Hearing Research, 51 (1991) 123-138 0 1991 Elsevier Science Publishers B.V. (Biomedical
HEARES
123 Division)
0378-5955/91/$03.50
01492
Electrically
evoked auditory
brainstem response: current level
Growth
of response with
Paul J. Abbas and Carolyn J. Brown * Department
of Speech Pathology and Audiology, University of Iowa, Iowa City, Iowa, U.S.A. (Received
4 August
1989; accepted
25 July 1990)
The electrically evoked brainstem response (EABR) was measured in cochlear implant users who had received either the Ineraid multichannel implant or the Nucleus multichannel implant. Although both implants use a multi-electrode array, they are different in a number of ways. In the Ineraid system the electrodes can be accessed directly through a percutaneous plug and stimulation is generally on four different intracochlear electrodes relative to a common ground outside the cochlea. In the Nucleus implant stimulation is accomplished via an internal coil and stimulation is bipolar between pairs along the 22 electrode array. The ABR waveforms were similar for both groups of subjects, consisting of a series of 3 or 4 positive peaks at the highest levels of stimulation. Using the normal stimulation mode (bipolar for Nucleus and monopolar for Ineraid), users of both devices demonstrated an increase in response amplitude and a decrease in response latency with increases in current level. The threshold of response tended to be higher and growth of the response with level tended to be more gradual for Nucleus users than for Ineraid users. However, with bipolar stimulation for both implant types, when the stimulating electrodes were closely spaced the threshold of response was higher and the growth of amplitude with level was more gradual than the case where the electrodes were separated further. When bipolar stimulation and similar electrode spacing was used, the response growth and threshold were similar for both implant types. Results from neither device showed a strong correlation with performance on word recognition tests. Electrical
stimulation;
Co&ear
implant;
EABR
Introduction Recent experience with cochlear implants has shown that although subject performance with the implant can vary widely, some individuals can gain significant benefit from the implant as compared to other forms of rehabilitation. In dealing with prospective or new cochlear implant users, there are several important decisions that need to be made. For example, whether or not to receive an implant, which type of device to implant, and the specific settings of the particular device that is used. There are a number of audiological and psychophysical tests that are used both with preimplant electrical stimulation and more commonly with stimulation through the implant which are
Correspondence to: Paul J. Abbas, Department of Speech Pathology and Audiology, The University of Iowa, Iowa City, IA 52242, U.S.A. * Present address: Department of Speech and Hearing Science, Arizona State University, Tempe, Arizona, U.S.A.
useful in making some of these decisions. We have been investigating the electrically evoked auditory brainstem response (EABR) as a possible addition or alternative to some of these procedures. The data reported thus far with pre-implant stimulation have not been promising as a predictive measure of performance. We have also been making measurements of brainstem potentials in response to stimulation through two different types of multielectrode implants in an effort to assess whether or not the characteristics of these responses are related to a subject’s performance. Although there are some conflicting data, several animal studies have suggested that growth of response as a function of current level may be indicative of the number of surviving neurons in the auditory nerve (Smith and Simmons, 1983; Lusted et al., 1984; Hall, 1989). If one reasons that a greater number of surviving neurons would have the capacity to transmit more information to the central nervous system, then one may expect a higher level of performance with the implant in
124
subjects with relatively more surviving neurons. The measures of EABR in response to extracochlear stimulation that have been made to date, including observations in our laboratory, have shown responses which are variable in form (Meyer et al., 1984; Game et al., 1987; Black et al., 1987; Simmons et al., 1984; Gantz et al., 1988, van den Honert and Stypulkowski, 1986). In our experience with stimulation through the implant, the form of the EABR is similar across subjects, across time within a subject, and similar in form to the acoustically evoked brainstem response. These observations suggest that with promontory stimulation the source of the recorded potential may in some subjects include activity from outside of the auditory pathway. The consistency of the waveforms with stimulation through the implant suggests that in that case the response primarily reflects activity of the auditory pathways. Any predictive measure would obviously be most useful if it could be recorded preoperatively. Our rationale for using stimulation through the implant is that these measures should provide the ‘best case’ that one may expect. Any measure that correlates with performance during pre-implant testing should also correlate with performance during post-implant testing. In this paper, we describe experiments in which we have attempted to characterize the properties of the EABR as recorded from users of two different multichannel cochlear implants. In particular, we have examined the growth of the response with stimulus level for several different stimulating electrode configurations in each implant type. In addition, measures of EABR threshold and the rate of growth of the evoked potential with stimulus level are correlated with measures of word recognition ability in the same subjects.
