Solar cells may sub for retinal receptors.

Researchers in Germany are developing an eye implant, based on a solar cell, for people who have lost their sight because of a disease of the retina.

The implant, just 1.5 micrometers thick and about 3 millimeters in diameter, "is not giant in size or market value, but giant in its goals," says Markus Schubert of the University of Stuttgart. He described the device last week in San Francisco at a meeting of the Materials Research Society.

Schubert and his colleagues are testing how animals tolerate the implant. A host of technical issues must be resolved before the device can be used in people.

Several groups of researchers are investigating retinal implants as a way of restoring partial sight to people with retinitis pigmentosa. This disease causes the light-sensitive rod and cone cells in the retina to waste away, producing tunnel vision or total blindness. About 20,000 people in the United States are blind because of retinitis pigmentosa.

The German team's implant consists of a thin layer of amorphous silicon deposited on a flexible titanium film and etched with an array of light-sensitive elements that turn light into electric impulses. Instead of converting light into electric power, however, the implant creates signals that are picked up by nerve cells in the retina. Those impulses then travel via the optic nerve to the brain.

Unlike other groups, the German researchers are designing their device to be implanted within the retina, where the rods and cones normally reside, rather than on its front surface. "It's a more direct approach," Schubert says. Their implant delivers signals to the first links in the chain of nerve connections leading from the eye to the brain and makes use of retinal nerve cells' ability to process those signals.

Implants on the surface of the retina, in contrast, convey information to nerve cells that are nearby but further along in the chain. This approach requires complicated preprocessing of the data, he says.

Mark S. Humayun of Johns Hopkins University in Baltimore argues that implants on the retinal surface do not necessarily call for more data processing and would be easier to implant and remove. Recent experiments by his group suggest that the form of the electric impulses, rather than physical proximity, determines which nerve cells are stimulated. Retinal surface signals could therefore trigger the earliest steps of visual processing.

"It's like tuning in a radio station," he explains. "It depends on the frequency, not how close the radio is to where the station is broadcast."