Eye implants that help blind patients read win approval in Europe - critical summary review

A retinal implant is a chip smaller than a grain of rice, placed surgically at the back of the eye where disease has killed the cells that normally sense light. Paired with a tiny camera built into a pair of glasses, it converts what the camera sees into electrical pulses that the eye's surviving wiring can still carry to the brain, standing in for the part of the eye that stopped working.

For two decades that described an experiment. In July it started describing a product: Europe's medical device regulator cleared the sale of one such implant, called PRIMA, made by the Californian company Science Corporation. The device is now authorised in 30 European countries, and the company has announced the start of commercial sales, with the first implants expected in Germany. Science Corporation says it is the first brain-computer interface device to win that certification for restoring functional central vision.

The approval landed in the same stretch of months in which a rival approach, which skips the eye altogether and wires a camera into the brain, picked up a regulatory label in the United States and a wave of scepticism from independent researchers.

How artificial vision actually works

In a healthy eye, light reaches the retina at the back, where cells called photoreceptors turn it into electrical signals. Those signals travel along the optic nerve to the brain, which reads them as images. Several diseases destroy the photoreceptors while leaving the rest of that chain intact, and that gap is what these devices try to fill.

PRIMA's chip is two millimetres square, about as thick as a human hair, and sits under the macula, the central part of the retina responsible for fine detail. It has no wires and no battery. The glasses capture the scene, a pocket processor sharpens it, and the image is projected onto the chip as near-infrared light. The chip works like a tiny solar panel, converting that light into electrical pulses that stimulate the retinal cells still alive underneath.

That condition is the whole design. "If all cells from the retina were dead, the chip wouldn't work at all," Frank Holz, head of ophthalmology at the University of Bonn and the trial's lead, told the German broadcaster DW. "It just replaces the photoreceptors."

A second family of devices bypasses the eye and stimulates the visual cortex at the back of the head directly. Those are aimed at people whose optic nerve no longer works, and they remain confined to small studies.

The chip that let patients read again

The trial behind the European approval, published in the New England Journal of Medicine, followed 38 people with geographic atrophy, an advanced form of dry age-related macular degeneration that erases central vision while leaving a blurred rim of peripheral sight. Of the 32 who completed a year, 27 could read again using central vision, gaining on average five lines on a standard eye chart.

"This is the first implant that's been demonstrated to give patients meaningful vision that they can use in their daily life, such as reading, writing," Mahi Muqit, the ophthalmic surgeon who led the British arm of the trial at Moorfields Eye Hospital in London, told the BBC. One of his patients, Sheila Irvine, 70, who is registered blind, said being able to read her post and do crosswords again was "out of this world". Max Hodak, Science Corporation's chief executive, told TechCrunch that a patient in France finished a 300-page novel and posted him the book.

What patients get is narrow. Hodak has described the implant as "like looking through a straw": black and white, sharp but tiny, a patch in the middle of the visual field. Learning to use it takes work, and the rehabilitation programme currently runs for about twelve months.

In the United States the device has only Breakthrough Device and Humanitarian Use Device designations from the Food and Drug Administration, labels that speed up review without amounting to approval. The company has also begun testing the implant in Stargardt disease, an inherited condition that destroys central vision and has no cure, with the first patient implanted in May at Sydney Eye Hospital.

A graveyard of bionic eyes

The field is littered with devices that worked and then vanished. The Argus II, made by Second Sight, cleared the FDA in 2013 for people blinded by retinitis pigmentosa, a genetic disease that kills light-sensing cells. It used camera glasses and a 60-electrode array laid on the retina, and produced an image of roughly 36 dots: shapes, outlines, no colour. It cost around 150,000 dollars. The company discontinued it in 2019 and stopped technical support, leaving patients with implants nobody maintains. "As long as nothing goes wrong, I'm fine," Terry Byland, who has the device in both eyes, told IEEE Spectrum. "The problem is if something goes wrong, because apparently there is no way to repair them."

