Auditory Brainstem Response to Tone Bursts in Quiet, Notch Noise, Highpass Noise, and Broadband Noise

Transcription

1 11 1'i +:l ; I I,iai J Am Acad Audiol 3 : (1992) Auditory Brainstem Response to Tone Bursts in Quiet, Notch Noise, Highpass Noise, and Broadband Noise Randall C. Beattie* Kristina M. Kennedy* Abstract This study investigated the effects of tone bursts (1000 Hz, 2000 Hz, and 4000 Hz) in quiet, notch noise, highpass noise, and broadband noise on the identifiability, latency, and amplitude of the auditory brainstem response (wave V). Normal listeners were presented with 40 db and 80 db nhl tone bursts having rise-plateau-fall times of 1 msec. Wave V was observed in all subjects at 40 db and 80 db nhl forthe quiet and noise conditions. The latency findings suggest that responses elicited by the 80 db nhl tone bursts in quiet were, in part, mediated by regions on the basilar membrane that did not correspond to the center frequency of the tone burst. To increase frequency-specificity, high-level tone bursts (e.g., 80 db nhl) should be mixed with notch, highpass, or broadband noise. The use of noise conditions for low intensity levels (e.g., 40 db nhl) does not appear necessary for isolating the response because both the notch and the highpass conditions yielded latencies similar to the quiet condition. Although si milar wave V amplitudes were found at all frequencies, amplitudes were smaller for the broadband noise than for the quiet, notch, and highpass conditions. Thus, the latter conditions seem preferred. Key Words : Auditory brainstem response, normal listeners, tone bursts, notch noise, highpass noise, broadband noise uditorybrainstem response (ABR) audiometry is useful to estimate auditory A thresholds with individuals who cannot be tested using conventional procedures (Jerger and Mauldin, 1978 ; Stein et al, 1981 ; Hall, 1984 ; Davis et al, 1985 ; Schwartz and Schwartz, 1991). Clicks commonly are used in ABR audiometry because they optimize response identifiability (Kileny, 1981 ; Hayes and Jerger, 1982 ; van Zanten and Brocaar, 1984 ; Hood and Berlin, 1986 ; Gorga and Thornton, 1989). However, clicks are characterized by a broad frequency spectrum that is shaped by the frequency response of the transducer. This lack of frequency specificity does not allow an accurate estimate of audiometric configuration (Jerger and Mauldin, 1978 ; Hood and Berlin, 1986). *Department of Communicative Disorders, California State University, Long Beach, Long Beach, California Reprint requests : Randall C. Beattie, Department of Communicative Disorders, California State University, Long Beach, Long Beach, CA Furthermore, clicks complicate comparisons between clinics because an identical electric waveform can yield substantially different spectra among earphones (Weber et a1,1981; Laukli, 1983a). Tone bursts are recommended because they are more frequency specific than clicks and their spectra are not highly dependent on the frequency response of the earphone (Laukli, 1983a). The degree of frequency specificity is dependent on rise, plateau, and fall times ; longer durations have narrower bandwidths. However, because ABRs are critically dependent on stimulus onset (Brinkmann and Scherg, 1979 ; Debruyne and Forrez, 1982), longer durations yield fewer responses and smaller amplitudes. Although it is desirable to test at 500 Hz, several investigators have reported that this frequency yields smaller amplitudes and/or larger intersubject variability than higher frequencies (Don et al, 1979 ; Hayes and Jerger, 1982 ; Laukli,1983b ; Gorga et a1,1988 ; Fjermedal and Laukli, 1989 ; Beattie and Spence, 1991). Thus, it appears that the most successful toneburst stimuli for eliciting ABRs are frequencies 349

2 Journal of the American Academy of Audiology/Volume 3, Number 5, September 1992 of 1000 Hz and above. As a compromise between frequency specificity and response identifiability, tone bursts with rise-plateau-fall times of approximately 1 millisecond (msec) are selected. Several authors have suggested that an ABR evaluation for auditory sensitivity should include at least one high-frequency stimulus and one low-frequency stimulus (Kileny, 1981 ; Eggermont, 1982 ; Jerger et al, 1985 ; Fjermedal and Laukli,1989 ; Gorga et al, 1991). This allows clinicians to construct a two-point audiogram, which serves as a basis for making rehabilitative decisions. Investigators do not agree, however, on preferred stimuli for estimating threshold (Don and Eggermont, 1978 ; Kileny, 1981 ; Beattie and Boyd, 1985 ; Stapells et al, 1985 ; Hood and Berlin, 1986; Gorga et a1,1988). Some authors suggest using tone bursts in quiet, whereas others recommend mixing tone bursts with highpass, notch, or broadband noises (Picton et al, 1979 ; Stapells and Picton, 1981 ; Laukh,1983b ; Beattie and Boyd,1985 ; Stapells et a1,1985 ; Gorga and Thornton, 1989 ; Purdy et al, 1989). These noises are designed to mask those frequencies outside the center frequency of interest. If masking is not used, the response may be elicited by the side lobes of the toneburst spectra, especially if linear windows are used (Stapells et al, 1985 ; Gorga and Thornton, 1989). Additionally, at high stimulus levels the response may be elicited by the upward and, to a lesser extent, downward spread of excitation to frequencies outside the nominal test frequency. This factor is particularly problematic for low-frequency stimuli because of the asymmetry of cochlear mechanics (Wegel and Lane, 1924 ; Teas et al, 1962 ; Chung, 1981). Notch-noise masking is used to prevent both high-frequency and low-frequency areas of the cochlea outside the nominal test frequency from contributing to the response (Picton et al, 1979 ; Stapells and Picton, 1981 ; Laukli, 1983b ; Beattie and Boyd, 1985 ; Stapells et al, 1985 ; Perez-Abalo et al, 1988 ; Purdy et al, 1989 ; Beattie and Spence, 1991). Notch-noise masking is limited, however, by the relatively complex instrumentation and by the substantial upward spread-of-masking at moderate and high masking levels (Picton et al, 1979 ; Chung, 1981 ; Beattie