A Different Approach to Using Neural Response Telemetry for Automated Cochlear Implant Processor Programming

In Brief Objective: This study explores the theoretical relation between the psychophysically measured current levels required for sound processor fitting in cochlear implants and the objectively measured compound action potential threshold (as measured by Neural Response Telemetry, NRT). The objective was to gain understanding of the variability across implantees in this relation and determine possible ways (using objective measures) of improving the predictability of NRT thresholds for behavioral levels needed for mapping. Design: A model of how rate of stimulation affects loudness is presented. The model can be used to understand differences among implantees in the way that rate affects loudness and hence explain the disappointing correlation between NRT and psychophysical measures. Suggestions are made, based on the model, for additional information that may improve the usefulness of NRT measurements. One such option (measuring the effect of interphase gap on NRT amplitude) was experimentally explored in eight subjects (26 electrodes). It was hypothesized that the current change required to maintain equal NRT amplitude when interphase gap was changed from 8 to 45 μsec would be correlated with the offset between behavioral and NRT thresholds. Results: The above hypothesis was not supported by the data, and several possible reasons for this outcome are discussed. Conclusions: The loudness model provides useful insights into why NRT thresholds are not good predictors of the behavioral levels needed for mapping and how NRT might be made more useful by additional objective information. These insights should be investigated in further experimental studies. In this paper, the theoretical bases of the poor correlations found between NRT thresholds and the behavioral levels needed for speech processor fitting were examined. It was shown that the average difference between NRT threshold and behavioral levels is predictable from the average equal-loudness curve for rate of stimulation. The reason for the poor correlation for individual measurements is the individual differences in how rate of stimulation affects loudness. The three factors that affect the shape of the equal-loudness curves for rate of stimulation are neural refractoriness, loudness growth functions, and individual preferences for loudness levels at the output of the processor. Using the insight of this theoretical model, several suggestions are made for objective measurements which might improve the predictability of NRT thresholds. One such hypothesis was tested and found not to be supported. However, hypotheses involving more-direct objective measures should be tested in the future.