Monopolar Detection Thresholds Predict Spatial Selectivity of Neural Excitation in Cochlear Implants: Implications for Speech Recognition.

Ning Zhou,Fan-Gang Zeng

doi:10.1371/journal.pone.0165476

Abstract

The objectives of the study were to (1) investigate the potential of using monopolar psychophysical detection thresholds for estimating spatial selectivity of neural excitation with cochlear implants and to (2) examine the effect of site removal on speech recognition based on the threshold measure. Detection thresholds were measured in Cochlear Nucleus® device users using monopolar stimulation for pulse trains that were of (a) low rate and long duration, (b) high rate and short duration, and (c) high rate and long duration. Spatial selectivity of neural excitation was estimated by a forward-masking paradigm, where the probe threshold elevation in the presence of a forward masker was measured as a function of masker-probe separation. The strength of the correlation between the monopolar thresholds and the slopes of the masking patterns systematically reduced as neural response of the threshold stimulus involved interpulse interactions (refractoriness and sub-threshold adaptation), and spike-rate adaptation. Detection threshold for the low-rate stimulus most strongly correlated with the spread of forward masking patterns and the correlation reduced for long and high rate pulse trains. The low-rate thresholds were then measured for all electrodes across the array for each subject. Subsequently, speech recognition was tested with experimental maps that deactivated five stimulation sites with the highest thresholds and five randomly chosen ones. Performance with deactivating the high-threshold sites was better than performance with the subjects’ clinical map used every day with all electrodes active, in both quiet and background noise. Performance with random deactivation was on average poorer than that with the clinical map but the difference was not significant. These results suggested that the monopolar low-rate thresholds are related to the spatial neural excitation patterns in cochlear implant users and can be used to select sites for more optimal speech recognition performance.

Highlights

One of the major factors that limit the processing of the electrical pulsatile stimulation with modern multichannel cochlear implants (CI) is channel interaction
The thresholds were compared to the traditional measure of spatial selectivity of neural excitation, i.e., slope of the forward-masking functions
For probe sites that were masked for both the apical and basal directions, the basal and apical slopes were averaged and the average was used as the measure of spatial selectivity for neural excitation

Summary

Introduction

One of the major factors that limit the processing of the electrical pulsatile stimulation with modern multichannel cochlear implants (CI) is channel interaction. The electrodes often excite broader areas in the cochlea than desired, resulting in spatial overlapping in the neural excitation between channels. This explains why speech recognition saturates in quiet when the number of activated electrodes increases beyond eight [1], and why CI users do not demonstrate better performance in fluctuating noises than in steady state noise as normal-hearing listeners do [2]. Because the variation patterns in pathology and electrode position are considerably different across the implanted ears [3,4], the neural spatial excitation pattern is expected to be ear specific and site specific within ears

Objectives

Methods

Results

Conclusion