and Spence, 1991). Highpass masking noise also has been used to increase frequency specificity (Don and Eggermont, 1978 ; Kileny, 1981 ; Laukli, 1983b ; Stapells et al, 1985). Highpass noise may provide larger amplitudes than notched noise and, thus, greater response identifiability due to the low-frequency energy on the skirts of the tone burst. Moreover, highpass noise does not contain low-frequency masking noise that may spread into the notch and reduce the signal-tonoise ratio and, consequently, response amplitude and identifiability. Highpass noise also is advantageous because it requires less complex instrumentation than notch noise. However, tone bursts in highpass noise are not as frequency specific as tone bursts in notch noise because the stimulus includes all frequencies below the cutoff frequency. A third procedure to obtain frequency-specific responses is to mix tone bursts with broadband noise. Broadband noise isolates the response to a particular region of the cochlea corresponding to the nominal frequency of the tone burst. Both the "line busy" and "suppression" hypotheses have been suggested to explain the physiologic mechanisms underlying masking (Pickles, 1982 ; Stapells et al, 1985). Without broadband noise, tone bursts activate wider areas of the cochlea and the response is derived in part from regions on the basilar membrane outside the nominal test frequency (Picton et al, 1979 ; Beattie and Boyd, 1985 ; Stapells et al, 1985 ; Perez-Abalo et al, 1988). Beattie and Boyd (1985) found no wave V latency differences in the presence of broadband noise and notch noise, and suggested that either procedure may yield more frequency-specific responses than when tone bursts are presented alone. Also, no differences were reported in the number of identifiable responses when tone bursts were presented in broadband noise or notch noise. That is, the additional stimulus energy within the notch did not yield lower thresholds than when the notch was filled with masking noise. These authors concluded that broadband noise is preferable to notch noise because the latter requires more complex instrumentation. However, Picton et al (1979) observed smaller amplitudes with broadband noise than with notch noise. Therefore, a possible disadvantage of broadband noise is reduced response amplitudes (Burkard and Hecox,1983) because of the elimination of acoustic energy in the notch (notch noise) or of frequencies below the high-frequency cutoff (highpass noise). Limited research has compared the effects of tone bursts in quiet, notch noise, highpass noise, and broadband noise on ABRs. Moreover, because of the upward and downward spreadof-masking, the relative effectiveness of the various tone burst-in-noise procedures may be 350

3 ABR to Tone Bursts in NoiseBeattie and Kennedy dependent on intensity (Teas et a1,1962 ; Chung, 1981). Therefore, the following research question was addressed : Are there differences in detectability, amplitude, or latency of wave V when normal listeners are presented with tone bursts (1000, 2000, and 4000 Hz ; 1 msec riseplateau-fall times) in quiet, notch noise, highpass noise, and broadband noise at 80 db and 40 db nhl? Our goal was to assess the relative value of the four types of stimuli to ascertain which stimulus provides the best compromise between frequency specificity and identifiability at both low and high intensities. Subjects METHOD Fifteen normal hearing subjects participated in the study. This group was composed of women ranging in age from 20 to 28 years (mean = 23 years) who reported no history of otoneurologic pathologies. They passed a 20 db HL (ANSI, 1989) screening at the octave frequencies from 250 Hz to 8000 Hz and had tympanometric peaks within ±50 dapa. The right ear was selected for testing. Instrumentation and Calibration The test stimuli were tone bursts with linear rise and fall times of 1 msec and a 1 msec plateau. Therefore, the rise-fall times allowed one complete cycle for 1000 Hz, two cycles for 2000 Hz, and four cycles for 4000 Hz. The tone bursts were produced by a generator (RC Electronics, Model 200BX), directed to an audiometer (Grason-Stadler, Model GSI 16), and then to an earphone (Telephonics, Model TDH-50P) encased in a cushion (Telephonics, Model 51). The frequency response ofthe earphone is shown in Figure 1. Tone bursts with rarefaction polarity were presented at a repetition rate of 25.6 per second FREQUENCY IN KHz Figurel Frequency response forthetelephonicstdh- 50P earphone. Gold-plated electrodes (Grass E5GH) were used to monitor the electric activity that occurred in response to auditory stimulation. These electrodes were connected to a physiologic amplifier (Grass, Model P511K), which provided a gain of 500,000. The responses were shaped by a filter with a bandpass from 30 Hz to 3000 Hz and having rejection rates of 6 db/ octave. The ABR was directed from the physiologic amplifier to the signal averager (RC Electronics). The dwell time was 40 microseconds and the analysis time was 20 msec. The artifact reject was adjusted so that approximately 10 to 20 percent of the trials were discarded (Hyde, 1985). Acoustic waveforms were obtained by directing the output of the audiometer to the earphone, which was situated on a 6-cc coupler (Quest, Model EC-9A) with an associated microphone (Quest, Model 7023) and sound level meter (Quest, Model 155). The output of the sound level meter was then directed to the signal averager (RC Electronics). The acoustic waveforms revealed linear-shaped envelopes that conformed closely to their nominal durations ; there was approximately 0.5 msec of after-ringing. The acoustic spectra of the tone bursts were obtained by directing the output of the sound level meter to the signal averager. The Power Spectrum Analysis Program (RC Electronics) was used to perform a fast Fourier transform on the tone bursts (resolution = 48 Hz). The results are shown in Figure 2 as A (1000 Hz), B (2000 Hz), and C (4000 Hz). Examination of this figure reveals that although the prominent peaks of energy were present at the nominal center frequencies, substantial acoustic energy was present at both the lower and upper frequency regions. The tone bursts were calibrated both psychoacoustically and acoustically. Psychoacoustic calibration was obtained by testing 10 normal hearing females with 1000, 2000, and 4000 Hz tone bursts having 1 msec rise-plateau-fall times. Thresholds were obtained using a two-alternative forced-choice method in which the subjects chose which of two intervals contained the stimulus (Penner,1978 ; Marshall and Jesteadt, 1986). The intervals were 1 to 2 seconds and the stimulus rate was 25.6 per second. Two db steps were used and threshold was defined as the 75 percent correct point. Thresholds were averaged across subjects and the means were specified as 0 db nhl. The intensity of these mean thresholds was deter- 351

