AI models could help blind people regain object vision with prosthetics

Researchers at the Swiss Federal Institute of Technology Lausanne (EPFL) are pioneering advancements in artificial intelligence that hold the promise of restoring a significant level of vision to blind individuals through sophisticated visual prosthetics. The NeuroAI Lab at EPFL has created AI models capable of predicting precise brain stimulation points to "evoke" clear images of objects and faces.

These AI models are designed to enhance existing cortical prosthetics. Current devices, used when the retina or optic nerve is damaged, stimulate the visual cortex directly. However, they typically produce only rudimentary flashes of light and simple shapes. The EPFL team's work aims to stimulate higher brain regions, enabling the generation of more detailed visual representations.

While the AI models have shown success in predicting object perception based on visual stimuli in experiments with monkeys, they cannot yet generate images entirely from scratch. The researchers believe this capability could be the next crucial step towards achieving meaningful vision restoration. Beyond visual prosthetics, EPFL suggests these AI advancements might also find applications in improving auditory prosthetics, like cochlear implants, which are more advanced than current visual aids but still have room for enhancement.

“

evoke