Predicting the Temporal Dynamics of Prosthetic Vision

TL;DR

This work tackles the problem of forecasting phosphene temporal dynamics in retinal prostheses by presenting two predictive frameworks: a Spectral model using a truncated Fourier representation and an Exponential model with multi-interval decays, both evaluated against real patient data from nine Argus II users. The Spectral model, which decomposes fading and persistence into spectral components, demonstrates strong generalization across subjects and stimulus conditions (up to $r ightarrow 0.88$ in some cases) and remains competitive for unseen stimuli, outperforming a Baseline two-exponential approach. The study advances prosthetic vision by providing mechanistic, data-validated timing predictions that can inform stimulation strategies, while also highlighting data scarcity as a key limitation and suggesting that 1–2 spectral components often suffice for generalization. Taken together, these models lay groundwork for more accurate, temporally aware phosphene simulations that could improve real-world usability of retinal implants.

Abstract

Retinal implants are a promising treatment option for degenerative retinal disease. While numerous models have been developed to simulate the appearance of elicited visual percepts ("phosphenes"), these models often either focus solely on spatial characteristics or inadequately capture the complex temporal dynamics observed in clinical trials, which vary heavily across implant technologies, subjects, and stimulus conditions. Here we introduce two computational models designed to accurately predict phosphene fading and persistence under varying stimulus conditions, cross-validated on behavioral data reported by nine users of the Argus II Retinal Prosthesis System. Both models segment the time course of phosphene perception into discrete intervals, decomposing phosphene fading and persistence into either sinusoidal or exponential components. Our spectral model demonstrates state-of-the-art predictions of phosphene intensity over time (r = 0.7 across all participants). Overall, this study lays the groundwork for enhancing prosthetic vision by improving our understanding of phosphene temporal dynamics.

Figures (2)

• Figure 1: Model fits for representative time courses, each elicited by distinct electrical stimuli (see Table II for details). Subjects' perceived brightness levels (gray; indicated by joystick position perez_fornos_temporal_2012) are depicted alongside predictions from three models: spectral (black), exponential (blue), and baseline avraham_retinal_2021 (orange).
• Figure 2: Training (gray) and validation (black) MSE $\pm$ Standard Error of the spectral model as a function of the number of spectral components ($m$), presented for each subject and averaged across stimulus conditions. The minimum training MSE is highlighted in blue, and validation in red.