It is apparent that auditory steady-state response (ASSR) testing, while still in the early phases of development, has excellent potential for use by dispensing professionals and diagnostics in real-world clinical environments. They can provide a non-behavioral measure of hearing loss by frequency that is often unavailable, unreliable, or impractical via acoustic reflexes, OAEs, and ABRs. This article provides an introduction to ASSRs and, through the use of many examples, demonstrates the utility and potential limitations of the ASSR tests.
In the world of audiometry, nothing beats a reliable, behavioral audiogram for revealing true hearing thresholds across the frequencies. A second-best option, however, is a non-behavioral test that can estimate hearing thresholds.
There are some clinical populations who cannot give a voluntary response, including infants, mentally disabled individuals, and those who do not want to reveal their true hearing thresholds. Examples of non-behavioral tests are acoustic reflexes, which came into clinical practice during the 1970s, the auditory brainstem response (ABR), which became popular during the 1980s, and otoacoustic emissions (OAEs) which arrived at our doorsteps during the 1990s. Over the last couple years, another non-behavioral test has emerged: the new auditory steady-state response (ASSR).
Threshold prediction via acoustic reflexes is not commonly utilized because the sheer variability among results precludes any easy conclusions about specific degrees of hearing loss. OAEs provide frequency-specific information about the state of one’s outer hair cells, but are not so effective at determining the degree of one’s hearing loss at specific frequencies. The tone-burst ABR can provide frequency-specific estimates about the degree of one’s hearing loss, but they can be time consuming.
ASSRs hold the promise of delivering information similar to the tone-burst ABR with much faster results. The tone-burst ABR can take some 2 hours to obtain ABRs to three or four frequencies from both ears. It is sometimes difficult for babies to remain calm for this length of time. The ASSR, however, can reduce this time dramatically—down to some 30-40 minutes! According to Cone-Wesson et al.,1 both the tone-burst ABR and the ASSR render similar estimates of actual behavioral pure-tone thresholds. In general, ABR and ASSR thresholds are at least 10 dB greater than behavioral thresholds for the mid-to-high frequencies, and this difference increases to some 20 dB for the low frequencies.2
Another potential advantage of the ASSR is that it may be used to estimate aided versus unaided hearing thresholds in a sound field.3 Distortion can occur when ABR tone-bursts (with their short-onset rise/fall times) are played over loudspeakers in a soundfield and when received by non-linear hearing aids. Preliminary research, however, suggests that continuous modulated tones of ASSR lend themselves much more readily to these conditions.
Currently, the new ASSR may be viewed as a work in progress; complete normalized data for final stimulus and recording parameters have not been finalized. In fact, new important research pertaining to ASSRs seems to be published every month (see suggested reading at the end of this article), and this information contributes to the utility and understanding of the new ASSR tests. The main purpose of this article is to provide a general description of the ASSR, and to show exactly what the title of this article says: namely, one clinician’s encounter with the new ASSR when testing some 17 subjects. I would stress that I am, by no means, an authority on the new ASSR; I’ve simply set about exploring its usage and application in a real-world clinical environment.
It is my hope that readers will look at some of the successes and pitfalls that may be encountered when first using this exciting new test. The particular ASSR methodology used in this study was one that involves the simultaneous presentation of four different frequencies to both ears at the same time. This multiple auditory steady-state evoked response paradigm has been given the acronym of MASTER™ and is implemented in a system developed by Bio-logic Systems Corp. It should be noted that several other ASSR systems exist, and may perform comparably.
What is a Steady State Response?
It may help by first looking at what ASSR is not. ASSR is not a transient response like the ABR. The ABR is a transient response to a single transient stimulus, and ABR testing measures the neural responses of the VIII nerve and lower brainstem over a time frame of about 10 milliseconds (ms). The ABR response stops after each stimulus presentation, and does not begin again until the next stimulus presentation. Stimulus repetition rate (per second) assesses the length of the response time-frame (100 ms), and thus cannot exceed 1 second. Repeated stimulation and computerized signal averaging improve the signal-to-noise ratio of the ABR, making it visible from the rest of the brain’s activity (ie, activity unrelated to sound stimulus).
The new ASSR on the other hand, is a “steady state” response. It is a continuous, ongoing neural response because its waveform follows the waveform of the continuous ongoing stimulus. Such a true sustained, steady-state response is phase-locked to the stimulus; it occurs slightly later in time than the stimulus, but faithfully follows the continuous temporal waveform envelope of the stimulus.