4 _,.~-C.. Journal of the American Academy of AudiologyA7olume 3, Number 5, September m Z -10 r -zo Z w z -30 > -40 a m 0-10 Z -20 w z -30 > -40 a w lioc6 T 1048 i FREQUENCYIN Hz FREQUENCYIN Hz A the associated programmable mainframe (Krohn-Hite, Model 3905). The outputs from the lowpass and highpass filters were connected to a custom mixer and the resultant noise was directed to an audiometer (Grason-Stadler, GSI 16) where the noise was combined with the tone bursts. The frequency responses of the filters were obtained by directing the output of the tone generator (Bruel & Kjaer, Model 1049) to the filters, through the audiometer, and then to the earphone situated on a 6-cc coupler with an associated sound level meter and one-third octave filter (Quest, Models 155 and OB-133). Figure 3 shows that the bandwidths centered around 1000 Hz ( Hz), 2000 Hz ( Hz), and 4000 Hz ( Hz) had notch depths of approximately 85 db. Rejection rates exceeded 175 db per octave. The same highpass filter frequency settings that were used for the notch noise conditions also were used for the highpass noise conditions (1600 Hz, 2800 Hz, and 5600 Hz). The noise was adjusted so that it was approximately 15 db below the level that would mask the tone bursts. That is, effective masking levels of 25 and 65 db nhl were used with the 40 and 80 db nhl tone bursts, respectively. Effective masking levels were ascertained by obtaining masked thresholds in broadband noise from 10 normal-hearing subjects. With the broadband noise held constant at 50 and 90 db SPL, the tone bursts were adjusted initially in 5- to 10-dB steps until the vicinity of threshold was identified. Next, the search for the masked Figure 2 Acoustic spectra for 1000 Hz (A), 2000 Hz (B), and 4000 Hz (C) tone bursts. The dotted lines in each panel represent the notch noises. 0 mined acoustically by placing the earphone on the 6-cc coupler that was attached to the sound level meter (Quest, Model 155). Peak sound pressure levels corresponding to 0 db nhl were 18 db for 1000 Hz and 4000 Hz, and 16 db for 2000 Hz. The highpass, broadband, and notch noises were produced by directing white noise from the noise generator (Grass, Model S10) to two cascaded highpass filters (Krohn-Hite, Models 31) and two cascaded lowpass filters (Krohn-Hite, Models 30). Two filters of each type were used to increase rejection rates from a nominal 115 db/octave to 230 db/octave. The highpass and lowpass cutoff frequencies were selected with m -20? 140 r z -60 w? FREQUENCY IN Hz Figure 3 Frequency responses of the notch filters centered around 1000 Hz, 2000 Hz, and 4000 Hz. 352 I "' 1 I I''I I~II'llir ~cpi

5 :1 1 1 :11!M1 ' ABR to Tone Bursts in Noise/Beattie and Kennedy threshold began by varying the tone bursts in 2- db steps over a range of about 16 db. The twoalternative forced-choice procedure previously described was used and threshold was defined as the level at which a 75 percent correct score was achieved on a total of 10 trials at each level. Thresholds were averaged across subjects to obtain effective masking levels. When testing at 80 db nhl, the broadband noise SPLs (65 db nhl effective masking) were 87 db at 1000 Hz, 88 db at 2000 Hz, and 94 db at 4000 Hz. The noise levels were 40 db less when testing at 40 db nhl. Procedure The electrode connected to the noninverting preamplifier input was placed on the vertex, the electrode connected to the inverting input of the amplifier was positioned on the neck ipsilateral to the test ear, and the electrode connected to the common input was situated on the neck contralateral to the test ear (Beattie et al, 1986). The neck electrodes were placed 7 cm below the lower edge of the mastoid. Impedances were less than 5000 ohms and within 1000 ohms of each other. The subjects were situated in a supine position on a cot located in a sound-treated suite. The room was darkened and the subjects were asked to lie as still as possible. The subjects were tested with tone bursts in quiet, notch noise, highpass noise, and broadband noise. Responses were obtained with tone bursts presented at a high intensity (80 db nhl) and a low intensity (40 db nhl). Testing was conducted during three sessions, with one frequency tested per session. The test frequencies and noise conditions were randomized to guard against order effects. Responses were considered identifiable if the waves were judged present by two trained examiners in at least two of three trials. Wave V latency was identified as the point just before the rapid negative inflection, or at the midpoint ofthe shoulder when a clear inflection point was not identifiable. Amplitude measurements were made from the point corresponding to the latency of wave V to the succeeding trough or plateau. The number of stimulus repetitions for each tracing ranged from 3000 to RESULTS Response Detectability One purpose of this study was to ascertain the frequency of occurrence of wave V when tone bursts were presented to normal hearing listeners in quiet, notch noise, highpass noise, and broadband noise. The results revealed that wave V was observed in all subjects at 40 db and 80 db nhl for the quiet and noise conditions. Latency A second purpose of this study was-to obtain wave V latencies for all frequencies and bandwidths. A