The Five Senses: Sight

August 30, 2017

Brain Implants May Be the Answer to Restoring Sight

Ken Shepard in a black and white striped polo shirt stands in front of his computer and next to a microscope on his desk.
Photo by Eileen Barroso


Retinal implants to treat blindness are now on the market, but even this cutting edge technology offers only vague impressions of objects and can only treat certain forms of blindness. Ken Shepard wants to do better.

By targeting the brain itself, Shepard hopes to both sharpen the picture quality and reach patients who have lost their sight due to retinal degeneration and damage to the optic nerve that connects the eye to the brain. His work, funded by the U.S. Defense Advanced Research Projects Agency, or DARPA, is part of a growing push in academia and industry to develop brain-computer interfaces to repair senses and skills lost to injury or disease.

An electrical engineering professor, Shepard develops electronics that interact with living systems. After spending his early career at IBM designing computer chips, he came to Columbia in 1997 and discovered a growing demand among biologists for new tools to study molecules and cells. He realized that modern microelectronics could help.

Explore the Five Senses
Ear: To HearEye: To SeeHand: To TouchMouth: To TasteNose: To Smell

“I looked at all the fun stuff going on, and realized that at Columbia I had the opportunity to explore problems I couldn’t pursue in industry,” he said from his Bioelectronics Systems Lab in the Northwest Corner Building.

His work in molecular diagnostics and microbiology eventually led to studying the brain. The $16 million DARPA project, announced in July, to develop a working brain implant to treat blindness, is Shepard’s latest collaboration with Columbia neuroscientist Rafael Yuste and follows several projects that involve applying modern electronics to neuroscience.

In four years, Shepard and his colleagues aim to have a working implant that can partially restore vision to people with retinal and optic nerve damage. A two-way wireless device, as thin as a piece of tissue paper, would rest on the surface of the brain, translating sights from the outside world into electrical signals that stimulate specific neurons to let the brain “see.”

So far the only device approved by the U.S. Food and Drug Administration to treat blindness is the Argus II, a retinal prosthetic made by California-based Second Sight. A camera mounted on a pair of eyeglasses sends visual inputs to a chip implanted on the retina, which translates the signals into electrical pulses that stimulate clusters of neurons. The brain learns to interpret the resulting bursts of light, or phosphenes, to make out nearby objects.

The brain implant Shepard and Yuste are developing for DARPA, with colleagues at Baylor College of Medicine, Caltech, Duke and NYU, would instead stimulate the visual cortex. Up to a million electrodes would target individual neurons, offering potentially much higher image resolution.

Shepard’s lab looks like a scene from a science fiction movie. A mouse retina kept alive in a nutrient-rich solution sits on a chip the size of a quarter, translating visual inputs from the world into neural responses which are read by the chip’s 65,000 electrode-sensors.

By analyzing the firing patterns of neurons across the retina, the Columbia researchers have mapped as much as 75 percent of neurons in the inner retina, a 10-fold increase over previous attempts. “We’ve observed neural networks at a level of detail that’s previously been impossible,” said David Tsai, a biomedical engineer at Columbia and a lead researcher on the DARPA project. This close-up view has revealed how retinal neurons communicate with the brain, and how specific neurons might be stimulated for therapeutic purposes.

The DARPA funding will allow Shepard and his team to adapt this 65,000-electrode array to the human brain, greatly expanding its recording capability. Shaving one side of the chip made it flexible enough to drape over the brain. To prevent it from overheating the brain tissue, the team will scale down its energy use to less than 5 percent of what a cellphone consumes.

A second device worn on the head would wirelessly power the chip and transfer data between the implant and a nearby computer or cellphone. The first experiments in humans are planned at the end of the DARPA program, pending FDA approval.

Advances in hardware and software have paved the way for a new era of research on brain-computer interfaces to augment human intelligence and restore lost capabilities. DARPA, the research arm of the military that helped develop the internet, has bet on Shepard’s team and five others in its goal to build a working brain-machine interface that can record one million of the brain’s 86 billion neurons.

“What we’re trying to do is extraordinarily ambitious, but I really think we can build this,” said Shepard. “It doesn’t violate any laws of physics. This is absolutely doable from where we are now.”

—By Kim Martineau