What About the 40 Hz Response; Isn’t That a Steady State Response?
The 40-Hz response is a variation of the middle latency response (MLR). The MLR typically uses a response time frame of 100 ms to view the responses of the temporal lobes. The main landmark in the MLR is a peak called “Pa,” which occurs at around 25 ms (Figure 1, right panel). Here, with the stimulus at the beginning of the 100 ms time frame, the stimulus presentation rate cannot exceed 10-per-second. What differentiates the 40-Hz response from the MLR is its stimulus repetition rate of around 40 times per second (hence, the name “40 Hz”). This 40-per-second repetition rate results in a deliberate overlapping of the 100 ms response time frames. As a result, the 40-Hz response has four peaks occurring at 25 ms intervals, making it look much like a 40 Hz sinusoid (Figure 1, left panel). The 40 Hz response is called a “steady state” response because it is a reflection of the stimulus repetition rate.
Figure 1. Comparison of 40 Hz tone burst (ABR) and the middle latency response (MLR) in an ASSR test for a subject age 40.
The great hope for the 40-Hz response was to reliably estimate low frequency hearing thresholds. Low frequency threshold estimation is a notorious audiometric shortcoming of the ABR, because the ABR is an “onset response.” This means the ABR is most easily seen if it arises from the simultaneous stimulation of thousands of basal, high-frequency cochlear hair cells. The best ABR stimulus is a “click,” which is a extremely short (1/10,000th of a second) sound wave with an almost instant onset time (ie, it really does sound like a click). Low frequency threshold estimation with the ABR requires tone bursts with longer rise/fall times, and these do not elicit robust ABRs. In general, the lower the tone burst frequency, the longer its onset time, and the harder it becomes to see the resultant ABR.
The 40-Hz response was seen as an alternative way to estimate low frequency thresholds, but it was found to have some major shortcomings. First, it is not very robust in infants—one of the main populations needing non-behavioral test measures! Furthermore, the 40-Hz response is very susceptible to sleep and attention factors in adults. These factors have greatly limited its clinical application.
The new ASSR, on the other hand, uses continuous tonal stimuli (or carrier frequencies). These carrier frequencies are rapidly increased and decreased in amplitude (modulated) at frequencies from around 80 Hz to 100 Hz. The new ASSR determines if the brain’s neural response follows these modulations over the much longer time frames (about 16 seconds compared to less than 1 second for ABRs). The main idea is to see if the brain’s activity is phase locking to the ongoing changes or modulations in the continuous tonal stimuli. The assumption is that, if the brain can detect stimulus modulation, then it can also detect the carrier stimuli. The faster rates used in the new ASSR work much better in infants; furthermore, they do not seem to be affected by the sleep state.4
Figure 2. Stimuli used in the Bio-logics Master™ system.
ASSR Stimulus Construction
Stimulus construction for the ASSR is quite complicated—but also fascinating. Many ASSR test devices enable presentation of one carrier frequency (along with its modulating frequency) at a time. The MASTER™ system for ASSR testing, for example, involves the presentation of four simultaneous carrier frequencies to both ears at the same time. The default carrier tones are 500 Hz, 1000 Hz, 2000 Hz, and 4000 Hz. Each carrier frequency in each ear is assigned its own unique modulation frequency for a total of 8 different modulation frequencies (Figure 2). Additionally, the modulation of these carrier tones involves both frequency modulation (FM) and amplitude modulation (AM), and the whole combination is called mixed modulation (MM). Preliminary research shows the MM-ASSR to be more robust,5 and the particular MM used is as follows:
- The AM is fixed at 100%, meaning that the amplitude fluctuates from full maximum to zero at rates between 80-per-second to almost 100-per-second. Recall that each carrier tone is assigned its own separate modulation frequency.
- The FM is set at 20%, meaning that the frequency fluctuates between +10% of the carrier tone frequency.
- Lastly, there is a specific relationship between the AM and FM, where the maximum AM occurs at -90° phase relative to the carrier tone center frequency.
The system has recently implemented exponential modulation in order to improve the ASSR acquisition at 500 Hz.