two-way analysis-of-variance (ANOVA) for repeated measures was performed for both the 40 and 80 db nhl conditions. Figure 4 illustrates mean latencies in msec and standard deviations for all three frequencies and bandwidths for 40 db nhl. Several observations are evident from this figure. First, standard deviations decreased as frequency increased from 1000 Hz (-0.66 msec) to 4000 Hz ( msec). Second, the broadband condition yielded longer latencies than the other three conditions. For example, at 1000 Hz the mean latency for the broadband noise was msec while latencies for the quiet, notch noise, and highpass noise conditions were approximately 10.2 msec. Third, latency increased as frequency decreased. For example, latencies for the tone bursts in quiet were 7.6 msec at 4000 Hz and 10.2 msec at 1000 Hz. Fourth, the quiet, notch noise, and highpass noise conditions yielded similar latencies at each frequency ; latencies were approximately 7.7, 8.7, and 10.2 msec at 4000, 2000, and 1000 Hz, respectively. Illustrative examples of the waveforms are shown in Figure 5 for the 40 db nhl stimuli. Each panel displays two tracings for each frequency and bandwidth. 12 SD -1,~ M {I ~a 10 " I ~- o - I,I S B C 6i 1 Mean - O NN HN BN 0 NN HN BN 0 NN HN BN 1000 HZ 2000 HZ 4000 HZ Figure 4 Means and standard deviations in milliseconds are shown for wave V latencies in response to 40 db nhl tone bursts centered around 1000 Hz, 2000 Hz, and 4000 Hz. The stimuli were presented in quiet (Q), notch noise (NN), highpass noise (HN), and broadband noise (BN). 353

6 Journal of the American Academy of Audiology/Volume 3, Number 5, September TIME IN MILLISECONDS Figure 5 Representative auditory brainstem responses (2 traces each) to 1000 Hz, 2000 Hz, and 4000 Hz tone bursts in quiet (Q), notch noise (NN), highpass noise (HN), and broadband noise (13N) at 40 db nhl. For the 40 db nhl stimuli, the ANOVA revealed no interaction between frequency and bandwidth (F [6,84] = 0.167, p >.05), statistically significant differences among frequencies (F [2,28] = 203.9, p <.01), and significant differences among bandwidths (F [3,42] = 55.3, p <.01). Tukey's post-hoc test was employed to identify statistically significant differences between means (Bruning and Kintz, 1987). This test revealed that wave V latency for 1000 Hz was significantly longer than those for 2000 Hz and 4000 Hz, and that latency for 2000 Hz was significantly longer than that for 4000 Hz (p <.01). Moreover, the broadband noise condition yielded a longer latency than the other bandwidth conditions (p <.01). No other differences between frequencies or bandwidths were statistically significant. Means and standard deviations for the 80 db nhl data are displayed in Figure 6. The following observations can be made from this figure. First, standard deviations decreased from about 0.6 msec at 1000 Hz to about 0.2 msec at 4000 Hz. Second, the quiet condition yielded shorter latencies than the other bandwidth conditions. At 1000 Hz the mean latency for the quiet condition was 6.86 msec while latencies for the notch, highpass, and broadband noises were approximately 8.3 msec. Third, latency increased as frequency decreased from 4000 Hz to 1000 Hz. Fourth, the notch, highpass, and broadband noise conditions yielded similar latencies at each frequency. Latencies were approximately 8.5 msec at 1000 Hz, 7.5 msec at 2000 Hz, and 6.9 msec at 4000 Hz. Illustrative examples of the waveforms are shown in Figure 7 for the 80 db nhl stimuli. Note that the first 3 milliseconds of the tracings are not shown because the electromagnetic stimulus artifact obscured the waveform during this time period. The ANOVA for the 80 db nhl stimuli revealed a statistically significant interaction (F [6,84] = 7.09, p <.01), frequency effect (F [2,28] = 62.78, p <.01), and bandwidth effect (F [3,42] = , p <.01). Tukey's post-hoc test revealed that latency for the quiet condition was significantly shorter (p <.01) than the notch, highpass, and broadband noise conditions for 1000 Hz and 2000 Hz, and the quiet condition was significantly shorter than the broadband condition for 4000 Hz. Latencies for 4000 Hz were shorter than 1000 Hz for the quiet and noise stimuli (p <.01). Latencies for 2000 Hz were shorter than 1000 Hz for all noise conditions (p <.01). However, latencies for the O NN HN BN O NN HN BN O NN HN BN 1000 HZ 2000 HZ 4000 HZ Figure 6 Means and standard deviations in milliseconds are shown for wave V latencies in response to 80 db nhl tone bursts centered around 1000 Hz, 2000 Hz, and 4000 Hz. The stimuli were presented in quiet (Q), notch noise (NN), highpass noise (HN), and broadband noise (BN). 354 le~l ~ I 'I III

7 IIRAI'JI11 1i',!1, 11111x11 ABR to Tone Bursts in Noise/Beattie and Kennedy similar for all conditions at 40 db nhl (--2.5 msec), and for the noise conditions at 80 db nhl (--1.5 msec). In contrast, the 80 db nhl quiet condition yielded a 1000 Hz to 4000 Hz latency difference of only 0.58 msec. Amplitude TIME IN MILLISECONDS Figure 7 Representative auditory brainstem responses (2 traces each) to 1000 Hz, 2000 Hz, and 4000 Hz tone bursts in quiet (Q), notch noise (NN), highpass noise (HN), and broadband noise (BN) at 80 db nhl. A third purpose of this study was to obtain wave V amplitudes for all frequencies and bandwidths. The 40 db nhl data are presented in Figure 8, which shows standard deviations and mean amplitudes in nanovolts (nv) as a function of frequency and bandwidth. The figure shows that the broadband condition yielded smaller amplitudes than the other three conditions at 2000 Hz and 4000 Hz. Figure 8 also reveals that the quiet, notch noise, and highpass noise conditions yielded similar amplitudes across frequency (-120 nv). For the 40 db nhl stimuli, the ANOVA revealed a statistically significant interaction (F [6,841= 3.27, p <.01) and bandwidth effect (F [3,42] = 11.08, p <.01). No significant effect for frequency was found (F [2,28] = 0.16, p >.05). Tukey's post-hoc test revealed that amplitude for the broadband noise was significantly smaller than the quiet, notch noise, and highpass