Abstract

Dynamic ranges, measured as the difference between threshold of perception and maximum comfortable level and loudness estimation for five arbitrary loudness levels, were measured in 13 postlingually and 5 prelingually deafened subjects, implanted with multichannel cochlear prostheses. Experimentes were conducted for varied widths of the pulsatile stimuli, namely pulses of 200, 100 and 50 µs/phase. The electrotests involved measurements of thresholds of perception, maximum comfortable levels and equal loudness countours which were obtained for five arbitrary loudness levels. The electric charges delivered at threshold levels were also calculated. The results show that for almost all the patients, T and C values increase and dynamic ranges expand for decreasing pulses width. The equal loudness contours for the different loudness levels tended to be least compressed by decreasing pulse widths; equal loudness contours are not equal charge contours. The threshold charges decrease when decreasing the width of the pulses. Speech perceptual data were obtained for varied pulse widths for postlingual patients using the speech processor alone, in order to study the usefulness of decrased pulse duration which yields favorable results in psychophysical tests. Results indicate that better fluent speech comprehension is achieved when shortening the width of the pulses. STUDY AIMS
In a cochlear prosthesis, the Dynamic Range (DR) of the acoustical signal, measured as the difference between C (maximum comfortable level) and T (threshold of perception), must be compressed into the electrical DR of each patient, which is usually restricted. The enlargement of electrical dynamic ranges would help variations in the acoustical signal levels to evoke less compressed loudness levels in the implanted subjects. As a consequence, speech features related to intensity variations could probably be better perceived and behavioral speech perception would be enhanced.
In an attempt to find the condition which yields better speech comprehension from the implanted subjects, this study explores the influence of pulse widths on the dynamic range (1-4) of 18 subjects fitted with the Nucleus 22 cochlear implant. DR and Equal Loudness Contours (ELC) determined for five loudness levels, were detrmined for pulse widths (PW) of 200, 100 and 50 µs/phase. This experiment attempts to study the separation (distance) between succesive contours as a function of the PW. The minimal delivery of electric charge at threshold was also determined as a fucntion of the PW. Finally, the ability to understand connected speech was measured in the postlingually deafened patients to support our psychophysical results.

Material and methods

measured as the difference between C (maximum comfortable level) and T (threshold) Eighteen patients, 13 postlingually and 5 prelingually deafened subjects participated in this study. Table I summarizes some of their details. The tests were conducted with the 6.6 or 6.9 versions of the Diagnosis and Programming System (DPS) of the Nucleus 22. The psychophysical experiments began after each patient had had enough experience to understand what was required of him/her and their T/C levels were stable and reliable. The subjects’ varying difficulties (limited understanding of the task, tinnitus) influenced the amount of training; threrefore, training periods varied among subjects.
For all the patients, the experiments began with the T/C determination in current levels, collected and averaged every two electrodes, in 5 work-sessions. In order to minimize learning effects, the sequence of tests for different pulse widths was randomized. Stimuli used were trains of biphasic pulses of 500ms duration, 500 ms interstimulus wait time and pulse rate of 125Hz.
Loudness was estimated for five stimulus levels: T, soft (s), medium (m), loud (l) and C every two electrodes. Each data point was averaged among five judgements for each electrode. As before, the sequence for the pulse widths was randomized. This test was made without previous training.
Electrical charge for threshold was determined as the product of the amplitude (calculated using the individual Output Current Tables) multiplied by the corresponding PW.
Fluent speech perception without visual aid was measured for the postlingual patients. Fifteen stories of 5 sentences each were used in 5 weekly work-sessions, one set for each pulse width. That made a total of five stories for each PW condition. The speech tracking method (5) was used as an evaluation procedure. The sequence of PW was again randomized in order to minimize training effects. This test was done in a quiet room, at a comfortable listening level with the speech proccesor of each patient. The speaker was a woman seated 1.5 m behind the patients.
Three statistical analyses to study the difference among the pulses were performed for the patients. First, DRs were examined using a one way analysis of variance (ANOVA). Second, the differences between the five isosonic curves were examined using a two-way analysis of variance. Finally, the word test was examined using a repeated measure analysis of variance; the results of 5 sections for each patient were considered. The first work session using pulses of 200 µs/phase (which was the worst response) was taken as the reference. A post hoc Newman-Keuls multiple comparison between the mean values using each pulse width condition and for each patient, was performed for the three analyses.

Results

The results obtained for the three PW conditions are depicted in the following figures: Dynamic Ranges averaged among the patients and work sessions in Fig 1; equal loudness contours in Fig 2a, b, and c respectively; threshold charges in Fig 3 and fluent speech comprehension measured in words per minute, averaged for poslingually deafened patients in five work sessions, in Figure 4.

Conclusions

For almost all the patients, T and C levels increase when decreasing PW for all the electrodes but increments in C levels are higher than increments in T levels values; as a consequence, shortening the pulses causes an expansion of the Dynamic Ranges. The ELC tend to be least compressed which in turn seem to improve the fluent speech comprehension, probably because speech cues related to changes in intensity are better perceived.
It is clear that equal charge does not produce equal loudness. Our results regarding threshold charges are identical to those obtained by Shannon (1-3): as pulse duration increases, threshold charge also increases; threshold is attained with less charge by using narrower pulse widths. However, when comparing DR data for different PW, our results are quite different: DR increases when the PW decreases.
This study provides data of psychophysical tests for patients whose etiologies differ. However, the result are similar for almost all of them: shortening the pulses brings about an improvement in speech understanding for postlingual patients and an expansion of auditory ranges for both pre and postlingual patients. Experiments involving place pitch and repetition rate perception using varying pulse widths (6) also indicate that results are better for short pulses.

References

1. Shannon RV. Growth of loudness for sinusoidal and pulsatil electrical stimulation. Ann Otol Rhinol Otolaryngol 1981;90 Suppl 82:13-14.
2. Shannon RV. Threshold and loudness function for pulsatil stimulation of cochlear implants. Hearing Research 1985;18:135-143.
3. Shannon RV. Threshold functions for electrical stimulation of the human cochlear nucleus. Hearing Research 1989; 40:173-178.
4. Shannon RV, Otto SR. Psychophysical measures from electrical stimulation of the human cochlear nucleus. Hearing Research 1990;47:159-168.
5. De Filippo CL, Scott BL. A method for training and evaluating the reception of ongoing speech.
J Acoust Soc Am 1978;64(4):1186-1192.
6. Aronson L, Rosenhouse J et al. Pitch perception in patients with a multichannel cochlear implant using various pulse widths. Medical Progress through Technology 1994;20:43-51.

Figure 1. Dynamic Ranges averaged among 18 patients in five work sessions.

Figure 2a. Equal loudness contours for pulses of 200µs/phase.

Figure 2b. Idem for pulses of 100µs/phase.

Figure 2c. Idem for pulses of 50µs/phase.

Figure 3. Threshold charge for different pulse durations averaged among 18 patients.

Figure 4. Fluent speech perception in words per minute (wpm) as a function of pulse durations for 13 patients.