To facilitate design of a multichannel vestibular prosthesis that can restore sensation to individuals with bilateral loss of vestibular hair cell function, we created a virtual labyrinth model. Model geometry was generated through 3-dimensional (3D) reconstruction of microMRI and microCT scans of normal chinchillas (Chinchilla lanigera) acquired with 30-48 μm and 12 μm voxels, respectively. Virtual electrodes were positioned based on anatomic landmarks, and the extracellular potential field during a current pulse was computed using finite element methods. Potential fields then served as inputs to stochastic, nonlinear dynamic models for each of 2,415 vestibular afferent axons with spiking dynamics based on a modified Smith and Goldberg model incorporating parameters that varied with fiber location in the neuroepithelium. Action potential propagation was implemented by a well validated model of myelinated fibers. We tested the model by comparing predicted and actual 3D angular vestibulo-ocular reflex (aVOR) axes of eye rotation elicited by prosthetic stimuli. Actual responses were measured using 3D video-oculography. The model was individualized for each animal by placing virtual electrodes based on microCT localization of real electrodes. 3D eye rotation axes were predicted from the relative proportion of model axons excited within each of the three ampullary nerves. Multiple features observed empirically were observed as emergent properties of the model, including effects of active and return electrode position, stimulus amplitude and pulse waveform shape on target fiber recruitment and stimulation selectivity. The modeling procedure is partially automated and can be readily adapted to other species, including humans.
- Finite element
ASJC Scopus subject areas