ASSR: How It’s Measured
Like the ABR, the ASSR employs the use of electrodes, filtering, differential amplification, and signal averaging. For the differential amplifier to do its job of increasing the response signal-to-noise ratio, the electrodes should be placed so that each one sends different voltages to the differential amplifier. The ASSR can use a single-channel vertical electrode montage, with one electrode placed at the mid-hairline of the forehead (Fz location), and the other at seventh cervical vertebra (C7). The ground electrode can be placed almost anywhere (eg, the clavicle or between the eyes). A real advantage of a single-channel vertical montage is that it can be used to test both ears individually without moving any electrode location.
The filters used to record the ASSR are set lower than those used for the ABR. The MASTER™ system, for example, utilizes 1 Hz-300 Hz filter settings because the spectral energy of the ASSR is at the 8 modulating frequencies of the 4 carrier tones to each ear—roughly between 80-100Hz. In contrast, the ABR has lots of spectral energy between 30-3000 Hz.
Figure 3. ASSRs, polar plots, and phase coherence.
The actual ASSR waveform is so tiny that it is measured in nanovolts or billionths of a volt. In contrast, the ABR waveform is measured in microvolts or millionths of a volt. Perhaps the tiny size of the ASSR waveform explains why there are presently numerous methods used to obtain ASSR measurements, as well as various ways of displaying the response itself. Pioneering research6 on the ASSR, using equipment developed in Australia, looked at ASSRs in terms of polar plots with phase coherence (Figure 3). Today, the GSI Audera system uses this method for ASSR analysis. Vivosonic Inc is looking into the usage of their own predictive algorithm (Vivography™) to determine “true ASSRs.”7 There are still some areas of disagreement relative to optimal tone-burst construction, electrode montages, and filtering for the ABR. Almost every system, however, looks at the ABR as a waveform in the time domain.
Figure 4A-c. The progressive recording of an ASSR test. The left panels show the ASSR waveform, while the right panels show the ASSR frequency spectrum (FFT). The vertical amplitude of the wave can be seen as a function of both time and frequency. Figure 4a shows the progress of ASSR acquisition after one sweep, Figure 4b after 16 sweeps, and Figure 4c after 20 sweeps. The height of the yellow bars (FFTs) indicate the modulating frequencies (ie, the higher the better).
In the case of the MASTER system, the frequency domain—the vertical amplitude as a function of frequency—is also displayed. Figures 4-6 show the progression of ASSR recording in a subject. The left panels show the ASSR waveform, and the right panels show the ASSR frequency spectrum or “fast fourier transform” (FFT). The large bump in the waveform of the left panels is the patient’s heartbeat, occurring at a rate of roughly once-per-second. The FFTs on the right display the amplitude of the modulating frequencies relative to those of adjacent frequencies. For each modulating frequency, an F ratio with a 95% confidence is required in order to determine if an ASSR is truly present. Again, it should be remembered that the underlying assumption regarding ASSRs is that, if the brain can detect stimulus modulation, then it can also detect the carrier stimuli.
Methods Used to Collect ASSRs
A total of 17 subjects (3 males, ages 38 to 48 years, and 14 females, ages 2 months to 75 years) were tested with the MASTER™ ASSR. While 10 subjects had normal hearing, the remaining 7 subjects had varying degrees of hearing loss. Electrodes were placed in the vertical (Fz-C7) montage, and the stimuli were binaurally sent through insert (ER-3A) headphones. The subjects were instructed to lie down, close their eyes, relax as much as possible, and try to sleep while being tested with the ASSR.
The collection method of the system is illustrated in Figures 4a-c, which show the progression of binaural ASSR acquisition at a 60 dB HL intensity level. The ASSR waveform on the left of each figure is shown over a time span of 1024 ms (about 1 second). Each of these time spans is called an epoch, and 16 of these epochs is called a sweep. Each sweep is added to a total running average of sweeps with relatively huge time frames of 16 seconds. The collection default is to obtain 32 sweeps at four intensity levels: 60 dB HL, 50 dB HL, 40 dB HL, and 30 dB HL (but one can decrease or increase the number of sweeps, depending on the criterion for obtaining a reliable ASSR).
Since the sweeps comprise roughly 16 seconds each, a collection of 32 sweeps takes about 9 minutes. As soon as these are collected, the intensity automatically drops by 10 dB HL to 50 dB HL, and the process begins again. Four of these uninterrupted acquisitions, from 60 dB HL down to 30 dB HL, thus take about 36 minutes. Figure 4a shows the progress of ASSR acquisition after only one sweep has been recorded, Figure 4b shows the progression after 16 sweeps, and Figure 4c shows the final successful ASSR acquisition after 20 sweeps. Note how the FFTs in these successive figures show increasing energy at the modulation frequencies of the carrier tones relative to the adjacent bars that represent adjacent (non-modulating) frequencies. In other words, the height of the yellow bars indicates the energy at the modulating frequencies—the higher the better.