The German firm Retina Implant AG took the other route, a light-sensitive chip under the retina, but its surgery lasted about eight hours and included cutting a cavity in the skull behind the ear for a power receiver. Patients saw flashes roughly matching the shape of what they looked at, and implants often broke over time. The company closed in 2019, citing a lack of medical progress.

PRIMA nearly joined them. Pixium Vision, the French company that developed it, went out of business in 2024 while the trial was still running. "I remember vividly the days not so long ago, when everything was just about to die at the time when the study was almost complete," José-Alain Sahel, who helped develop the system, told DW. Science Corporation bought the assets.

Writing in The Conversation, Tim Jackson, a retinal research professor at King's College London, notes why this one may hold: age-related macular degeneration is far more common than the inherited diseases earlier implants targeted, which makes the device far more attractive to investors. He also flags the ethical cost. Unlike patients with inherited retinal disease, most people with macular degeneration still have usable peripheral vision, so they have a lot to lose, and the surgery is far from risk-free.

The brain-implant promise, and the scientists who dispute it

Neuralink's device for blindness, Blindsight, received the FDA's breakthrough device designation in September 2024. Elon Musk said on X that it would let even people who have lost both eyes and the optic nerve see, that resolution would start low, "like Atari graphics", and that over time it had the potential to be better than natural vision.

The FDA told Estadão that the designation evaluates only whether a technology might offer a more effective treatment than current standards, that it does not assure safety or efficacy, and that it does not allow the device to be sold. Blindsight has not begun human trials, and the company has given no timetable.

The first results Neuralink has shown came at a conference in June 2025, where engineer Joseph O'Doherty reported that a monkey moved its eyes towards a target researchers were trying to make its brain perceive, at least two-thirds of the time.

Researchers at the University of Washington ran simulations of what such implants can produce and published them in Scientific Reports. "Engineers often think of electrodes as producing pixels. But that's not how biology works," said Ione Fine, the lead author. In their model, an image built from 45,000 pixels is crisp; the same image simulated through 45,000 electrodes in the visual cortex is a blur. Fine was blunt about the marketing: "When Elon Musk says things like, 'This is going to be better than human vision,' that's a dangerous thing to say."

Eduardo Fernández, the Spanish neuroscientist running one of the few cortical implant studies in humans, puts the ceiling plainly: artificial vision is "not like seeing again". His stated goal is better orientation and mobility.

What these devices still cannot do

Eligibility is set by diagnosis, not by blindness. PRIMA was approved for one condition, and the trials are not expected to help people whose optic nerve has stopped working. It only addresses loss of photoreceptors.

Nor is it a cure. In the trial, 26 serious adverse events occurred in 19 participants, almost all raised pressure inside the eye and small haemorrhages in the first eight weeks after surgery; none caused total vision loss. The study was small, ran for twelve months and had no control group. "For the first time, a large clinical study showed that an implant can restore part of the functional central vision of patients with geographic atrophy," said Ricardo Japiassu, a director of the Brazilian Society of Ophthalmology, who added that the device does not return normal vision and that durability, long-term safety and performance in larger numbers still need confirming.

Price is unresolved. Each device is expected to cost in the hundreds of thousands of dollars, and Science Corporation is negotiating reimbursement with European health systems. "Even when they work, they tend to be expensive and to require specialised, complex infrastructure," the ophthalmologist Claudio Lottenberg told Veja. There is no announced timetable for Brazil or the rest of Latin America.

What to do with this information

Two markers are worth more than the next round of headlines. The first is whether the FDA grants PRIMA full approval in the United States, where it currently holds only the two accelerated-review designations. The second is whether Neuralink or any other brain-implant maker publishes peer-reviewed results from human trials, rather than monkey experiments and statements from executives. Those are different kinds of evidence, and only one of them is what regulators act on.

The other detail is the one that decides everything for an individual. These devices are not approved for blindness in general. They are approved for named diseases, and so far for one: geographic atrophy, with Stargardt disease in trial and retinitis pigmentosa the target of a generation of implants that no longer exists. If sight loss has touched someone close to you, do you know which of these diagnoses it is, and whether any approved device was ever meant for it?