plant ranged in age from 28 to 71 years with the time between their loss of hearing and date of implantation ranging from 1 to 34 years. Subjects with the Ineraid implant ranged in age from 33 to 71 years with the time between their loss of hearing and date of implantation ranging from 1 to 44 years. Several experimental manipulations are reported here. Not all subjects were run in all conditions. Some subjects were run as many as 4 times spanning a period of up to 30 months. Thirty of the subjects were implanted at University of Iowa Hospitals and Clinics (designated with the prefix I in subject identification number). Twelve were implanted at other centers and brought to our program for testing (designated with the prefix C in subject identification number). All subjects participated in an extensive program of audiological, psychological, and psychophysical tests as part of the Iowa Cochlear Implant Program. Stimuli Stimuli used for all of the experiments reported here were biphasic square pulses, either 100 ps per phase or 200 11s per phase. For both implant types, the speech processor was bypassed, but because of the differences between implant types, the details of stimulus delivery were quite different. In the Ineraid subjects, the stimuli were delivered directly to the electrode via the percutaneous plug. The stimuli were delivered through a custom-built, ground-isolated current generator which was capacitively coupled to the electrodes to avoid passing any DC current. We monitored the waveform delivered to the electrodes through
Methods Subjects We report data from 42 subjects, 23 of whom have previously been implanted with the Nucleus 22-electrode cochlear implant and 19 of whom have been implanted with the Ineraid Cochlear implant. Each subject had at least one month of experience with the implant before the measurements were made. Subjects with the Nucleus im-
I
I
0
0.5
1 .o
1.5
2.0
MS
Fig. 1. Typical current stimulation waveforms for the Ineraid implant. The current waveform in the Ineraid device was measured through a resistor in series with stimulating electrodes.
125
an isolated amplifier built into the current stimulator current. A typical waveform recorded from a subject is shown in Fig. 1. Waveforms were similar across the range of current levels used. The current to each electrode was measured on-line for each stimulus used in the experiment using this method. Current pulses were delivered with alternating initial phase, i.e., the presentation of a positive-negative pulse was followed by a negative-positive pulse. For the Nucleus subjects, the stimuli were delivered through the Speech Processor Interface. The specific characteristics of the Nucleus current source are described eleswhere (Crosby et al., 1985). The stimulus waveform is similar to that used in the Ineraid device, the primary difference being a short interval (in our experiments, 20 /.Ls) between the positive and negative pulses forming the biphasic waveform. A computer program was written to stimulate a given electrode pair with a train of biphasic pulses and simultaneously to trigger a separate computer that was used for response averaging. Stimuli were biphasic but not alternating in phase. Current could not be monitored on-line; the actual current values delivered were calculated subsequently from the individual current calibration data supplied by the manufacturer.
1 PV
1
!
0
2
6
4
a
10
MS
Fig. 2. Typical averaged and filtered EABR subjects with the Ineraid cochlear implant. time calibration is shown. For each subject current stimulus was chosen in the upper part dynamic range.
traces from five The voltage and the level of the of that subject’s
Procedure Recording
Recording electrodes were placed on the vertex (positive), contralateral mastoid (reference), and forehead (ground). A ground-isolated, low-noise amplifier with gain of 500,000 (bandwidth 0.1 Hz to 1.5 kHz, 6 dB/octave cutoff) preceded the input to the averaging computer. Special care was taken to shield the amplifier from RF to reduce stimulus artifact generated by the high frequency carrier signal used in the Nucleus system. Two averages of 1000 or 2000 sweeps were used in each stimulus condition. Response traces with voltage exceeding 20 PV were automatically rejected by the program. The number of rejected sweeps was typically less than 200 for each averaged trace. the stimulus artifact was After averaging, eliminated from the beginning of each trace and the signal was then digitally filtered with an FIR bandpass filter (57 order) with cutoff values 150 Hz to 3 kHz.