The eight colored circles below each carrier frequency for right and left ears indicate whether an ASSR has been successfully obtained or not. Default colors are red, yellow, and green (these can be changed to suit the clinician). For any carrier frequency, red means the F value is greater than 0.1 (ie, no ASSR available for that frequence; there is less than 90% probability that a statistically significant difference exists between the amplitude of the modulating frequency versus those for the adjacent frequencies in the FFT). In simpler terms, there is a greater than 10% probability that the differences in amplitude are simply due to chance. A yellow circle indicates an F value of 0.1 to 0.05. That is, caution should be used in determining an ASSR value. A green circle indicates an F value of less than 0.05, indicating that there is an ASSR for that particular carrier frequency. More specifically, there is a probability of more than 95% that a statistically significant difference exists between the amplitude of the modulating frequency (for that specific carrier frequency) versus those for the adjacent frequencies in the FFT. In other words, there is less than 5% probability that the differences in amplitude are simply due to chance.
These decision criteria bring up yet another important difference between the ABR and the ASSR. Most clinicians determine the presence or absence of an ABR by subjectively analyzing the waveform for latencies, amplitudes, and overall morphology. In contrast, artificial intelligence is used to determine presence versus absence of an ASSR.
Subject ASSR Results
In this field study, the ASSRs were actually more robust when the clients were sleeping. In the author’s experience, a completely relaxed subject state seems to be even more important for optimally recording the ASSR than for the ABR.
Figures 5-13 show the summarized results for 8 of the 17 subjects tested. Two normal-hearing subjects are shown in Figures 5 and 6, while Figures 7-13 show show a variety of audiometric configurations. Each figure shows the behavioral audiogram and the ASSR results, where the carrier frequencies are displayed across the top, and the dB levels are displayed along the left vertical side. The box colors indicate whether or not ASSRs were successfully obtained for a specific carrier frequency at some intensity level. For any particular box, if significance (p) of the numbers are less than 0.05, then an ASSR is present (on this system, the box is green); if the number is between 0.05 and 0.10, an ASSR is to interpreted with caution (the box is yellow); if the number is 0.10 or more, an ASSR is absent (the box has no color). For some subjects, blank boxes appear across all frequencies at some specific intensity level. If the number of sweeps is low (eg, 1 or 2 sweeps), this means the measurement was aborted and tried again later. For many subjects, it was also noted that, at times, the ASSR would appear at a particular frequency and then, disappointingly, it would disappear again.
Of the 10 normal-hearing subjects, only 2 had consistently measurable ASSRs for both ears at all frequencies (500 Hz, 1000 Hz, 2000 Hz, and 4000 Hz) at all 10 dB steps down to 30 dBHL. Of the normal-hearing subjects, 5 had consistently measurable ASSRs for both ears at all frequencies at all 10 dB steps down to 40 dBHL. All remaining normal-hearing subjects, however, did show ASSRs at some frequencies at intensities of 40 dBHL and 30 dBHL; however, at these lower intensities, responses were absent at one or more frequencies (usually 500 Hz and 4000 Hz).
Figure 5. ASSR of a 44-year-old female with normal hearing. Not complete responses from 60 dB HL down to 30 dB HL at all 10 dB steps. (A few responses are even apparent at 20 dB HL!)
Figure 5 shows the final summarized ASSR results for the normal-hearing subject shown in Figures 4a-c. Note the green boxes showing present ASSRs for the right ear (upper right panel) and left ear (lower right panel) at all intensities (30 dBHL-60 dBHL in 10 dB steps) at 500 Hz, 1000 Hz, 2000 Hz, and 4000Hz. Note also how ASSRs in this subject are also present at 20 dBHL at some frequencies for each ear. This is also a subject who routinely renders a pristine ABR for 250 Hz tone bursts at a sensation level of only 10 dB!
Figure 6. ASSR of a 48-year-old male (the author) with normal hearing. Note no response at 2000 Hz at 60 dB for left ear, incomplete responses at 30 dB HL and 40 dB HL for both ears, and no responses at 20 dB HL.
