Design of a High–Resolution Optoelectronic Retinal Prosthesis

Abstract

Abstract: : Purpose: To develop a system for electrical stimulation of the retina that would produce functional levels of sight. Design of such prosthesis presents several major challenges: (a) To activate thousands of microelectrodes while avoiding metal erosion, cross–talk between pixels and overheating of tissue, the electrodes need to be located very close to the target cells; (b) To allow for natural scanning of images, information delivered to the implant should be coupled to eye movements; (c) To adjust the stimulation map to retinal architecture, real–time location–dependent image processing will be required. Methods: Methodology that underlies our design includes: (a) Achievement of neural proximity by promoting the migration of retinal cells into perforated membranes and protruding electrode arrays. (b) Delivery of processed images from a video camera by projection of a goggle–mounted display onto the retina, in order to activate an array of powered photodiodes in the retinal implant. (c) Adjustment of the projected images to retinal architecture using a tracking system that monitors the direction of gaze in real time. Results: Our optoelectronic subretinal implant is designed for pixel densities up to 2500 pix/mm² (pixel size of 20 µm) which would allow functional levels of visual acuity. Each pixel has a photodiode connected to a common bi–phasic pulsed power line and thus converts local light intensity into a biphasic pulse of current. Intimate proximity between electrodes and cells (a few µm) is achieved using microfabricated arrays of either chamber electrodes with apertures 10–15 µm in width and 20–50 µm in depth or pillar electrodes 10 µm in diameter and 50 – 70 µm in length. The processed image from the video camera is displayed with pulsed infrared collimated LEDs through an LCD screen and projected onto the retinal implant. A tracking system mounted on the same goggles monitors the position of the retinal implant and adjusts the image processing between the camera and the IR goggles at a rate of 30 frames/s. Conclusions: We present a design for a retinal prosthesis that would solve some of the major problems associated with the realization of a functional retinal stimulation: high pixel density, proximity of electrodes to target cells, natural eye scanning capability, and real–time image processing adjustable to retinal architecture.

Keywords: retinal connections, networks, circuitry • image processing • eye movements