noise conditions at 4000 Hz, and that amplitude for the broadband noise was significantly smaller than the notch noise at 2000 Hz (p <.01). No other differences among frequencies or bandwidths were statistically significant. Means and standard deviations for the 80 db nhl amplitude data are illustrated in Figure 9. The figure shows similar standard devia- quiet condition between 2000 Hz and 1000 Hz were not statistically significant (p >.01). Comparison of the 40 db nhl stimuli with the 80 db stimuli reveals three items of interest. First, the 40 db nhl stimuli had longer latencies than the 80 db nhl tone bursts, and these differences decreased as frequency increased. For example, the notch noise condition yielded latency differences (40 db versus 80 db) of 1.83 msec at 1000 Hz, 1.29 msec at 2000 Hz, and 0.85 msec at 4000 Hz. Second, latency differences between 1000 Hz and 4000 Hz were larger for the 40 db nhl stimuli. For example, the notch noise stimuli yielded a 1000 Hz to 4000 Hz latency difference of 2.45 msec at 40 db nhl and 1.47 msec at 80 db nhl. Third, latency differences between 1000 Hz and 4000 Hz were = so Mean O NN HN BN Q NN HN BN O NN HN BN 1000 HZ 2000 HZ 4000 HZ Figure 8 Means and standard deviations in nanovolts are shown for wave V amplitudes in response to 40 db nhl tone bursts centered around 1000 Hz, 2000 Hz, and 4000 Hz. The stimuli were presented in quiet (Q), notch noise (NN), highpass noise (HN), and broadband noise (BN). 355

8 Journal of the American Academy of Audiology/Volume 3, Number 5, September 1992 N 350 V T 200 S 150 M SD 50 JIM.Mean 80 db nhli O NN HN BN O NN HN BN O NN HN BN 1000 HZ 2000 HZ 4000 HZ Figure 9 Means and standard deviations in nanovolts are shown for wave V amplitudes in response to 80 db nhl tone bursts centered around 1000 Hz, 2000 Hz, and 4000 Hz. The stimuli were presented in quiet (Q), notch noise (NN), highpass noise (HN), and broadband noise (BN). subjects at both 40 and 80 db nhl in all conditions. These findings are consistent with previous investigators who used procedures similar to those employed in the present study (Kileny, 1981 ; Stapells and Picton, 1981 ; Hayes and Jerger, 1982 ; Gorga et al, 1988). The present results demonstrate that response detectability for wave V was not affected by stimulus conditions at either 40 or 80 db nhl. The broadband noise conditions yielded smaller amplitudes, which made detectability more difficult, but responses in the presence of broadband noise were observed in every subject. However, we did not compare identifiability at threshold. Testing at levels less than 40 db nhl may reveal differences among stimuli. Latency tions across frequency. Figure 9 also shows that the broadband noise yielded smaller amplitudes than the other conditions for all frequencies. For example, at 1000 Hz the mean amplitude was 133 nv for the broadband condition while amplitudes for the quiet, notch noise, and highpass noise conditions were approximately 313 nv. Finally, the figure reveals similar wave V amplitudes across frequency. The ANOVA for the 80 db nhl stimuli revealed a significant interaction (F [6,841 = 7.26, p <.01) and bandwidth effect (F [3,421 = 43.80, p <.01). There are no statistically significant effects for frequency (F [2,281 = 0.08, p >.05). Tukey's post-hoc test revealed that amplitude for the broadband noise was smaller than the other conditions at 1000 Hz, smaller than the quiet and highpass noise at 2000 Hz, and smaller than the highpass condition at 4000 Hz (p <.01). Furthermore, at 2000 Hz, amplitude for the quiet condition was larger than the three noise conditions (p <.01). No other differences among frequency or bandwidth were statistically significant. DISCUSSION Response Detectability One purpose of this study was to compare response detectability for wave V to 1000, 2000, and 4000 Hz tone bursts in quiet, notch noise, broadband noise, and highpass noise. The data revealed that response detectability for wave V did not vary for the quiet and noise conditions. We found that wave V was present in all 15 Latencies decreased when intensity was increased from 40 to 80 db nhl, and this latency change was greater for the lower frequencies. These findings are consistent with previous research (Kileny, 1981 ; Stapells and Picton, 1981 ; Hayes and Jerger, 1982 ; Beattie and Boyd,1985 ; Davis et a],1985 ; Stapells et al, 1985 ; Gorga et a1,1988 ; Perez-Abalo et a1,1988). Several hypotheses have been advanced to explain this phenomenon. Teas et al (1962) stated that latency is determined by the position at which the cochlear traveling wave first rises above threshold, but that as intensity is increased, this position shifts toward the base with a consequent decrease in travel time. The greater latency shift for the low frequencies with increasing intensity may be explained by an associated greater spread of low-frequency excitation toward the base of the cochlea. Davis (1976) commented that synaptic transmission time is decreased as intensity increases with a resultant decrease in latency. Brinkman and Scherg(1979)suggested that latency isdependent on that point in time at which the traveling wave first rises above threshold. Likewise, Gorga et al (1988) suggested that a certain amplitude may trigger a neural response and that this threshold amplitude will be reached earlier for higher levels of stimulation. Finally, Kileny (1981) suggested that higher intensities are associated with an increase in the traveling wave velocity, which decreases latency because less time is required to reach the place of stimulation on the basilar membrane. The present study also showed that latency decreased as the frequency increased from 1000 to 4000 Hz. These results are consistent with 356

