Development of a Closed-loop Approach for Automatically Adjusting Thresholds in Cortical Visual Prostheses

FABRIZIO GRANI; CRISTINA SOTO-SANCHEZ; J SUAREZ; P GONZALEZ; MARÍA DOLORES GRIMA-MURCIA; FERNANDO DANIEL FARFÁN; LEILI SOO; ALFONSO RODIL; A ALFARO; EDUARDO FERNANDEZ

Detroit

Congreso; The Eye and the Chip - Virtual Event 12th World Research Congress; 2021

Detroit Institute of Ophthalmology

Purpose: Cortical visual prostheses are a potential method to restore vision in profoundly blind people. An effective cortical prosthesis may include thousands of electrodes and prospective users need to adjust the thresholds individually for each single electrode. This process is time consuming, and users might experience fatigue during the determination of thresholds. Our main goal is the development of an automated procedure, based on neural recordings, to adjust stimulation thresholds without the need of patient interaction. Method: We implanted an intracortical microelectrode array (consisting of 96 electrodes) in the visual cortex of a 57-year-old adult with complete blindness. Standard psychometric experiments with different stimulus amplitudes, ranging from 1 to 160 A, were performed with simultaneous electrophysiological recordings during electrical stimulation for a 6-month period. After removing the electrical artifacts recorded on the electrode array, we extracted the sum of 4-28 Hz spectra components of the recordings during the stimulation and analyzed how these spectra changed with stimulation intensity obtaining a curve that we compared with the standard psychophysics curves based on subject answers. We assumed as perception threshold the stimulation intensity at which the subject reported to see a phosphene 50% of the time, and we compared this intensity, to the stimulation intensity at which the 4-28 Hz spectra curve reached half of his maximum. Results: The normalized 4-28 Hz spectra correlates with the standard psychometric curves obtained with the subject answers (mean r=0.85). The automated thresholds obtained with the proposed closed-loop approach, without the need of subject interactions, correlated well (r=0.64, p=0.001) with perception thresholds obtained with the subject answers. Conclusion: Our results showed that neural recordings provide direct insight into the efficacy of stimulation. Although more studies are still needed, our results suggest that the 4-28 Hz signal spectra contains information related to perception and can be used to automatically estimate current thresholds