The typical experimental session lasted 3-4 hours. Subjects were seated in a reclining chair and asked to relax and/or sleep during the session. Some subjects received 5-10 mg of Valium to help them to sleep. Before any measurements were made, the subjects were asked to indicate behavioral threshold and uncomfortable level for each stimulus to be used in the session. The current levels of all stimuli were subsequently kept below these uncomfortable levels. The specific conditions that were completed with each subject varied because of time limitations. Results The EABR waveforms recorded from different subjects, different implants, and different electrodes within an implant were all consistent in their morphology. Examples of responses recorded from users of the two implants are shown in Fig. 2
126
(Ineraid) and Fig. 3 (Nucleus). For all of the traces in Figs. 2 and 3, the current level chosen was in the upper part of the subjects’ dynamic range. Two replications of the EABR are shown superimposed for each subject. Typically, the most prominent peak occurs about 4 ms after stimulus onset. Two earlier peaks occurring at approximately 1.3 and 2.0 ms were often observed at higher current levels. In a very few traces, a fourth peak is observed at a latency of approximately 3 ms. All of the analysis of amplitude and latency that will be discussed in the rest of the paper involves the measurement of the amplitude and latency of the latest and most prominent peak (positive to negative) of the EABR which most likely corresponds to wave V of the acoustically evoked ABR. EABRs were recorded from each subject using a set of standard stimuli and electrode configura-
CNlO
IN1 1
CN4
CN13
1 /I”
Fig. 3. Typical averaged and filtered EABR traces from subjects with the Nucleus cochlear implant. The voltage time calibration is shown. For each subject, the level of current stimulus was chosen to be in the upper part of subject’s dynamic range.
five and the that
tions. Since the implants’ design and normal stimulation modes were quite different, these standard stimulus parameters were chosen to be more like the normal mode of stimulation rather than attempting to be the same for both implant types. For the Ineraid implant, these stimuli consisted of monopolar stimulation with a pulse duration of 100 pss/phase. In the Nucleus implant, the stimulation was bipolar and a 200 ps/phase pulse duration was used. It was necessary to use a longer stimulus duration with Nucleus subjects because with bipolar stimulation, the response is generally less sensitive than with monopolar stimulation. The use of longer duration stimuli generally resulted in lower current thresholds. Fig. 4 illustrates measurements of the amplitude and latency of the fourth peak of the EABR recorded as a function of the current level of the bipolar pulse for both a Ineraid and Nucleus cochlear implant user. In each case, the parameter is the stimulating electrode pair. In the Ineraid implant, there are 8 electrodes: electrode 1 is the most apical and electrode 8 is placed in the temporalis muscle and is usually used as a common reference. The electrodes placed within the cochlea are spaced 4 mm apart. Stimulation in these cases is accomplished from each electrode in the cochlea to a common ground. The amplitude growth functions in Fig. 4A show data for four different electrode pairs as indicated in the legend. These functions generally increase monotonically with level and typically do not saturate over the range of levels used. This particular subject shows a clear difference in sensitivity among the 4 electrodes, electrode pair l-8 being the most sensitive and electrode pair 4-8 being the least. Many of the Ineraid subjects show a similar pattern; others show little difference in sensitivity across electrodes. Fig. 4B shows similar data from a Nucleus implant subject. For the Nucleus implant, the electrodes are numbered 1 to 22 with 22 being the most apical. The spacing of adjacent electrodes is 0.75 mm. The stimulation in these cases is bipolar +4, meaning that if electrode 1 is chosen, the stimulation is between electrodes 1 and 6. In each case, the stimulated electrode pair is indicated in the legend. In all cases, the stimulation with the Nucleus implant is with bipolar electrode pairs.