9 ABR to Tone Bursts in Noise/Beattie and Kennedy previous authors who have used tone bursts in quiet and/or noise conditions (Suzuki et al, 1977 ; Kileny, 1981 ; Hayes and Jerger, 1982 ; Beattie and Boyd,1985 ; Davis et a1,1985 ; Gorga et al, 1988 ; Purdy et al, 1989). This increase in latency reflects the longer travel time required by low-frequency stimuli to progress to more apical regions ofthe cochlea (Stapells and Picton, 1981). That is, the place of maximum excitation shifts to the apex of the cochlea as the frequency is decreased, thus creating an increase in latency. Also, it is of interest to note that travelingwave velocities are more rapid at the basal end of the cochlea (Eggermont and Odenthal, 1974 ; Parker and Thornton, 1978). When noise is added to tone bursts to mask the basal areas, the response is derived from the region of the cochlea that corresponds more closely to the nominal frequency of the tone burst. Slightly longer latencies were observed for broadband noise at 40 db nhl as compared to the quiet, notch noise, and highpass noise conditions. Although this result was somewhat unexpected (Picton et al, 1979 ; Beattie and Boyd, 1985), at least three explanations may be offered. First, the shorter latencies for the tone burst in quiet may be due to the additional acoustic energy present in the high-frequency and low-frequency side-lobes. Second, the quiet, highpass noise, and notch noise conditions all contained acoustic energy in the notch region that was not present in the broadband noise. The broadband noise masks tone-burst energy in the notch, which may result in less neural activity related to the stimulus with a consequent increase in latency. Burkard and Hecox (1983) observed increased wave V latency when 60 db nhl 1000 Hz and 4000 Hz tone bursts were mixed with broadband effective noise levels of only 20 db to 30 db. These shifts increased to about 2 msec when the effective masking levels increased to 50 db. The tendency for broadband noise to increase the latency ofabrs has been verified in subsequent studies (Burkard and Hecox, 1987 ; Hecox et al, 1989). Burkard and Hecox (1987) concluded that their findings cannot be explained adequately by a shift in place along the basilar membrane. Instead, they hypothesized that the latency shifts occur in the central auditory mechanism, perhaps due to an increase in the time required for synaptic or postsynaptic processes. Third, latency may be dependent on that point in time at which the traveling wave first rises above threshold (Brinkmann and Scherg, 1979). The broadband noise may mask the first cycle or few cycles of the stimulus with a resultant increase in latency. However, the foregoing hypotheses do not explain why latency differences among noise conditions were observed at 40 db nhl but not 80 db nhl. The quiet condition at 80 db nhl yielded shorter latencies than the noise conditions. Moreover, the 1000 Hz to 4000 Hz latency difference was only about 0.5 msec for the quiet condition but msec for the noise conditions. Also, the latency shift as intensity decreased from 80 db nhl to 40 db nhl was greater for the quiet condition than for the noise conditions. For example, for the tone bursts in quiet, latency decreased about 3.0 msec at 1000 Hz, 2.0 msec at 2000 Hz, and 1.5 msec at 4000 Hz. In contrast, the noise conditions decreased latency approximately 2.2 msec at 1000 Hz, 1.5 msec at 2000 Hz, and 1.0 msec at 4000 Hz. These results are consistent with previous research (Picton et al, 1979 ; Stapells and Picton, 1981 ; Hayes and Jerger, 1982 ; Gorga et al, 1988 ; Beattie and Spence, 1991) and suggest that the responses elicited by the tone bursts in quiet were, in part, mediated by regions on the basilar membrane that did not correspond to the center frequency of the tone burst. Instead, the shorter latencies for the quiet conditions may reflect the contribution of the high-frequency components of the tone burst spectrum and/or the spread of excitation to the higher frequencies. The basal region of the cochlea may contribute to a more synchronized neural response for the quiet condition, as contrasted to the noise conditions that mask the more basal regions of the cochlea (Picton et al, 1979 ; Stapells and Picton, 1981 ; Stapells et al, 1985 ; Beattie and Spence, 1991). To increase frequency specificity, the present study suggests that notch, highpass, or broadband noise should be mixed with highlevel tone bursts having linear rise-fall times. The use of noise conditions for low intensity levels does not appear necessary for isolating the auditory brainstem response because both the notch and the highpass conditions yielded latencies similar to the quiet condition. Amplitude Similar amplitudes were found at all frequencies. These results are consistent with Beattie and Boyd (1985) who observed a similar number of responses for 500 Hz, 1000 Hz, and 2000 Hz in quiet, notch noise, and broadband noise. In contrast, Picton et al (1979) observed ,.,ti, m

10 Journal of the American Academy of Audiology/Volume 3, Number 5, September 1992 larger wave V amplitudes in notch noise for 1000 Hz tone bursts than for 4000 Hz tone bursts. Studies that have reported amplitudes in response to clicks in notch noise and highpass noise have been quite variable. For example, Beattie and Spence (1991) found that amplitudes increased with increasing frequency whereas Don and Eggermont (1978) reported larger amplitudes at 500 Hz than at 2000 Hz. With the exception of 1000 Hz at 40 db nhl, wave V amplitudes were smaller with the broadband noise than amplitudes for the quiet, notch, and highpass conditions. Apparently masking noise desynchronizes some of those neural elements that contribute to the unmasked response, thereby attenuatingthe whole nerve potential (Stapells and Picton,1981; Davis and Owen, 1985). Our results are in agreement with Burkard and Hecox (1983) who observed reduced wave V amplitudes when 60 db nhl 1000 Hz and 4000 Hz tone bursts were mixed with effective masking levels (broadband noise) of only 20 db. They found that amplitudes decreased from approximately 250 nv in quiet to about 100 nv when the effective masking level was increased to 50 db. Our results are in contrast with those of Perez-Abalo et al (1988) and Picton et al (1979) who reported that both notch noise and broadband noise produced similar results. Moreover, Beattie and Boyd (1985) found that response detectability