Decoding network-mediated retinal response to electrical stimulation: implications for fidelity of prosthetic vision

Abstract

AbstractObjectivePatients with the photovoltaic subretinal implant PRIMA demonstrated letter acuity by ~0.1 logMAR worse than the sampling limit for 100μm pixels (1.3 logMAR) and performed slower than healthy subjects, which exceeded the sampling limit at equivalently pixelated images by ~0.2 logMAR. To explore the underlying differences between the natural and prosthetic vision, we compare the fidelity of the retinal response to visual and subretinal electrical stimulation through single-cell modeling and ensemble decoding.ApproachResponses of the retinal ganglion cells (RGC) to optical or electrical (1mm diameter arrays, 75μm pixels) white noise stimulation in healthy and degenerate rat retinas were recorded via MEA. Each RGC was fit with linear-non-linear (LN) and convolutional neural network (CNN) models. To characterize RGC noise level, we compared statistics of the spike-triggered average (STA) in RGCs responding to electrical or visual stimulation of healthy and degenerate retinas. At the population level, we constructed a linear decoder to determine the certainty with which the ensemble of RGCs can support the N-way discrimination tasks.Main resultsAlthough LN and CNN models can match the natural visual responses pretty well (correlation ~0.6), they fit significantly worse to spike timings elicited by electrical stimulation of the healthy retina (correlation ~0.15). In the degenerate retina, response to electrical stimulation is equally bad. The signal-to-noise ratio of electrical STAs in degenerate retinas matched that of the natural responses when 78±6.5% of the spikes were replaced with random timing. However, the noise in RGC responses contributed minimally to errors in the ensemble decoding. The determining factor in accuracy of decoding was the number of responding cells. To compensate for fewer responding cells under electrical stimulation than in natural vision, larger number of presentations of the same stimulus are required to deliver sufficient information for image decoding.SignificanceSlower than natural pattern identification by patients with the PRIMA implant may be explained by the lower number of electrically activated cells than in natural vision, which is compensated by a larger number of the stimulus presentations.