The compound action potential (CAP) of the auditory nerve reflects the collective response of auditory nerve fibers to a brief stimulus. In cochlear implant (CI) users with residual hearing in the implanted ear, CAPs can be evoked both acoustically and electrically and recorded through CI electrodes. Previous research has shown that electric and acoustic stimuli can mask each other, resulting in reduced evoked responses and impaired auditory perception. The CAP in subjects with combined electric-acoustic stimulation provides a unique opportunity to objectively study such electric-acoustic interactions.
This study presents a computational modeling framework that simulates CAPs evoked by both electric and acoustic stimulation considering refractive interaction in the auditory nerve. The proposed framework consists of three components: a 3D finite element method model of an implanted cochlea to predict electric voltage spread, a point neuron model to efficiently predict spike times in response to electric-acoustic stimulation, and a multi-compartment neuron model to simulate excitation and propagation of action potentials along the nerve fibers. Previous models have typically focused on either electric or acoustic stimulation, often assuming a “unitary response” of all nerve fibers. This study introduces a more sophisticated approach that replaces the unitary response with individual single-fiber CAP contributions for the nerve fibers that are obtained with the multi-compartment model.
We show that the new modeling framework replicates human CAP morphologies for electric as well as acoustic stimulation and predicts CAP amplitude growth and suppression of electrically evoked CAPs by acoustic stimuli. In addition, the new model design can account for the dependence on stimulation and recording sites in the cochlea. The model has the potential to guide future applications such as objective assessment of hearing thresholds and neural health using acoustically or electrically evoked CAPs in CI users.
The modeling framework is available in GitLab