Figure 6 shows results that are more typical for normal-hearing subjects (specifically, this figure shows the summarized results for the author, who has normal hearing). Some boxes show missing ASSRs, especially at 40 dB and 30 dB levels, and at 500 Hz and 4000 Hz. For the left ear, no ASSR is indicated at 2000 Hz, even at 60 dBHL. At a lower intensity level, an ASSR is indicated. Basic intuition should make the reader ask how it is possible to see an ASSR at a lower intensity while not at a higher intensity. As such, all ASSRs for the left ear at 2000 Hz should theoretically be ignored. Interestingly, when testing my own ABRs, I routinely render a barely readable ABR for 250 Hz tone bursts at a sensation level of 20 dB.
In the author’s experience, the ASSRS were not quite as close to actual behavioral thresholds as those reported by Herdman et al.8 The results reported here, however, do show normal-hearing subjects to produce ASSRs at 40 dbHL and at 30 dBHL. In most normal-hearing subjects, the ASSRs for some carrier frequencies were missing, especially at 30 dBHL.
A large advantage of the ASSR is that it is much faster (35-45 minutes) than tone-burst ABR. On conventionally available clinical equipment, tone-burst ABRs done separately for each ear at 500 Hz, 1000 Hz, 2000 Hz, and 4000 Hz—along with the required replications—would take much longer. Another strength of the ASSR appears to be that it shows the general configuration of the hearing loss. It should be noted that the author usually adhered to the default number of sweeps (32) in order to obtain ASSRs for each subject. Bio-logic has since found that the time to obtain ASSRs can be reduced even further by employing a simple rule of thumb: look for 2-3 consecutive sweeps where low noise is evident (10-15 nV), then reduce stimulus intensity and begin recording again. For very high intensities, if there is no ASSR after 10-12 sweeps, then an ASSR will probably not be obtained.
Figure 7. ASSR of 45-year-old female with otosclerosis. She had moderate hearing loss. The ASSR intensity levels begin at 80 dB HL and descend in 10 dB steps to 40 dB HL. For this subject, the ASSRs were obtained only at 1000 Hz and 2000 Hz for the right ear, and at 1000 Hz for the left ear at 70 dB HL.
Figure 8. ASSR of 33-year-old female with sudden idiopathic sensorineural hearing loss (SSNHL). The SSNHL occurred about 2 years ago. ASSRs show the normal hearing for the right ear and the HL configuration for the left ear (response was obtained for the left ear at 500 Hz at 70 dB HL). Low-level responses are invalid if not present at higher levels.
Figure 9. ASSR of a 46-year-old female with severe-to-profound high frequency hearing loss. Note how the ASSR indicates the configuration of HL (ie, no responses at 2000 Hz and 4000 Hz for both ears with responses present at 500 Hz and 1000 Hz for both ears).
Figure 10. ASSR of a 21-year-old female with hereditary SNHL. She had severe-to-profound mid-high frequency hearing loss. In this case, the ASSR results were disappointing because no ASSRs are indicated at all—even though the client has only a mild SNHL for 500 Hz (ie, subsequently, the clinician would expect a 500 Hz ASSR response).
Figure 11. ASSR of a 75-year-old female with presbycusis. She had a typical moderately sloping SNHL. The ASSR summary faithfully indicates the sloping configuration of HL. Note consistent responses at low frequencies.
Figure 12. ASSR of a 40-year-old female with otitis media. Note the missing responses at low frequencies. Responses again follow the configuration of the hearing loss.
Figure 13. ASSR of the author, who has normal hearing, tested for a second time 2 weeks after the testing shown in Figure 6. Note the somewhat better responses here compared to 2 weeks earlier.
A few general observations from this field study:
1. Subjects known to exhibit “textbook ABRs” also tend to exhibit the best ASSRs; that is, their ASSRs were measured the fastest and at lowest intensity levels.
2. There was a tendency for normal-hearing subjects not to produce ASSRs at either 500 Hz or 4000 Hz, and this tended to occur especially at 30 dBHL.
3. With the exception of only 1 in 17 subjects (Figure 12), the ASSR seems quite effective at estimating the slope or configuration of the behavioral audiogram.
4. Subjects who were sleeping actually tended to render the best ASSRs. This is even more noticeable with the ASSR than the ABR, and it is in direct opposition to the 40-Hz steady state response.