was similar for both broadband noise and notch noise. The differences among studies may be due to stimulus variables such as rise-plateau-fall times, filter bandwidths, and/or the intensity relationship between the tone burst and the noise. The bandwidth of the physiologic filters also may have a substantial effect on the ABR (Suzuki and Horiuchi, 1977 ; Beattie et al, 1984). Moreover, the reader should recall that all of our subjects yielded identifiable responses at 40 db nhl with the broadband noise, and we did not compare wave V thresholds using the various stimuli. Because lower amplitudes were found with the broadband noise than with the quiet, notch noise, and highpass noise, the latter conditions seem preferred. At low levels of stimulation, the tone burst in quiet may be preferred because of the apparent frequency specificity and the simplicity of instrumentation. However, at moderate and high intensities, notch and highpass conditions are preferred to the quiet condition because of the improved frequency specificity provided by the masking. The highpass noise is advantageous because it required less complex instrumentation than notch noise, and because the low-frequency spread-of-masking is absent. The notch noise may provide better frequency specificity than highpass noise. Future studies should investigate the effects of rise-fall times and signal-to-noise ratio on ABRs using highpass, notch, and broadband noises. Moreover, before our procedures can be employed clinically, research is required using tone bursts in quiet and noise with subjects having hearing losses of varying degrees, configurations, and etiologies. Some reports that have studied the electrophysiologic responses of eighth nerve fibers suggest that data from normal hearing systems may not generalize readily to hearing impaired systems (Evans, 1974 ; Liberman and Kiang, 1978 ; Kiang et al, 1986). Several investigators have reported that stimulation at moderate-to-high intensities can cause the spread of excitation to adjacent cochlear regions, and that this spread is primarily toward the basal, high-frequency fibers (Kiang et al, 1967 ; Evans, 1972 ; Kiang and Moxon, 1974 ; Liberman and Kiang, 1978). Frequency-threshold (tuning) curves for normal hearing systems show that high-frequency eighth nerve fibers have thresholds at the characteristic (best) frequency that are 40 to 60 db better than low-frequency thresholds on the tail of the curve (Kiang and Moxon, 1974). That is, the tip-to-tail ratio is approximately 50 db. Highpass noise may be used to occupy fibers with high characteristic frequencies so that they are unresponsive to low-frequency stimuli. Gorga and Worthington (1983) show that the presence of broadband noise can elevate the frequency-threshold curve for high-frequency fibers but may leave the shape of the curve unaltered. In contrast, cochlear pathology can alter the shape ofthe frequency-threshold curve in quiet and in noise so that tip-to-tail ratio may be less than 10 db (Evans, 1974 ; Dallos and Harris, 1978 ; Liberman and Kiang, 1978 ; Liberman and Dodds, 1984 ; Kiang et al, 1986). Because of the reduced threshold difference between the low and high frequencies, highpass noise may not be effective in masking highfrequency fibers when cochlear pathology is present. Therefore, the application of ABR data obtained on normal-hearing subjects may not be applicable to individuals with hearing impairment (Gorga and Worthington, 1983). Although the present investigation suggests that mixing tone bursts with notch noise or highpass noise may be feasible procedures for estimating auditory sensitivity, several other 358

11 ~uiyd~~f 11 1! ~ { I+ Plwti ABR to Tone Bursts in Noise/Beattie and Kennedy frequency-specific procedures have been suggested. These include tone bursts using Blackman windows (Gorga and Thornton,1989), early/middle responses (Beattie and Boyd,1985 ; Davis et a1,1985), and middle responses (Mendel and Wolf, 1983 ; Kraus and McGee, 1990). Comparative studies are required using hearing impaired subjects to assess the relative advantages of these procedures. REFERENCES American National Standards Institute. (1989)American National Standard Specification for Audiometers (ANSI S ). New York : ANSI. Beattie RC, Beguwala FE, Mills DM, Boyd RL. (1986) Latency and amplitude effects of electrode placement on the early auditory evoked response. JSpeech HearDisord 51 : Beattie RC, Boyd RL. (1985) Early/middle evoked potentials to tone bursts in quiet, white noise and notched noise. Audiology 24 : Beattie RC, Moretti M, Warren V. (1984) Effects of risefall time, frequency, and intensity on the early/middle evoked response. J Speech Hear Disord 49 : Beattie RC, Spence J. (1991) Auditory brainstem response to clicks in quiet, notch noise, and highpass noise. J Am Acad Audiol 2: Brinkmann RD, Scherg M. (1979) Human auditory onand off-potentials of the brainstem. Scand Audiol 8: Bruning JL, Kintz BL. (1987) Computational Handbook of Statistics. Glenview, IL: Scott, Foresman. Burkard R, Hecox K. (1983) The effect of broadband noise on the human brainstem auditory evoked response. II. Frequency specificity. JAcoust Soc Am 74 : Burkard R, Hecox KE. (1987) The effect of broadband noise on the human brain-stem auditory evoked response Anatomic locus. J Acoust Soc Am 81 : ChungDY. (1981) Tone-on-tone masking in subjects with normal hearing and with sensorineural hearing loss. J Speech Hear Res 24 : Dallos P, Harris D. (1978) Properties of auditory nerve responses in absence of outer hair cells. J Neurophysiol 41 : Davis H. (1976) Principles of electric response audiometry. Ann Otol Rhinol Laryngol 85 (Supp128) :1-96. Davis H, Hirsh SK, Turpin LL, Peacock ME. (1985) Threshold sensitivity and frequency specificity in auditory brainstem response audiometry. Audiology 24 : Davis H, Owen J. (1985) Auditory evoked potentials. In : Owen JH, Davis H, eds. Evoked Potential Testing-Clinical Applications. New York: Grime & Stratton, Debruyne F, Forrez G. (1982) On-effect in brainstem electric response audiometry-consequences for the