Optimizing Selectivity of Epiretinal Stimulation using Local Returns

Abstract

Purpose : A limitation of epiretinal prostheses is that current spread can produce non-selective activation of many retinal ganglion cells (RGCs), resulting in an imprecise artificial visual signal. The purpose of this work is to test whether using a local return (ground) arrangement with high-density multi-electrode arrays, rather than a distant return, can more selectively activate an individual targeted RGC while avoiding activation of other RGCs.

Methods : A custom 512-electrode system (60μm pitch, 8-15μm diameter, hexagonal grid) was used to stimulate and record from hundreds of RGCs in isolated macaque retina. Electrical stimuli were provided on one or more channels while recording RGC activity from all channels. The distant return pattern consisted of a charge-balanced triphasic pulse on a single electrode, with a bath ground ~1 cm away. The local return pattern consisted of the same central electrode current, with simultaneous current waveforms of opposite sign and 1/6 amplitude on the six immediately surrounding return electrodes. Spiking activity was manually analyzed to separate spikes from electrical artifact. The locations of cell bodies and axons were determined from the electrical image, i.e. the average spatiotemporal pattern of voltage deflections produced across the array during a spike. The geometric arrangement of cells relative to the stimulating electrode was examined to test how it influenced the selectivity of activation.

Results : Selectivity was defined as the ratio of activation threshold of a non-target cell to the threshold of a target cell; thus, high ratios indicated activation of the target cell without activation of the non-target cell. Selectivity was examined directly in recordings from 7 cell pairs, and indirectly in surrogate data from 112 pairs of cells recorded on different electrodes. In all cases, the target cell was near the central electrode and the non-target cell was near one of the surrounding return electrodes. In all 7 pairs and 102/112 surrogates, selectivity increases (medians 2.1 and 1.5 respectively) were observed with local returns. In other geometric arrangements, local returns were less effective.

Conclusions : Restricting the electric field with local returns can increase selectivity of stimulation at single-cell resolution. In a future high-resolution prosthesis, this may reduce unwanted neuronal activity and permit a more faithful reproduction of the neural code of the retina.

This is an abstract that was submitted for the 2017 ARVO Annual Meeting, held in Baltimore, MD, May 7-11, 2017.