Axon activation with focal epiretinal stimulation in primate retina

Abstract

Purpose: Epiretinal prostheses aim to mimic natural vision by direct stimulation of retinal ganglion cells (RGCs). Selective activation of individual RGCs of different types is necessary to accurately recreate the activity that occurs during normal vision. Our experiments indicate that in some cases, activation at single-cell single-spike resolution is possible, suggesting high potential for future prostheses. However, these data also reveal unwanted activation of RGC axons, from distant areas of the visual field, consistent with the phosphenes produced in recent clinical studies. The purpose of the present work is to develop approaches to selectively activate of RGCs without simultaneously activating axons.

Methods: A custom 512-electrode stimulation and recording system was used to measure electrical activation in hundreds of RGCs from isolated macaque retina. Stimulation current was provided through one or more electrodes while RGC spikes were simultaneously recorded on all electrodes. Electrodes had diameters of 8-15 μm and were separated by 60 μm. Cell types and axon locations of RGCs were identified using their recorded responses to white noise visual stimuli.

Results: At stimulation current levels similar to those that produced somatic RGC activation, bidirectional signal propagation, apparently attributable to axon bundle activation, was observed originating from the vicinity of the stimulating electrode. To investigate if spatially patterned electrical stimulation can reduce unwanted axon activation, responses of RGCs stimulated with various individual electrodes and pairs of electrodes near their somas and along their axons were recorded. Preliminary results show that in many cases, testing multiple spatial stimulation patterns for a given target RGC revealed a particular pattern that activated the cell at its soma without activating the axon bundle. However, significant variability was observed in the efficacy of this approach between retinal preparations. Ongoing studies aim to determine the conditions in which spatial patterns, with two or more electrodes, may provide an improvement in specific somatic activation.

Conclusions: The use of spatial patterns of electrical stimulation can, in some cases, increase the odds of evoking somatic activation without axon activation. However, variability in the results points to the need for a more systematic approach to spatially patterned stimulation and axon avoidance.