Optimizing retinal prosthesis stimulation: Enhancing the avoidance of unintended axonal activation

By electrically stimulating surviving retinal neurons, retinal prostheses improve vision for individuals with outer retinal degeneration. Epiretinal prostheses, commonly used commercially, mainly target retinal ganglion cells (RGCs). However, current retinal prosthetic devices often lead to elongated phosphenes instead of a focal spot of light, affecting visual clarity due to the unwanted activation of axon fibers. In this study, we developed a computational model of morphologically realistic RGCs to study their response to epiretinal electrical stimulation. While previous studies have often investigated pulse parameters in isolation, our model systematically quantified the critical, synergistic interaction between pulse duration, waveform, and electrode-to-cell distance on the prevention of unwanted passing axon fiber activation. Four rectangular pulse waveforms with two durations (50 μs and 0.5 ms) were tested at various distances from the stimulating electrode to the RGC. Our findings reveal that short anodic-first biphasic (BA) pulses proved most effective at avoiding unwanted passing axon fiber activation across all distances, showing more than a 2.5-fold difference in thresholds between the axon initial segment (AIS) and the axon as distance increased. This represents a significant improvement in selectivity over commonly used cathodic-first waveforms. Short cathodic-first biphasic (BC) pulses became viable at larger distances. Furthermore, we demonstrate that increasing the electrode-retina distance, contrary to some expectations, can enhance focal activation when paired with an optimized waveform. Notably, electrode-RGC distance does not directly avoid unwanted passing axon fiber activation; its effectiveness relies on the pulse waveform shape. In developing the stimulation strategy, it is crucial to consider how pulse duration, pulse waveform, and the electrode-RGC distance interact, as these factors are intricately linked.