Analysis of Factors Affecting the Performance of Retinal Prostheses Using Finite Element Modelling of Electric Field Distribution in the Retina

This dissertation proposes a computational framework targeted at improving the design of currently employed retinal prostheses. The framework was used for analysing factors impacting the performance of prostheses in terms of electrical stimulation for retinal neurons, which might lead to a perception of pixelated vision. Despite their demonstrated effectiveness, the chronic and safe usage of these retinal prostheses in human and animal trials is jeopardised due to high stimulation thresholds. This is related to the distance between the stimulating electrodes and the retinal neurons resulting from the implantation procedure. The major goal of this dissertation was to evaluate the stimulation efficacy in current implantable planar microelectrode-based retinal prostheses and consequently demonstrate their weakness, thereby providing scope for the development of future implants. The effect of geometrical factors i.e., electrode-retina distance and electrode size on stimulation applied to the retina by retinal prostheses was studied. To this end, a finite element method based simulation framework to compute electric field distribution in the retina was constructed. An electrical model of the retina was an integral part of the framework, essentially represented by a resistivity profile of the multi-layered retina. The elements of a retinal prosthesis were modelled by incorporating realistic electrode sizes, an anatomical and electrical model of the retina, a precise positioning of stimulation and return electrodes and the location of the implant with respect to the retina representing the epiretinal and subretinal stimulation schemes. The simulations were carried out both in quasi-static and direct current (DC) modes. It was observed that electrode-electrolyte interface and tissue capacitance could be safely neglected in our model based on the magnitude of the applied voltage stimulus and frequencies under consideration. Therefore, all simulations were conducted in DC mode. Thresholds and lateral extents of the stimulation were computed for electrode sizes corresponding to existing and self-fabricated implants. The values and trends obtained were in agreement with experiments from literature and our collaborators at the les Hôpitaux Universitaires de Genève (HUG). In the subretinal stimulation scheme, the computed variation of impedance with electrode-retina distance correlated well with time varying in vivo impedance measurements in rats conducted in collaboration with the Institut de la Vision, INSERM, Paris. Finally, it was also reiterated that the currently employed retinal prostheses are not very efficient due to a significant distance between the stimulation electrode and the retinal cells. In addition, I present a new experimental technique for measuring the absolute and local resistivity profile in high-resolution along the retinal depth, based on impedance spectroscopy using a bipolar microprobe. This experiment was devised to extract the resistivity profile of an embryonic chick retina to construct an electrical model for the simulation framework to simulate in vitro retinal stimulation experiments conducted by HUG collaborators. We validated the capability of the technique in rat and embryonic chick retinas. In conclusion, the computational framework presented in this dissertation is more realistic than those found in literature, but represents only a preliminary step towards an accurate model of a real implantation scenario in vivo. The simulation results are in agreement with results from clinical trials in humans for epiretinal configuration (literature) and with in vitro results for epiretinal and subretinal stimulation applied to chick retinas (HUG). The developed simulation framework computes quantities that can form a reference for quality control during surgery while inserting implants in the eye and functionality checks by electrophysiologists. Furthermore, this framework is useful in deciding the specifications of stimulation electrodes such as optimal size, shape, material, array density, and the position of the reference electrode to name a few. The work presented here offers to aid in optimising retinal prostheses and implantation procedures for patients and eventually contributes towards improving their quality of life.