Speech understanding in noise with a Med-El COMBI 40+ cochlear implant using reduced channel sets.

Objective: The objective of the investigation described in this paper was the determination of the number of (widely spaced) active electrodes needed for users of a COMBI 40+ cochlear implant to achieve asymptotic performance in the recognition of speech against a background of wideband noise.

Design: This study measured the performance in speech tests of patients using the Med-El implementation of continuous interleaved sampling with widely spaced electrode pair subsets of 2, 3, 4, 6, 8, and 10 out of a possible maximum of 12. An eight-vowel test, a 16-consonant test, and BKB sentences were presented against a background of pink noise. Additionally, AB monosyllabic words were presented both in quiet and in noise to processors with 6, 8, and 11 widely spaced electrodes. 11 subjects participated in the study.

Results: Using moderate signal-to-noise ratios, for these patients the curve relating percentage score to increasing numbers of active channels approached an asymptote before the 10-channel data point was reached. Asymptotic performance was achieved using four channels for consonants, and eight channels for sentences. Understanding of monosyllabic words reached a maximum value at a similar number of channels for both quiet conditions and against a background of pink noise, and the mean increase in test score between 6 and 11 channels was only 7%.

Conclusions: These results are similar to those of previous experiments carried out in quiet listening conditions. The data suggest that 12 frequency channels (the number implemented by the COMBI 40+ cochlear implant) are more than adequate for users to achieve asymptotic performance levels in clinical speech tests applied in the presence of wideband noise at moderate signal-to-noise ratios.