith normal hearing or sensorineural hearing loss.Ear Hear 16:361–371.

Stapells DR, Herdman A, Small SA, Dimitrijevic A, Hatton J. (2005)Current status of the auditory steady-state responses for estimating aninfant’s audiogram. In: Seewald RC, Bamford JM, eds.ASound Foun-dation Through Early Amplification 2004. Basel: Phonak AG, 43–59.

Stapells DR, Linden D, Suffield JB, Hamel G, Picton TW. (1984)Human auditory steady state potentials. Ear Hear 5:105–113.

Stapells DR, Picton TW, Durieux-Smith A, Edwards CG,Moran LM. (1990) Thresholds for short-latency auditory-evokedpotentials to tones in notched noise in normal hearing and hear-ing-impaired subjects. Audiology 29:262–274.

Stroebel D, Swanepoel D, Groenewald E. (2007) Aided auditorysteady-state responses in infants. Int J Audiol 46:287–292.

Suzuki T, Kobayashi K. (1984) An evaluation of 40-Hz eventrelated potentials in young children. Audiology 23:599–604.

Swanepoel de W, Steyn K. (2005) Short report: establishing nor-mal hearing for infants with the auditory steady-state response.S Afr J Commun Disord 52:36–39.

ASSRs/Korczak et al

169


Tlumak AI, Rubinstein E, Durrant JD. (2007) Meta-analysis ofvariables that affect accuracy of threshold estimation via meas-urement of the auditory steady-state response (ASSR). Int JAudiol 46:692–710.

vanderReijdenCS,MensLHM,SnikAFM. (2001)Comparing signalto noise ratios of amplitude following responses from four EEG-derivations in awake normally hearing adults. Audiology 40:202–207.

van der Reijden CS, Mens LH, Snik AFM. (2005) EEG derivationsproviding auditory steady-state responses with high signal-to-noise ratios in infants. Ear Hear 26:299–309.

Van Maanen A, Stapells DR. (2009) Normal multiple auditorysteady-state response thresholds to air-conducted stimuli ininfants. J Am Acad Audiol 20:196–207.

Van Maanen A, Stapells DR. (2010) Multiple-ASSR thresholds ininfants and young children with hearing loss. J Am Acad Audiol21:535–545.

Yang C-H, Chen H-C, Hwang C-F. (2008) The prediction of hearingthresholds with auditory steady-state responses for cochlearimplanted children. Int J Pediatr Otorhinolaryngol 72:609–617.

Yoshinaga-Itano C, Sedley AL, Coulter DK, Mehl AL. (1998) Lan-guage of early- and later- identified children with hearing loss.Pediatrics 102:1161–1171.

Appendix A. Glossary

Key Terms

Carrier frequency Associated with the region in the

cochlea where the hair cells are activated in response to

the presentation of a stimulus

Modulation frequency The frequency at which elec-

troencephalography (EEG) activity is synchronized tofire and can be derived by calculating the period of

the modulation frequency.

Types of Stimuli Used in ASSR

Amplitude modulated (AM) tone A pure tone that

changes in amplitude over time.Blackman-gated tone Commonly used type of RSG

(repeating sequence gated tone) tone. These tones differ

from other RSG tones in three ways: (1) the width of the

main peak of energy, (2) the height of the side lobes of

energy, and (3) the rate of decay for the side lobes of energy.

Chirp A type of stimulus that covers a broader range of

frequencies than traditional modulated pure tones,

activating more hair cells.ClickAverybrief-duration stimulus (usually 100microsec)

with a broad frequency spectrum (z100–10,000Hz), which

is produced by a transient electrical pulse (Hall, 2007).

Frequency modulated (FM) tone A pure tone that

changes in frequency over time.

Mixedmodulated (MM) toneA pure tone that changes

in both frequency and amplitude over time.

Repeating sequence gated (RSG) tone A series ofgated tones that can be combined to form either a single

frequency tone or a multiple frequency tone.

Tone burst A brief (,1 sec) tonal stimulus that is fre-

quency specific.

Stimulation Techniques

Multiple frequency Amethod of stimulation that pre-

sentsmultiple carrier frequency tones (up to four in each

ear) simultaneously. These carrier frequency tones are

presented either to one ear (monaural test condition) or

to both ears (binaural test condition).

Single frequency A method of stimulation that pre-sents one carrier frequency tone at one modulation fre-

quency to one ear at a time.

Analyses Techniques

Fast-Fourier transform (FFT) analysis A compu-

terized technique for separating a complex waveform

consisting ofmultiple frequencies into its individual fre-

quency components (Hall, 2007).

F-test (or F-ratio) A statistical method that is applied

in auditory steady state response (ASSR) testing to esti-mate the probability that the amplitude of an ASSR

found at a particular modulation frequency is statisti-

cally different from the energy found at the surrounding

frequencies that are attributed to the ongoing electro-

encephalography (EEG) noise.

Phase coherence Phase coherence “is related to the

signal (response)-to-noise (background EEG and myo-

genic) ratio” (Cone and Dimitrijevic, 2009, p. 333).

Neuro-Imaging Techniques

Brain Electrical Source Analysis (BESA) Software

for source analysis and dipole localization that is used in

electroencephalography (EEG) and magnetoencephalo-

graphy (MEG) research.

Functional magnetic resonance imaging (fMRI) Atype of magnetic resonance imaging (MRI) that meas-

ures the changes in blood flow in various areas of the

brain that are related to underlying neural activity.

Magnetoencephalography (MEG) Technique used

to measure magnetic fields produced by electrical activ-

ity in the brain.

Terms Associated with ThresholdFrequency specificity of the response “How inde-

pendent a threshold at one stimulus frequency is of con-

tributions from surrounding frequencies” (Oates and

Stapells, 1998, p. 61). This refers to behavioral threshold

estimations.

Mean difference score (MDS) The behavioral pure

tone threshold minus the ASSR threshold equals the

difference score. This is calculated separately for eachcarrier frequency.

Place specificity Place specificity of the response ref-

ers to the specific point along the basilarmembrane that

has beenmaximally activated by the stimulus (Herdman

et al, 2002).


170

tutorial auditory steady-state responses · response, as well as information regarding how...

Documents