Optimizing electrode and filter selection in cochlear implant speech processor maps.

This study examined two hypotheses: that speech understanding of cochlear implantees could be improved by removing electrodes that exhibit nontonotopic percepts from the speech processor map and that speech understanding could be improved by extending the range of high frequencies that are mapped to the electrodes. Electrodes producing nontonotopic percepts were identified using a multidimensional scaling procedure with seven users of the Nucleus CI22 implant and Spectra processor. Two experimental maps were created with those electrodes removed: the first using the same set of filters as their clinical map and the second using the complete set of filters available. After periods of take-home experience, speech perception was tested and compared for the two experimental maps and their own clinical map. It was found that removing nontonotopic electrodes did not improve speech perception, possibly due to the deleterious side effect of shifting the frequency-to-electrode allocation. Also, extending the high-frequency range of the map did not improve speech perception, possibly due to the poor sensitivity of this processor to high-frequency sounds.