SAN DIEGO—Tens upon tens of thousands of people in the U.S. use cochlear implants, a bionic ear-like device that can restore hearing to the profoundly hearing impaired. Biomedical technologists have labored for decades to create the visual equivalent, a working prosthesis to restore sight to those with damaged retinas. Thus far, these efforts have met with limited success.
Here at this week’s Society for Neuroscience annual meeting, Sheila Nirenberg of Weill Cornell Medical College presented a novel approach that appears to improve on previous labors by fashioning a device that mimics the brain’s innate software. The prosthesis replaces retinal circuits by processing light received and then outputting a code—technically, action potentials—that are relayed internally to the processing centers in the brain that construct a conscious image.
Previous technology produced simpler images, Nirenberg said at a press briefing. (The pictures above show natural vision on the left, an old-style artificial retinal image on the right and different stages of processing Nirenberg’s coded images in the two middle pictures.) Nirenberg and her student, Chethan Pandarinath, hail from the neural coding field, which brings a new way of approaching the problem. The standard focus has been on improving prosthetics by increasing the number of electrical nerve stimulators that are placed in the patient’s retina. Nirenberg’s group has been focusing on how to turn the stimulators on and off in the right pattern, so that the retina sends normal signals to the brain. Their device mimics the retina’s software to replicate the signalling used to produce an image in the brain.
The team built a device to calculate the requisite code from the light pattern captured from an eye-tethered camera that is part of the prosthesis. The encoder replicates a portion of the retinal circuitry and was able to produce images in blind mice that had a fidelity that approached that of an actual eye. “There was enough information in the system’s output to construct images, of faces, animals, scenes of Central Park, basically anything we attempted, she said.