Generating Natural Neural Response in Visual Neuroprosthetics Using the Artificial Bee Colony Algorithm

A significant challenge that hampered visual prosthetic systems is its ability to produce a meaningful perception of vision. It is hypothesised that mimicking the natural spiking pattern of the visual cortex would lead to a more natural perception of vision. In this study, an optimisation algorithm, namely the Artificial Bee Colony (ABC) algorithm, is utilised to automatically generate a stimulus waveform to produce a response that best matches a target neural pattern. An in silico model of realistic retinal ganglion cells (RGC) was used to simulate neural responses in a visual prosthesis and assess the performance of the ABC algorithm. The ABC algorithm could generate a stimulus waveform that produces target neural responses with 91±7% accuracy compared to the target within 258±83 cycles of iteration.Clinical Relevance- ABC algorithm can be used to facilitate a stimulus waveform to trigger natural spiking patterns from the retinal cells. This method could potentially be implemented together with artificial intelligence (AI) techniques to improve the quality of vision provided by visual prosthetic systems.