use of tone-bursts. Otorhinolaryngol 44 : Don M, Eggermont JJ. (1978) Analysis of the click-evoked brainstem potentials in man using highpass noise masking. JAcoust Soc Am 63 : Don M, Eggermont JJ, Brackmann DE. (1979) Reconstruction of the audiogram using brainstem responses and high-pass noise masking. Ann Otol Rhinol Laryngol 88 (Suppl 57):1-20. Eggermont JJ. (1982) The inadequacy of click-evoked auditory brainstem responses in audiological applications. Ann N YAcad Sci 388: Eggermont JJ, Odenthal DW. (1974) Frequency selective masking in electrocochleography. Rev Laryngol 95 : Evans EF. (1972) The frequency response and other properties of single fibers in guinea-pig cochlear nerve. J Physiol 226: Evans EF. (1974) Auditory frequency selectivity and the cochlear nerve. In : Zwicker E, Terhardt E, eds. Facts and Models in Hearing. New York : Springer-Verlag, Fjermedal 0, Laukli E. (1989) Low-level 0.5 and 1 khz auditory brainstem responses. Scand Audiol 18 : Gorga MP, Worthington DW. (1983) Some issues relevant to the measurement of frequency-specific auditory brainstem responses. Semin Hear 4: Gorga MP, Kaminski JR, Beauchaine KA, Jesteadt W. (1988) Auditory brainstem responses to tone bursts in normally hearing subjects. J Speech Hear Res 31 : Gorga MP, Kaminski JR, Beauchaine KL. (1991) Effects of stimulus phase on the latency of the auditory brainstem response. J Am Acad Audiol 2:1-6. Gorga MP, Thornton AR. (1989) The choice of stimuli for ABR measurements. Ear Hear 10: Hall III JW. (1984) Auditory brainstem response audiometry. In : Jerger J, ed. Hearing Disorders in Adults- Current Trends. San Diego : College-Hill Press, Hayes D, Jerger J. (1982) Auditory brainstem response (ABR) to tone-pips : results in normal and hearing-impaired subjects. Scand Audiol 11 : Hecox KE, Patterson J, Birman M. (1989) Effect of broadband noise on the human brain stem auditory evoked response. Ear Hear 10 : Hood LJ, Berlin CI. (1986) Auditory Evoked Potentials. Austin: Pro-Ed. Hyde ML. (1985) Instrumentation and signal processing. In : Jacobson JT, ed. The Auditory Brainstem Response. San Diego : College-Hill Press, Jerger J, Mauldin L. (1978) Prediction of sensorineural hearing level from the brain stem evoked response. Arch Otolaryngol 104: Jerger J, Oliver T, Stach B. (1985) Auditory brainstem response testing strategies. In : Jacobson JT, ed. The Auditory Brainstem Response. San Diego : College-Hill Press, Mang NY, Liberman MC, Sewell WF, Guinan JJ. (1986) Single unit clues to cochlear mechanisms. HearRes 22 :

12 Journal of the American Academy of Audiology/Volume 3, Number 5, September 1992 Kiang NY, Maxon EC. (1974) Tails of tuning curves of auditory-nerve fibers. JAcoust Soc Am 55 : KiangNY, Sachs MB, Peake WT. (1967) Shapes oftuningcurves for single auditory nerve fibers. JAcoust Soc Am 42 : Kileny P. (1981) The frequency specificity of tone-pip evoked auditory brainstem responses. Ear Hear 2: Kraus N, McGee T. (1990) Clinical applications of the middle latency response. J Am Acad Audiol 1: Laukli E. (1983a) Stimulus waveforms used in brainstem response audiometry. Scand Audiol 12 : Laukli E. (1983b) High-pass and notch noise masking in suprathreshold brainstem response audiometry. Scand Audiol 12: Liberman MC, Kiang NY. (1978) Acoustic trauma in cats-cochlear pathology and auditory-nerve activity. Acta Otolaryngol (Suppl) 358:1-63. Liberman MC, Dodds LW. (1984) Single-neuron labelling and chronic cochlear pathology. IV. Stereocilia damage and alterations of threshold tuning curves. Hear Res 16 : Marshall L, Jesteadt W. (1986) Comparison of pure-tone audibility thresholds obtained with audiological and twointerval forced-choice procedures. J Speech Hear Res 29 : Mendel MI, Wolf KE. (1983) Clinical applications of the middle latency responses. Audiology : J Contin Ed 8: Parker DJ, Thornton ARD. (1978) Cochlear travelling wave velocities calculated from the derived components of the cochlear nerve and brainstem evoked responses of the human auditory system. Scand Audiol 7: Penner MJ. (1978) Psychophysical methods and the microcomputer. In : Mayzner MS, Dolan TR, eds. Minicomputers in Sensory and Information-Processing Research. New York : John Wiley & Sons, Perez-Abalo MC, Valdes-Soso MJ, Bobes MA, Galan L, Biscay R. (1988) Different functional properties of on and off components in auditory brainstem responses to tone bursts. Audiology 27 : Pickles JO. (1982) An Introduction to the Physiology of Hearing. New York : Academic Press. Picton TW, Ouellette J, Hamel G, Smith AD. (1979) Brainstem evoked potentials to tone-pips in notched noise. J Otolaryngol 8: Purdy SC, Houghton JM, Keith WJ, Greville KA. (1989) Frequency-specific auditory brainstem responses. Audiology 28 : Schwartz DD, Schwartz JA. (1991) Auditory evoked potentials in clinical pediatrics. In : Rintelmann WF, ed. Hearing Assessment. 2nd Ed. Austin : Pro-Ed. Stapells DR, Picton TW. (1981) Technical aspects of brainstem evoked potential audiometry using tones. Ear Hear 2: Stapells DR, Picton TW, Perez-Abalo M, Read D, Smith A. (1985) Frequency specificity in evoked potential audiometry. In : Jacobson JT, ed. The Auditory Brainstem Response. San Diego : College-Hill Press, Stein LK, Ozdamar O, Schnabel M. (1981) Auditory brainstem responses (ABR) with suspected deaf-blind children. Ear Hear 2: Suzuki T, Hirai Y, Horiuchi K. (1977) Auditory brainstem responses to pure tone stimuli. Scand Audiol 6: Suzuki T, Horiuchi K. (1977) Effect of high-pass filter on auditory brain stem responses to tone pips. Scand Audiol 6: Teas DC, Eldredge DH, Davis H. (1962) Cochlear responses to acoustic transients : an interpretation of whole nerve action potentials. J Acoust Soc Am 34 : van Zanten GA, Brocaar MP. (1984) Frequency-specific auditory brainstem responses to clicks masked by notched noise. Audiology 23 : Weber BA, Seitz MR, McCutcheon MJ. (1981) Quantifying click stimuli in auditory brainstem response audiometry. Ear Hear 2: Wegel RL, Lane CE. (1924) The auditory masking of one pure tone by another and its probable relation to the dynamics of the inner ear. Phys Reu 23 :