Objective Identification of Simulated Cochlear Implant Settings in Normal-Hearing Listeners Via Auditory Cortical Evoked Potentials

Objectives: Providing cochlear implant (CI) patients the optimal signal
processing settings during mapping sessions is critical for facilitating
their speech perception. Here, we aimed to evaluate whether auditory
cortical event-related potentials (ERPs) could be used to objectively
determine optimal CI parameters.
Design: While recording neuroelectric potentials, we presented a set of
acoustically vocoded consonants (aKa, aSHa, and aNa) to normal-hearing
listeners (n = 12) that simulated speech tokens processed through
four different combinations of CI stimulation rate and number of spectral
maxima. Parameter settings were selected to feature relatively fast/
slow stimulation rates and high/low number of maxima; 1800 pps/20
maxima, 1800/8, 500/20 and 500/8.
Results: Speech identification and reaction times did not differ with
changes in either the number of maxima or stimulation rate indicating
ceiling behavioral performance. Similarly, we found that conventional
univariate analysis (analysis of variance) of N1 and P2 amplitude/latency
failed to reveal strong modulations across CI-processed speech conditions.
In contrast, multivariate discriminant analysis based on a combination
of neural measures was used to create “neural confusion matrices”
and identified a unique parameter set (1800/8) that maximally differentiated
speech tokens at the neural level. This finding was corroborated by
information transfer analysis which confirmed these settings optimally
transmitted information in listeners’ neural and perceptual responses.
Conclusions: Translated to actual implant patients, our findings suggest
that scalp-recorded ERPs might be useful in determining optimal signal
processing settings from among a closed set of parameter options and
aid in the objective fitting of CI devices.