AI-powered bionic eye approach outperforms standard methods in blind participant

Researchers have demonstrated that artificial intelligence can make future visual cortical prostheses — devices sometimes called "bionic eyes" — more precise and adaptive, according to a proof-of-concept study published August 7 in the journal Neuron.eurekalert+1

AI-Designed Stimulation Outperforms Conventional Methods

A team from UC Santa Barbara, ETH Zurich, and Miguel Hernández University trained a deep neural network to predict how a blind participant's brain would respond to different electrical stimulation patterns delivered through a 96-channel electrode array implanted in the visual cortex. The model then worked in reverse, identifying the stimulation settings most likely to produce a desired pattern of neural activity.bioengineer+1

When tested in the participant — a 27-year-old man who lost his vision following a traumatic brain injury — the AI-designed patterns reproduced targeted brain activity more accurately while requiring less electrical current than conventional approaches. The researchers also found that recorded neural activity was more informative about what the participant perceived than the stimulation settings alone, underscoring that what the brain does with a signal matters as much as the signal itself.eurekalert+1

"Engineers naturally want to treat phosphenes like pixels: stimulate more electrodes, and you should get a more complete image," said Michael Beyeler, associate professor of computer science at UC Santa Barbara and a senior author on the study. "But the brain does not work that way."eurekalert

From Fixed Recipes to Adaptive Devices

The implant, placed at Hospital IMED Elche in Spain as part of a broader feasibility trial, could both deliver electrical current and record the brain's response — a capability that allowed the deep-learning model to train on actual neural data rather than relying solely on predetermined stimulation rules. The model also incorporated measurements of resting brain activity before each stimulation event, enabling it to adjust for fluctuations in neural excitability from moment to moment.bioengineer+1

The participant's perceptions remained limited to phosphenes — spots, flashes, or shapes of light — rather than detailed images. But the framework points toward a future in which prostheses continuously learn how an individual brain responds and modify stimulation in real time.news.ucsb+1

A Path Toward Personalized Prostheses

The study builds on computational work supported by Beyeler's 2022 National Institutes of Health Director's New Innovator Award. The project was led by co-first authors Pehuén Moure of ETH Zurich, Jacob Granley of UC Santa Barbara, and Fabrizio Grani of Miguel Hernández University.eurekalert

Cortical prostheses bypass the eyes and optic nerves entirely, potentially benefiting people whose blindness resulted from strokes, neurodegenerative diseases, or brain injuries. "A useful visual prosthesis cannot rely on a fixed recipe," Beyeler said. "It has to learn how an individual brain responds and adapt the stimulation accordingly. Ultimately, the device should adapt to the person, not the other way around."eurekalert