127 Subject
IS9 (Ineraid)
Sub[ect
IN8 (Nucleus)
0.5 -
o-o
l-8
O-0
2-8
A-A
3-8
A-A
4-8
B
0.4 --
s
o-o
l-5
O-0
5-10
A-A
10-15
A-A
15-20
0.3--
i I.4
4
5.5 -
. I
O-O
l-8
0-a
2-8
A-A
3-8
A-A
4-8
D
o-o O-0
5.2 --
.
1-5 5-10 10-15 15-20
A-A
A-A 9
4.9 --
z :
4.6 --
4 4.3 -0
0.1
0.3
0.2
Current
(mA)
7 I3.4
4.0 7 0.0
0.1
0.2
Current
0.3
0.4
(mA)
Fig. 4. Amplitude (positive peak to following negative trough) and latency (positive peak) of the fourth peak in the EABR response is plotted as a function of the stimulus current level for one subject with the Ineraid implant and one subject with the Nucleus implant. The stimulated electrode pair is the parameter in each graph and is indicated in the legend.
The sensitivity of each electrode pair can be quite different as seen with electrode lo-15 in this subject, which has a higher threshold than the other stimulated pairs. Nevertheless, we have observed no consistent pattern across subjects in which a particular electrode pair is most sensitive. The latency of the measured responses are shown for each of the two subjects in Figs. 3C and 3D. The latency generally decreases slightly with increasing current level, but the magnitude of the latency changes observed are much smaller than corresponding changes in latency observed with acoustic stimulation. In some cases this trend was not evident, but over 95% of the measured latency functions had a negative slope when fit with linear regression. The measures of amplitude growth are sum-
marized in the bar graphs in Fig. 5. The individual amplitude functions were fit with a linear function and the slope of that function and threshold (defined to be the current level necessary to elicit a response amplitude of 0.1 CLV) was determined. For subjects where we had made repeated measures for different electrodes, the functions for each repetition were fit and the values of threshold and slope averaged. Data for the Nucleus and Ineraid implants are reported separately in the figure. Fig. 5A and 5C show average data for Ineraid subjects. Both the threshold and slope data show a consistent change across electrode configuration. Stimulation of basal electrodes results in higher thresholds and shallower growth functions than is found with stimulation of apical electrodes. To test these differences, we performed
128 Inerold 0.30, A 0.25
-5 0.20 3 0 0.15 f $
0.10 0.05
2. 0.00
L
0.25
Nucleus
herald 20.0
0.0
4.0 C
I--
1-8
2-0
3-a
D
4-0
Electrode polr
Electrode polr
Fig. 5. Threshold of EABR response was averaged across subjects with each implant type and is plotted in (A) for Ineraid users and in (B) for Nucleus users. Slope of the growth function was calculated using linear regression. These values were averaged across subjects and are plotted in (C) for lneraid users and in (D) for Nucleus users.
a multiple test analysis of variance where stimulating electrode was the within subject factor. For the group of Ineraid subjects, the analysis for both threshold (F = 3.73, (Y= 0.021) and slope (F = 6.06, (Y= 0.002) showed significant differences across electrodes. Follow-up paired samples t-tests showed that electrode pairs 1-8 to 4-8 were significantly different (0.05 level) in threshold and electrode pairs 1-8 to 3-8 and l-8 to 4-8 were significantly different in slope. Similar tests with Nucleus subjects showed no significant differences for either threshold or slope measures. The plots in Fig. 5 show slope calculated in linear current units. In doing this, the relative size of the stimulus increment relative to threshold is not taken into account. By calculating the slope in logarithmic units (in dB), the differences between the threshold values for the different electrodes
should not affect the measure of growth. When the calculations were done in this way, the differences among electrodes tended to be smaller, but the statistical analysis still showed significant differences among electrodes for the Ineraid implant with follow-up t-tests showing differences between electrode pairs 1-8 to 3-8 and 2-8 to 3-8. The average latency data are illustrated in Fig. 6. To compare values of absolute latency, the latency at the highest current level used was measured for each subject and stimulus condition. These values are shown in the upper two graphs. In addition, each latency vs. current level function was fit by linear regression. The slope of that function is shown in the lower two graphs of Fig. 6. For both the Nucleus and Ineraid subjects,
129 herald
Inerald -10.0
C
2
HEARES
123 Division)
0378-5955/91/$03.50
01492
Electrically
evoked auditory
brainstem response: current level
Growth
of response with
Paul J. Abbas and Carolyn J. Brown * Department
of Speech Pathology and Audiology, University of Iowa, Iowa City, Iowa, U.S.A. (Received
4 August
1989; accepted
25 July 1990)
