Our long term goal is to understand the of the high variability in speech performance in implant listeners and to reduce the variability by improving performance, particularly in poorer performers (NIH Consensus Statement on Cochlear Implants, 1995). Our hypothesis is that the high variability across implant listeners is due to the differences in individual nerve survival and to differences in the coupling of the implanted electrodes to those surviving neurons. We hypothesize that speech recognition will be optimized if we can quantify the number and location of potential receiving channels in an individual implant patient and customize the delivery of speech information to those channels. Being able to deliver optimum speech cues to a given number of receiving channels depends on understanding of how to divide the speech cues into maximally information-bearing channels, a problem we term the analysis problem. Specifically, we will: (1) Quantify the number and location of potentiaI receiving channels in cochlear implant and brainstem implant patients. Seven measures (loudness summation, forward masking, simultaneous masking, gap detection, electrode discrimination, cortical dipole source localization, mismatch negativity) of electrode interaction will be collected to define the tonotopic specificity with which a spectral pattern can be represented in an individual patient. These measures will attempt to define the homogeneity and location of surviving neural elements, which will define the effective number and location of receiving channels in individual patients. Simultaneous and nonsimultaneous, psychophysical and electrophysiological, measures will be used at threshold and suprathreshold intensity levels. (2) Evaluate the spectral distribution of envelope cues in speech that are most critical for normal speech recognition (the analysis problem). Spectral and intensity cues will be independently and selectively degraded in an acoustic speech processor simulation of a cochlear implant. Phoneme and sentence recognition will be measured as a function of the number of channels to evaluate the importance function of each parameter. A secondary hypothesis is that we can trade off number of channels and corruption of some cues, e.g. that additional channels can offset the deleterious effect of an amplitude distortion or a spectral location shift. (3) Evaluate new speech processor designs to optimize the transmission of the most critical spectral-temporal cues of speech to individual implant patients based on electrode interaction measures. Psychophysical measures of electrode interaction (from #1 above) will be used to select the number and location of electrodes for use in a CIS-type speech processor. Given the number and location of the channels in an individual patient, speech envelope cues will be divided and presented to the selected electrodes in a manner that optimizes performance (determined in #2 above). Basic research in electrical stimulation of the human auditory system is important to advance our knowledge of basic auditory processing in normal and impaired listeners, and to improve the design and performance of implanted prostheses. This proposal will investigate two areas that we feel are fundamental to future improvements in cochlear implant performance: electrode interactions (which will define the number and location of potential receiving channels in an individual patient) and the tonotopic distribution of speech envelope cues (which will define the optimal way to analyze the speech information for delivery to a given number of channels).
Showing the most recent 10 out of 43 publications
