Three-dimensional spiraling finite element model of the electrically stimulated cochlea.

Abstract

The objective of the article is to provide an accurate model of the human cochlea with which potential distributions and thus neural excitation patterns around cochlear implant electrodes can be determined. Improvements on previous models of the implanted cochlea are that this model 1) includes the spiral nature of the cochlea as well as many other anatomical details (and it is a model of the human cochlear rather than the guinea pig cochlea), and 2) facilitates modeling of different electrode geometries, array locations and electrode separations without changing the structure of the model. A three-dimensional spiraling finite element model of the human cochlea was created. The model incorporates the effect of neighboring canals and conduction along the fluid-filled canals of the cochlea. Potential distributions are used as inputs to a nerve fiber model to investigate auditory nerve excitation patterns around intracochlear electrode arrays. Potential distributions around intracochlear electrodes generated with the finite element model are presented. The effects of electrode separation, electrode geometry and array location on excitation threshold, excitation spread and ectopic excitation (i.e., excitation of nerve fibers at an undesirable location) are demonstrated. The following conclusions should be considered preliminary, as their accuracy depends on the exactness of the underlying model. The spiraling geometry of the cochlea causes asymmetry in potential distributions. The location of electrodes along the length of the basilar membrane has a stronger influence on the site of excitation than the polarity of the leading phase of the stimulus. Array location is the primary parameter that controls excitation spread. Threshold currents and the effect of ongoing loss of peripheral dendrites on threshold currents can be limited by placing arrays close to the modiolus. Point electrode geometries are recommended above banded electrode geometries only when the array can be placed close to the modiolus. There is a tradeoff between array location and the degree of ectopic stimulation caused by a specific array location. Bimodal excitation patterns exist at comfortable stimulus intensities for longitudinal bipolar electrode configurations. It is shown that an electrode configuration with an electrode separation of approximately half that of the bipolar electrode separation of the Nucleus electrode can be used instead of radial and offset radial electrode configurations to create unimodal excitation patterns. The stimulation resolution of cochlear implant electrode arrays can potentially be improved by increasing the number of electrode contacts in an array.