The electrically evoked brainstem response (EABR) was measured in cochlear implant users who had received either the Ineraid multichannel implant or the Nucleus multichannel implant. Although both implants use a multi-electrode array, they are different in a number of ways. In the Ineraid system the electrodes can be accessed directly through a percutaneous plug and stimulation is generally on four different intracochlear electrodes relative to a common ground outside the cochlea. In the Nucleus implant stimulation is accomplished via an internal coil and stimulation is bipolar between pairs along the 22 electrode array. The ABR waveforms were similar for both groups of subjects, consisting of a series of 3 or 4 positive peaks at the highest levels of stimulation. Using the normal stimulation mode (bipolar for Nucleus and monopolar for Ineraid), users of both devices demonstrated an increase in response amplitude and a decrease in response latency with increases in current level. The threshold of response tended to be higher and growth of the response with level tended to be more gradual for Nucleus users than for Ineraid users. However, with bipolar stimulation for both implant types, when the stimulating electrodes were closely spaced the threshold of response was higher and the growth of amplitude with level was more gradual than the case where the electrodes were separated further. When bipolar stimulation and similar electrode spacing was used, the response growth and threshold were similar for both implant types. Results from neither device showed a strong correlation with performance on word recognition tests. Electrical
stimulation;
Co&ear
implant;
EABR
Introduction Recent experience with cochlear implants has shown that although subject performance with the implant can vary widely, some individuals can gain significant benefit from the implant as compared to other forms of rehabilitation. In dealing with prospective or new cochlear implant users, there are several important decisions that need to be made. For example, whether or not to receive an implant, which type of device to implant, and the specific settings of the particular device that is used. There are a number of audiological and psychophysical tests that are used both with preimplant electrical stimulation and more commonly with stimulation through the implant which are
Correspondence to: Paul J. Abbas, Department of Speech Pathology and Audiology, The University of Iowa, Iowa City, IA 52242, U.S.A. * Present address: Department of Speech and Hearing Science, Arizona State University, Tempe, Arizona, U.S.A.
useful in making some of these decisions. We have been investigating the electrically evoked auditory brainstem response (EABR) as a possible addition or alternative to some of these procedures. The data reported thus far with pre-implant stimulation have not been promising as a predictive measure of performance. We have also been making measurements of brainstem potentials in response to stimulation through two different types of multielectrode implants in an effort to assess whether or not the characteristics of these responses are related to a subject’s performance. Although there are some conflicting data, several animal studies have suggested that growth of response as a function of current level may be indicative of the number of surviving neurons in the auditory nerve (Smith and Simmons, 1983; Lusted et al., 1984; Hall, 1989). If one reasons that a greater number of surviving neurons would have the capacity to transmit more information to the central nervous system, then one may expect a higher level of performance with the implant in
124
subjects with relatively more surviving neurons. The measures of EABR in response to extracochlear stimulation that have been made to date, including observations in our laboratory, have shown responses which are variable in form (Meyer et al., 1984; Game et al., 1987; Black et al., 1987; Simmons et al., 1984; Gantz et al., 1988, van den Honert and Stypulkowski, 1986). In our experience with stimulation through the implant, the form of the EABR is similar across subjects, across time within a subject, and similar in form to the acoustically evoked brainstem response. These observations suggest that with promontory stimulation the source of the recorded potential may in some subjects include activity from outside of the auditory pathway. The consistency of the waveforms with stimulation through the implant suggests that in that case the response primarily reflects activity of the auditory pathways. Any predictive measure would obviously be most useful if it could be recorded preoperatively. Our rationale for using stimulation through the implant is that these measures should provide the ‘best case’ that one may expect. Any measure that correlates with performance during pre-implant testing should also correlate with performance during post-implant testing. In this paper, we describe experiments in which we have attempted to characterize the properties of the EABR as recorded from users of two different multichannel cochlear implants. In particular, we have examined the growth of the response with stimulus level for several different stimulating electrode configurations in each implant type. In addition, measures of EABR threshold and the rate of growth of the evoked potential with stimulus level are correlated with measures of word recognition ability in the same subjects.
