A machine-learning algorithm correctly classifies cortical evoked potentials from both visual stimulation and electrical stimulation of the optic nerve

Gaillet et al. used the NeuroNexus ECoG array, E32-1000-30-200, to record cortical activity along with a support vector machine classification algorithm to classify cortical responses originating from visual and electrical stimuli. They confirmed an increase in classification accuracy with increased center-to-center separation on patterned visual stimulation. Additionally, their results demonstrated the classification accuracy’s dependence on the current amplitude, with higher accuracy at higher amplitudes. They also used a regression model to add a predictive capacity. Using a regression model, they showed that cortical activities elicited by electrical stimulation are meaningfully different, as it highlights features that vary in a linear manner, which can be expected from cortical activity patterns resulting from the stimulation of a gradually shifted portion of the visual field. These results represent a necessary, although not sufficient, condition for an optic nerve prosthesis to deliver vision with non-overlapping phosphene.

Related Topics