A Computational Model Simulates Light-Evoked Responses in the Retinal Cone Pathway.

Abstract

Partial vision restoration on degenerated retina can be achieved by electrically stimulating the surviving retinal ganglion cells via implanted electrodes to elicit a signal corresponding to the natural response of the cells. Realistic computational models of electrical stimulation of the retina can prove useful to test different stimulation strategies and improve the performance of retinal implants. Simulation of healthy retinal networks and their dynamical response to natural light stimulation may also help us understand how retinal processing takes place via a series of electrical signals flowing through different stages of retinal processing, ultimately giving rise to visual percepts. Such models may provide further insights on retinal network processing and thus guide the design of retinal prostheses and their stimulation protocols to generate more natural percepts. This work aims to characterize the photocurrent generated by healthy cone photoreceptors in response to a light flash stimulation and the resulting membrane potential for the photoreceptors and its postsynaptic cone bipolar cells. A simple network of ten cone photoreceptors synapsing with a cone bipolar cell is simulated using the NEURON environment and validated against patch-clamp recordings of cone photoreceptors and ON-type bipolar cells (ON-BC). The results presented will be valuable in modeling light-evoked or electrically stimulated retinal networks that comprise cone pathways. The computational models and methods developed in this work will serve as an integral building block in the development of large and realistic retinal networks.Clinical Relevance- Accurate computational model of a retinal neural network can help in predicting cell responses to electrical stimulation in vision restoration therapies using prostheses. It can be leveraged to optimize the stimulation parameters to match the natural light response of the network as closely as possible.