plant ranged in age from 28 to 71 years with the time between their loss of hearing and date of implantation ranging from 1 to 34 years. Subjects with the Ineraid implant ranged in age from 33 to 71 years with the time between their loss of hearing and date of implantation ranging from 1 to 44 years. Several experimental manipulations are reported here. Not all subjects were run in all conditions. Some subjects were run as many as 4 times spanning a period of up to 30 months. Thirty of the subjects were implanted at University of Iowa Hospitals and Clinics (designated with the prefix I in subject identification number). Twelve were implanted at other centers and brought to our program for testing (designated with the prefix C in subject identification number). All subjects participated in an extensive program of audiological, psychological, and psychophysical tests as part of the Iowa Cochlear Implant Program. Stimuli Stimuli used for all of the experiments reported here were biphasic square pulses, either 100 ps per phase or 200 11s per phase. For both implant types, the speech processor was bypassed, but because of the differences between implant types, the details of stimulus delivery were quite different. In the Ineraid subjects, the stimuli were delivered directly to the electrode via the percutaneous plug. The stimuli were delivered through a custom-built, ground-isolated current generator which was capacitively coupled to the electrodes to avoid passing any DC current. We monitored the waveform delivered to the electrodes through
Methods Subjects We report data from 42 subjects, 23 of whom have previously been implanted with the Nucleus 22-electrode cochlear implant and 19 of whom have been implanted with the Ineraid Cochlear implant. Each subject had at least one month of experience with the implant before the measurements were made. Subjects with the Nucleus im-
I
I
0
0.5
1 .o
1.5
2.0
MS
Fig. 1. Typical current stimulation waveforms for the Ineraid implant. The current waveform in the Ineraid device was measured through a resistor in series with stimulating electrodes.
125
an isolated amplifier built into the current stimulator current. A typical waveform recorded from a subject is shown in Fig. 1. Waveforms were similar across the range of current levels used. The current to each electrode was measured on-line for each stimulus used in the experiment using this method. Current pulses were delivered with alternating initial phase, i.e., the presentation of a positive-negative pulse was followed by a negative-positive pulse. For the Nucleus subjects, the stimuli were delivered through the Speech Processor Interface. The specific characteristics of the Nucleus current source are described eleswhere (Crosby et al., 1985). The stimulus waveform is similar to that used in the Ineraid device, the primary difference being a short interval (in our experiments, 20 /.Ls) between the positive and negative pulses forming the biphasic waveform. A computer program was written to stimulate a given electrode pair with a train of biphasic pulses and simultaneously to trigger a separate computer that was used for response averaging. Stimuli were biphasic but not alternating in phase. Current could not be monitored on-line; the actual current values delivered were calculated subsequently from the individual current calibration data supplied by the manufacturer.
1 PV
1
!
0
2
6
4
a
10
MS
Fig. 2. Typical averaged and filtered EABR subjects with the Ineraid cochlear implant. time calibration is shown. For each subject current stimulus was chosen in the upper part dynamic range.
traces from five The voltage and the level of the of that subject’s
Procedure Recording
Recording electrodes were placed on the vertex (positive), contralateral mastoid (reference), and forehead (ground). A ground-isolated, low-noise amplifier with gain of 500,000 (bandwidth 0.1 Hz to 1.5 kHz, 6 dB/octave cutoff) preceded the input to the averaging computer. Special care was taken to shield the amplifier from RF to reduce stimulus artifact generated by the high frequency carrier signal used in the Nucleus system. Two averages of 1000 or 2000 sweeps were used in each stimulus condition. Response traces with voltage exceeding 20 PV were automatically rejected by the program. The number of rejected sweeps was typically less than 200 for each averaged trace. the stimulus artifact was After averaging, eliminated from the beginning of each trace and the signal was then digitally filtered with an FIR bandpass filter (57 order) with cutoff values 150 Hz to 3 kHz.
