A collaborative effort to design an intracortical visual prosthesis has progressed to the point that tests of a prototype in a human volunteer are being planned. The results of psychophysical testing in normal volunteers indicate that the prosthesis could provide sufficient visual functionality to enable users to perform useful visual tasks.
Now, after many years of tackling technologic barriers, funding problems, and other issues, a multi-institutional team is nearly ready to test a prototype system in a human volunteer, according to Philip R. Troyk, PhD, associate professor, biomedical engineering, Illinois Institute of Technology (IIT) in Chicago. Dr. Troyk leads the neuroprosthetic research program at IIT.
"We're certainly right now creating a road map for how we would test this in a volunteer," he said. "This is the first time that we have thought that there's enough available technology that has converged to a known state where it makes sense actually to go ahead and create a plan to try this in a volunteer."
Dr. Troyk presented a portion of the team's recent work, describing psychophysical testing in normal humans.
"This was done in order to come as close as we could to seeing if, with a presentation of visual percepts called phosphenes, they could or could not use that to do some useful sensory tasks," he explained. "This has convinced us that within the limits of what can be done with normal human psychophysical testing, we have a basis for proceeding on the functional grounds."
The system includes a subminiature camera-like device that captures the image in real time; an electronic version of the image then is translated into a series of wireless commands that cross the scalp to a group of modules implanted in the visual region of the brain (Figure 1). Each module contains 16 intracortical electrodes, each with an exposed tip approximately the size of the neurons they are intended to stimulate.
"With the correct spatial-temporal pattern of stimulation, it should be possible to manipulate the neural machinery of vision to create the perception of the image in the person's brain," Dr. Troyk said.
He estimated that anywhere from 300 to 1,000 electrodes would need to be implanted to produce the desired results. These electrodes would penetrate the surface of the cortex to a depth of 1 to 2 mm, reaching the neuronal layer where projections from the optic nerve can be found. This site is also known as Area V1, the primary visual cortex.
The objective is to create a grid of non-uniformly sized dots or phosphenes. If enough of these dots are apparent as the camera scans around the image, the subject in whom the device is implanted should be able to interpret the image.
"It's sort of like looking through the world through a soda straw," Dr. Troyk explained, except the straw would be scattered with unevenly spaced holes. The hypothetical straw would be moved around to give the prosthesis user limited views of the world.
The experiments that Dr. Troyk described involved normal subjects who were timed while performing a series of psychophysical tasks. They were asked to scan a checkerboard and count the number of white squares, place black checkers on the white squares, and complete a simulated mobility task. For this assignment, the subjects were shown a simulated maze on a computer screen and asked to navigate from room to room using a joystick while overcoming various visual obstacles. Ultimately, they learned to navigate at about 11 seconds per room in a maze of approximately 20 rooms.
With as few as 325 phosphenes, the above-average performance of these normal subjects doing pattern recognition tasks was statistically significant.
"We were pleasantly surprised at how well they could do with an impoverished display," Dr. Troyk said. "When you look at the display that they got and then look at what they're doing, it's pretty amazing that they're able to do it. It speaks to the ability of the brain to adapt when that's the only information that it's presented with."
Processing sparse information
This ability to process something meaningful from sparse information is the hope with all visual prosthesis systems currently under development, he continued. This ability apparently is due to the fact that about half of the brain is dedicated to processing vision, and the neural machinery can interpret even minimal information and use it for a sensory benefit.
"It would not be vision as we normally know it, but it might be some sensory information that people would be able to use in daily tasks," Dr. Troyk said. "A visual prosthesis is not of much use if all you can do is learn to see a big E on a wall chart after training for half a day. You have to be able to walk into an unknown environment and determine what's in that environment."
He also discussed his views of the future of visual prostheses; several projects are under way in various locations around the world, targeting several different areas of the brain in addition to the cortex.
"There are going to be visual prostheses. Initially, the success is going to be sporadic and scattered, but the technology will become more sophisticated. What's desperately needed is a quantum leap in technology for the artificial interface to the biology. Right now that's where the bottleneck is," Dr. Troyk said. "How do we have the technological-biological interface stable so that we can manipulate the neural network the way we would like to?"
When that hoped-for quantum leap in technology occurs, the field of visual prosthetics will "explode like we can't even imagine," he predicted. "With the present technology, we can do a lot, we can learn a lot, and we can possibly even provide some limited user population with sensory benefits. Some of those [benefits] might include having the perception of a circadian rhythm, which does have a significant health impact."
Dr. Troyk suggested, though, that visual prostheses probably won't be used for reading because better technologies specifically for this purpose constantly are being developed. But visual prostheses could be used in combination with other technologies to help with navigation. For example, a prosthesis might provide supplemental information that could help a blind person navigate with a cane.
A prosthesis also could offer the wearer a limited degree of facial recognition, and there also could be psychological and emotional benefits that are hard to quantify, he continued.
"We won't know what people can do with them until we can deploy them into use 24 hours a day," Dr. Troyk said. "That's where everybody is right now. We want to deploy them, get them out of the research laboratory, and just see what the human brain is capable of and how well people can do with this impoverished visual information."