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

Abstract

Objective. Patients 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.Approach. Responses 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 linearnon-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 results. Although 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.Significance. Slower 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.

Correlated Activity in the Degenerate Retina Inhibits Focal Response to Electrical Stimulation

“…A possible physiological explanation of the low vision found in RP patients is a decrease in the signal-to-noise ratio (SNR) due to spontaneous hyperactivity of the RGCs ( Goo et al, 2011b , 2016 ; Euler and Schubert, 2015 ; Ivanova et al, 2016 ; Haselier et al, 2017 ). This induces a less reliable RGC response to repetitive electrical stimulation ( Ho et al, 2020 ; Yoon et al, 2020 ). Therefore, higher stimulation intensities may be required to electrically activate degenerate RGCs than that required for normal RGCs ( Jensen and Rizzo, 2009 ; Goo et al, 2011b ; Cha et al, 2021 ).…”

Differences in the spatial fidelity of evoked and spontaneous signals in the degenerating retina

“…The definition of evoked and spontaneous activity in a spontaneously firing cell maybe somewhat semantic, although it serves to highlight the challenges associated with resolving evoked information carrying visual information, from background activity. The idea that increased spontaneous firing will decrease the signal to noise for restored visual signals has received considerable theoretical and experimental support ( Dräger and Hubel, 1978 ; Sauvé et al, 2001 ; Stasheff, 2008 , 2018 ; Toychiev et al, 2013 ; Yee et al, 2014 ; Barrett, 2015 ; Ivanova et al, 2016 ; Telias et al, 2019 , 2022 ; Ho et al, 2020 ), and to some extent the challenges that exist for experimentally separating evoked and spontaneous spikes will exist for the brain during vision restoration. Our findings show this distinction between evoked and spontaneous activity is important as they have different spatio-temporal properties and are mediated by different circuit mechanisms, and thus may be differentially targeted.…”

“…It is widely accepted that spontaneous activity will negatively impact vision restoration strategies, such as electronic retinal prosthetics or gene therapy mediated optogenetics, because it will reduce the signal to noise ratio of restored prosthetic visual signals ( Dräger and Hubel, 1978 ; Sauvé et al, 2001 ; Stasheff, 2008 , 2018 ; Toychiev et al, 2013 ; Yee et al, 2014 ; Barrett, 2015 ; Barrett et al, 2016 ; Ivanova et al, 2016 ; Telias et al, 2019 , 2022 ; Ho et al, 2020 ). Here, we found an additional functional consequence of the disease mediated circuit changes beyond a simple degradation in the signal to noise.…”

“…Although, much work has focused on the implications for spontaneous activity, fewer studies have examined how evoked responses are altered during degeneration, and results vary widely ( Lorach et al, 2015a ; Stutzki et al, 2016 ; Jalligampala et al, 2017 ; Ho et al, 2020 ; Sekhar et al, 2020 ; Ahn et al, 2022 ). In addition, few have investigated how evoked activity alters the properties of spontaneous activity.…”

Artificial intelligence techniques for retinal prostheses: a comprehensive review and future direction

“…The predicted results of these computational models can match the recorded natural responses well. A recent study implemented a CNN model into a PRIMA [161]. In the degenerate retina, the response stimulated by multiple electrical arrays (MEAs) of the implant was unable to fit with natural responses.…”