The typical experimental session lasted 3-4 hours. Subjects were seated in a reclining chair and asked to relax and/or sleep during the session. Some subjects received 5-10 mg of Valium to help them to sleep. Before any measurements were made, the subjects were asked to indicate behavioral threshold and uncomfortable level for each stimulus to be used in the session. The current levels of all stimuli were subsequently kept below these uncomfortable levels. The specific conditions that were completed with each subject varied because of time limitations. Results The EABR waveforms recorded from different subjects, different implants, and different electrodes within an implant were all consistent in their morphology. Examples of responses recorded from users of the two implants are shown in Fig. 2
126
(Ineraid) and Fig. 3 (Nucleus). For all of the traces in Figs. 2 and 3, the current level chosen was in the upper part of the subjects’ dynamic range. Two replications of the EABR are shown superimposed for each subject. Typically, the most prominent peak occurs about 4 ms after stimulus onset. Two earlier peaks occurring at approximately 1.3 and 2.0 ms were often observed at higher current levels. In a very few traces, a fourth peak is observed at a latency of approximately 3 ms. All of the analysis of amplitude and latency that will be discussed in the rest of the paper involves the measurement of the amplitude and latency of the latest and most prominent peak (positive to negative) of the EABR which most likely corresponds to wave V of the acoustically evoked ABR. EABRs were recorded from each subject using a set of standard stimuli and electrode configura-
CNlO
IN1 1
CN4
CN13
1 /I”
Fig. 3. Typical averaged and filtered EABR traces from subjects with the Nucleus cochlear implant. The voltage time calibration is shown. For each subject, the level of current stimulus was chosen to be in the upper part of subject’s dynamic range.
five and the that
tions. Since the implants’ design and normal stimulation modes were quite different, these standard stimulus parameters were chosen to be more like the normal mode of stimulation rather than attempting to be the same for both implant types. For the Ineraid implant, these stimuli consisted of monopolar stimulation with a pulse duration of 100 pss/phase. In the Nucleus implant, the stimulation was bipolar and a 200 ps/phase pulse duration was used. It was necessary to use a longer stimulus duration with Nucleus subjects because with bipolar stimulation, the response is generally less sensitive than with monopolar stimulation. The use of longer duration stimuli generally resulted in lower current thresholds. Fig. 4 illustrates measurements of the amplitude and latency of the fourth peak of the EABR recorded as a function of the current level of the bipolar pulse for both a Ineraid and Nucleus cochlear implant user. In each case, the parameter is the stimulating electrode pair. In the Ineraid implant, there are 8 electrodes: electrode 1 is the most apical and electrode 8 is placed in the temporalis muscle and is usually used as a common reference. The electrodes placed within the cochlea are spaced 4 mm apart. Stimulation in these cases is accomplished from each electrode in the cochlea to a common ground. The amplitude growth functions in Fig. 4A show data for four different electrode pairs as indicated in the legend. These functions generally increase monotonically with level and typically do not saturate over the range of levels used. This particular subject shows a clear difference in sensitivity among the 4 electrodes, electrode pair l-8 being the most sensitive and electrode pair 4-8 being the least. Many of the Ineraid subjects show a similar pattern; others show little difference in sensitivity across electrodes. Fig. 4B shows similar data from a Nucleus implant subject. For the Nucleus implant, the electrodes are numbered 1 to 22 with 22 being the most apical. The spacing of adjacent electrodes is 0.75 mm. The stimulation in these cases is bipolar +4, meaning that if electrode 1 is chosen, the stimulation is between electrodes 1 and 6. In each case, the stimulated electrode pair is indicated in the legend. In all cases, the stimulation with the Nucleus implant is with bipolar electrode pairs.
