Investigating Preferential Activation of Rat Retinal Ganglion Cell Classes with Electrical Stimulation

Abstract

Purpose : Retinal prostheses (bionic eyes), which aim to return vision to the blind through electrical stimulation of retinal ganglion cells (RGCs) currently present the only available clinical treatment for retinal degenerative diseases such asretinitis pigmentosa. However, the image perceived by the patients is still of poor quality. This limitation partially stems from the inability to selectively activate specific RGCs with electrical stimulation. In most mammalian retinas, there are 10-15 classes of RGCs with distinct morphological and physiological properties, which form distinct processing streams in normal vision. We aim to obtain the preferred electrical stimulation waveform of each morphological class of rat RGCs and use it to preferentially activate that class.

Methods : Firstly, we used intracellular Gaussian white noise electrical stimulation and spike triggered average analysis to recover the preferred stimulation waveforms of individual rat RGCs. We then clustered the preferred waveforms of RGCs using a K-means clustering algorithm and correlated the clusters with RGC morphological classes. Secondly, we stimulated a number of RGCs with their own preferred waveforms and the waveforms recovered from other RGCs to test the efficacy of each waveform on each other.

Results : We recovered the preferred stimulation waveforms of 55 individual rat RGCs. The waveforms were classified into four different categories. Interestingly, the waveform clusters correlated accurately with the RGCs morphological classes A, B, C and D. As a result, we have developed an alternative method for identifying RGCs that eliminates the need to recover their morphology. In addition, a number of RGCs were stimulated with their own preferred waveforms and the waveforms recovered from other RGCs. We found a preferential stimulation of almost every cell to its own preferred waveform. Furthermore, many RGCs show preferential responses to waveforms of other RGCs that are from the same waveform cluster.

Conclusions : Our findings show that RGCs in rat retina are selective to specific features of electrical stimuli. Hence we could improve the efficacy of retinal prostheses by developing stimulation strategies that target specific RGC classes critical for extracting the attributes of a visual scene.

This abstract was presented at the 2019 ARVO Annual Meeting, held in Vancouver, Canada, April 28 - May 2, 2019.