5. The ASSR, once having appeared, has an annoying tendency to disappear after further signal averaging. That is, sometimes when displaying a “green circle” at a carrier frequency, the circle would later turn red. Although this reduction of signal-to-noise ratio can also be noted in the ABR, it seems to occur more frequently with ASSR testing.
6. There is as yet a plethora of methods of ASSR stimulation, recording, and analysis among research groups and manufacturers.
7. Unlike the ABR, the ASSR relies on artificial intelligence in order to determine presence (versus absence) of a response.
As mentioned in the introduction, the specifics of the new ASSR remains unfinished, and there is further work to be done. For example, Bio-logic is now implementing “exponential” rise/fall times for the carrier tone modulations, which they report is showing improved responses at 500 Hz and 4000Hz.
The promise of the ASSR is bright, and the horizons are colorful. One needs only to be reminded about its potential usage in comparing aided and unaided thresholds to understand that ASSRs may provide invaluable data for dispensing professionals in everyday testing environments.
This article and field study was prepared without remuneration from outside entities. The author did consult and was assisted in the preparation of this article by several parties including his wife, Laura Venema; Tang Chow of Electro Medical, Mississauga, Ontario, who graciously lent the lab the Bio-logic equipment; Lynne Hemmerich, a hearing instrument specialist in Kitchener, Ontario; Vivosonic Inc, Toronto; and the Biologic staff in Chicago.
|Ted Venema, PhD, is an assistant professor at the University of Western Ontario, London, Ontario. He is a frequent lecturer at the major hearing-related conventions and is author of the book, Compression for Clinicians.
Correspondence can be addressed to HR or Ted Venema at [email protected].
1. Cone-Wesson B, Dowell R, Tolin D, Rance G, Min WJ. The auditory steady-state response: Comparisons with the auditory brainstem response. J Amer Acad Audiol. 2002; 13 (4).
2. Stapells D. The tone-evoked ABR: Why it’s the measure of choice for infants. Hear Jour. 2002; 55(11).
3. Picton T, Dimitrijevic A, van Roon P, Sasha John M, Reed, Finkelstein. Possible roles for the auditory steady-state responses in fitting hearing aids. In: Proceedings of A Sound Foundation Through Early Amplification, 2nd Int’l Conference; Warrenville, Ill: Phonak; 2002.
4. Stach B. Page Ten: The auditory steady-state response: A primer. Hear Jour. 2002; 55(9).
5. Sasha JM, Dimitrijevic A, van Roon P, Picton T. Multiple auditory steady-state responses to AM & FM stimuli. Audiol Neuro-otol. 2001; 5:12-27.
6. VanderWerff KR, Brown CJ, Gienapp BA, Schmidt-Clay KM. Comparison of auditory steady-state response and auditory brainstem resonse thresholds in children. J Am Acad of Audiol. 2002; 13(5):227-235.
7. Li X, Wodlinger H, Sokolov Y. A new method for measuring DPOAEs and ASSRs. Hearing Review. 2003; 10(2):44-47.
8. Herdman AT, Stapells D. Thresholds determined using monotic and dichotic multiple auditory steady-state response technique in normal-hearing subjects. Scand Audiol. 2001; 30:41-49.
1. Dimitrijvic, Sasha John, van Roon, et al, Estimating the audiogram using multiple auditory steady-state responses J Am Acad Audiol. 2000; 13: 205-224.
2. Payne-Stueve M, ORourke C. Estimation of hearing loss in children: Comparison of auditory steady-state response, auditory brainstem response, and behavioral test methods. Am J Audiol. 2003; 12(2).
3. Picton T, Dimitrijevic A, Sasha John M, van Roon P. The use of phase in the detection of auditory steady-state responses. Clin Neurophysiol. 2001; 112:1698-1711.
4. Rance G, Rickards F, Cohen L, De Vidi S, Clark G. The automated prediction of hearing thresholds in sleeping subjects using auditory steady-state evoked potentials. Ear Hear. 1995; 16: 499-507.
5. Rance G. Prediction of hearing thresholds using steady-state evoke potentials. J Am Acad Audiol. 2002; 13(5).
6. Sasha JM, Dimitrijevic A, Picton T. Weighted averaging of steady-state responses. Clinical Neurophysiol. 2001; 112:555-562
7. Sasha JM, Purcell D, Dimitrijevic A, Picton T. Advantages & caveats when recording steady-state responses to multiple simultaneous stimuli J Am Acad Audiol. 2002;13: 246-259.