127 Subject
IS9 (Ineraid)
Sub[ect
IN8 (Nucleus)
0.5 -
o-o
l-8
O-0
2-8
A-A
3-8
A-A
4-8
B
0.4 --
s
o-o
l-5
O-0
5-10
A-A
10-15
A-A
15-20
0.3--
i I.4
4
5.5 -
. I
O-O
l-8
0-a
2-8
A-A
3-8
A-A
4-8
D
o-o O-0
5.2 --
.
1-5 5-10 10-15 15-20
A-A
A-A 9
4.9 --
z :
4.6 --
4 4.3 -0
0.1
0.3
0.2
Current
(mA)
7 I3.4
4.0 7 0.0
0.1
0.2
Current
0.3
0.4
(mA)
Fig. 4. Amplitude (positive peak to following negative trough) and latency (positive peak) of the fourth peak in the EABR response is plotted as a function of the stimulus current level for one subject with the Ineraid implant and one subject with the Nucleus implant. The stimulated electrode pair is the parameter in each graph and is indicated in the legend.
The sensitivity of each electrode pair can be quite different as seen with electrode lo-15 in this subject, which has a higher threshold than the other stimulated pairs. Nevertheless, we have observed no consistent pattern across subjects in which a particular electrode pair is most sensitive. The latency of the measured responses are shown for each of the two subjects in Figs. 3C and 3D. The latency generally decreases slightly with increasing current level, but the magnitude of the latency changes observed are much smaller than corresponding changes in latency observed with acoustic stimulation. In some cases this trend was not evident, but over 95% of the measured latency functions had a negative slope when fit with linear regression. The measures of amplitude growth are sum-
marized in the bar graphs in Fig. 5. The individual amplitude functions were fit with a linear function and the slope of that function and threshold (defined to be the current level necessary to elicit a response amplitude of 0.1 CLV) was determined. For subjects where we had made repeated measures for different electrodes, the functions for each repetition were fit and the values of threshold and slope averaged. Data for the Nucleus and Ineraid implants are reported separately in the figure. Fig. 5A and 5C show average data for Ineraid subjects. Both the threshold and slope data show a consistent change across electrode configuration. Stimulation of basal electrodes results in higher thresholds and shallower growth functions than is found with stimulation of apical electrodes. To test these differences, we performed
128 Inerold 0.30, A 0.25
-5 0.20 3 0 0.15 f $
0.10 0.05
2. 0.00
L
0.25
Nucleus
herald 20.0
0.0
4.0 C
I--
1-8
2-0
3-a
D
4-0
Electrode polr
Electrode polr
Fig. 5. Threshold of EABR response was averaged across subjects with each implant type and is plotted in (A) for Ineraid users and in (B) for Nucleus users. Slope of the growth function was calculated using linear regression. These values were averaged across subjects and are plotted in (C) for lneraid users and in (D) for Nucleus users.
a multiple test analysis of variance where stimulating electrode was the within subject factor. For the group of Ineraid subjects, the analysis for both threshold (F = 3.73, (Y= 0.021) and slope (F = 6.06, (Y= 0.002) showed significant differences across electrodes. Follow-up paired samples t-tests showed that electrode pairs 1-8 to 4-8 were significantly different (0.05 level) in threshold and electrode pairs 1-8 to 3-8 and l-8 to 4-8 were significantly different in slope. Similar tests with Nucleus subjects showed no significant differences for either threshold or slope measures. The plots in Fig. 5 show slope calculated in linear current units. In doing this, the relative size of the stimulus increment relative to threshold is not taken into account. By calculating the slope in logarithmic units (in dB), the differences between the threshold values for the different electrodes
should not affect the measure of growth. When the calculations were done in this way, the differences among electrodes tended to be smaller, but the statistical analysis still showed significant differences among electrodes for the Ineraid implant with follow-up t-tests showing differences between electrode pairs 1-8 to 3-8 and 2-8 to 3-8. The average latency data are illustrated in Fig. 6. To compare values of absolute latency, the latency at the highest current level used was measured for each subject and stimulus condition. These values are shown in the upper two graphs. In addition, each latency vs. current level function was fit by linear regression. The slope of that function is shown in the lower two graphs of Fig. 6. For both the Nucleus and Ineraid subjects,
129 herald
Inerald -10.0
C
2
