Retinal prosthesis · Grey Matter

A retinal prosthesis is an implant that restores some sight to people whose photoreceptors have died, by stimulating the retinal cells that survive and leaving the brain to interpret the result, as a cochlear implant does for hearing.

Retinal prosthesis. A retinal prosthesis is an implant that restores some sight to people whose photoreceptors have died, by stimulating the retinal cells that survive and leaving the brain to interpret the result, as a cochlear implant does for hearing.

It needs a retina whose output still works. In retinitis pigmentosa and in the geographic atrophy of age-related macular degeneration, the rods and cones die while the bipolar and ganglion cells above them, and the optic nerve, mostly remain. A camera on a pair of glasses captures the scene and the implant turns it into a pattern of stimulation; what the person perceives are spots of light (phosphenes) that the visual cortex has to assemble into shapes. Two designs have reached patients. Epiretinal devices such as Argus II sit on the inner surface and stimulate ganglion cells through a grid of 60 electrodes. Subretinal devices sit where the photoreceptors were: PRIMA, a 2 mm square of silicon with 378 pixels of 100 µm, turns near-infrared light projected by the glasses into current for the bipolar cells.

Resolution is the limit. Sixty electrodes give a picture of sixty dots at best, and current spread blurs it further, so Argus II served mostly to find large, high-contrast things such as a doorway or a window.

The subretinal route reached reading. In the PRIMA trial of 38 people with geographic atrophy, 26 of the 32 assessed at a year improved their acuity by at least two lines, on average by about five.

The device can outlive its maker. Second Sight stopped producing Argus II around 2020 and nearly shut down, leaving more than 350 implanted people without repairs or support, the standing case for long-term duties in informed consent.

PRIMA changed hands before it reached the market. Developed at Stanford and tested by Pixium Vision, it was bought by Science Corporation when Pixium was liquidated in 2024, and received its CE mark in July 2026.

A retinal prosthesis borrows the retina's own wiring.

The implant only replaces the light sensors, so the retinal circuits, the optic nerve and the cortex still do their usual work on what it writes.

Questions: Could light-sensitive proteins replace a retinal implant? Optogenetics tries exactly that: a gene therapy makes the surviving ganglion cells of the retina respond to light, so no chip is needed. In 2021 a man blind from retinitis pigmentosa, treated this way and wearing goggles that projected intensified light onto his retina, could perceive, locate and count some objects on a table. Both approaches still depend on glasses that turn the scene into a strong signal, and for now the implant gives the sharper result: PRIMA patients gained on average about five lines on a reading chart. How much can a person see through a retinal implant with sixty electrodes? At best a grid of sixty spots of light, and in practice less, because current from an electrode on the retina's surface also excites the axons passing beneath it, so many spots are drawn out into streaks. In the Argus II trial nearly everyone could find a bright square on a screen with the system on, more than half did better at telling the direction of a moving bar, and the best acuity measured was 20/1260. That is enough to find a door, a window or a line on the floor, and not enough to recognise a face. What does a retinal implant need from the eye, and what is left when the eye cannot help? A retinal implant replaces only the dead photoreceptors, so it needs living bipolar or ganglion cells and an intact optic nerve to carry its signal to the visual cortex. That is why it suits retinitis pigmentosa and macular degeneration and cannot help after glaucoma, optic nerve damage or the loss of the eye. For those cases the alternative is to skip the eye and stimulate the visual cortex directly: in 2021 a 96-electrode array in the occipital cortex of a woman blind for 16 years produced points of light that let her make out some letters. Why did a brain-computer interface company reach patients first with a retinal implant? The retina is part of the brain that a surgeon can reach through the eye, and in macular degeneration its output cells and the optic nerve still work after the photoreceptors die. A chip under the retina only has to replace the light sensors, and the brain's own visual wiring does the rest. Science Corporation also bought a device already deep in testing: PRIMA, designed at Stanford, came with Pixium Vision's trial when Pixium was liquidated in 2024, and its results in the New England Journal of Medicine in 2025 led to the CE mark in July 2026.