A Model-Based Approach for Optimizing Cochlear Implant Stimulation

University Of Washington, Seattle WA

Investigators

Linked publications & trials

Abstract

DESCRIPTION (provided by applicant): The long-term goal of this research is to develop novel stimulation strategies that improve speech and music perception for cochlear implant (CI) users. The approach is based on computational modeling because these techniques, when carefully constrained by experimental studies, can provide deep insight into neural mechanisms that limit how sound information is transmitted to the brain. Moreover, computationally efficient models present the opportunity to evaluate a broader set of stimulation strategies than can be tested in experimental and clinical studies. This opens the door for the discovery of novel stimulation patterns that optimize the delivery of sound information to CI listeners. This research is consistent with the stated goals of the National Institute on Deafness and Other Communication Disorders which include supporting research training in the disordered processes of hearing and supporting efforts to create devices which substitute for lost and impaired sensory function. There are two specific aims of the proposed research. First, a computationally efficient point process model will be derived from an existing biophysically-detailed model that has been parameterized to represent the mammalian auditory nerve response to cochlear implant stimulation. This aim will be accomplished by fitting model parameters to spike train output of the detailed model using maximum-likelihood methods. The second aim is to analyze neural encoding of cochlear implant stimuli by simulating psychophysical tests of temporal processing. Amplitude modulation detection, gap detection, and Schroeder phase discrimination experiments will be simulated using a two alternative forced choice paradigm. By comparing simulation results to known experimental data, it will be possible to identify situations in which the neural response limits performance on psychophysical tests due either to. specific neural mechanisms or to deficiencies in current speech processing strategies. In addition, the computationally efficient model can be used to rapidly explore the space of possible stimulation strategies using global optimization algorithms to search for stimulation strategies that improve the transmission of temporal information to CI listeners. It is hoped that simulations will identify new speech processing strategies that, when implemented in future CI devices, will improve speech and music perception for CI listeners.