Word Recognition and Frequency Selectivity in Cochlear Implant Simulation: Effect of Channel Interaction.

Pierre-Antoine Cucis,Christian Berger-Vachon,Ruben Hermann,Stéphane Gallego,Eric Truy,Hung Thaï-Van

doi:10.3390/jcm10040679

Abstract

In cochlear implants (CI), spread of neural excitation may produce channel interaction. Channel interaction disturbs the spectral resolution and, among other factors, seems to impair speech recognition, especially in noise. In this study, two tests were performed with 20 adult normal-hearing (NH) subjects under different vocoded simulations. First, there was a measurement of word recognition in noise while varying the number of selected channels (4, 8, 12 or 16 maxima out of 20) and the degree of simulated channel interaction (“Low”, “Medium” and “High”). Then, there was an evaluation of spectral resolution function of the degree of simulated channel interaction, reflected by the sharpness (Q10dB) of psychophysical tuning curves (PTCs). The results showed a significant effect of the simulated channel interaction on word recognition but did not find an effect of the number of selected channels. The intelligibility decreased significantly for the highest degree of channel interaction. Similarly, the highest simulated channel interaction impaired significantly the Q10dB. Additionally, a strong intra-individual correlation between frequency selectivity and word recognition in noise was observed. Lastly, the individual changes in frequency selectivity were positively correlated with the changes in word recognition when the degree of interaction went from “Low” to “High”. To conclude, the degradation seen for the highest degree of channel interaction suggests a threshold effect on frequency selectivity and word recognition. The correlation between frequency selectivity and intelligibility in noise supports the hypothesis that PTCs Q10dB can account for word recognition in certain conditions. Moreover, the individual variations of performances observed among subjects suggest that channel interaction does not have the same effect on each individual. Finally, these results highlight the importance of taking into account subjects’ individuality and to evaluate channel interaction through the speech processor.

Highlights

Modern cochlear implants (CIs) provide unique results in the rehabilitation of severe and profound deafness [1]
In this study, when we changed the degree of simulated channel interaction there was a strong within-subject correlation between psychophysical tuning curves (PTCs) frequency selectivity and average speech recognition in noise
The results showed a correlation between the evolution of the frequency selectivity and the average speech recognition in noise across the “Low” and the “High” degree of simulated spread of excitation

Summary

Introduction

Modern cochlear implants (CIs) provide unique results in the rehabilitation of severe and profound deafness [1]. Electrode arrays are currently composed of 12 to 22 electrodes depending on the manufacturer [2]. Thanks to multi-electrode technology, speech perception, and quality of life of CI users have been considerably enhanced [3,4]. An inherent outcome of multiplying the number of channels or electrodes is that it may lead to channel interaction. The overlap of electrical fields stimulates a large number of nerve fibers and can create an overlap among the “neural channels”. Depending on the overlap degree, signals, information, neural integration, and neural processing can be degraded [5,6]

Objectives

Methods

Discussion

Conclusion