22 December 2009—Several teams of scientists and engineers have been trying for years to produce a practical retinal prosthesis for people afflicted by a progressive loss of photoreceptor cells. One problem all the researchers face is how to get power and data (the image) to a retinal chip that’s implanted at the back of a person’s eye. Some groups’ implants, such as those from the University of Southern California’s Doheny Eye Institute and an MIT-Harvard team get their power and data from RF signals beamed in from the outside, while other groups, including one at the University Eye Hospital in Tübingen, Germany, are working on getting the data as light entering the eye using RF energy to beam in the power. But a team from Stanford University has been working on what might seem like the obvious solution: using light entering the eye for both power and data.
The Stanford implant is designed as an array of miniature solar cells. The device—technically a subretinal implant, because it is placed behind the retina—is part of a system that includes a video camera that captures images, a pocket PC that processes the video feed, and a bright near-infrared LCD display built into video goggles. The pulsed 900-nanometer-wavelength image that shines into the eyes is enough to produce electricity in the chip. (A chip driven by just the ambient light coming in to your eye would produce current that is one-thousandth or less the strength required to trigger retinal neurons.) The researchers chose a near-infrared display because it is invisible. Some patients’ retinas might still have some working photoreceptors that could be stimulated by visible light. Visible light bright enough to stimulate cells would yield artifacts that would muddy the image generated in the brain.
Stanford's 3-millimeter-wide chip is configured in three layers that together are 30 micrometers thick. The array is a series of pixels, each formed from a three photovoltaic cells of three different sizes.
The purpose of the multiple subpixels, say the Stanford researchers, is to boost the amount of current each pixel sends to the still functional intermediate layers of the retina that perform the eye’s natural image processing and data compression. (These layers perform compression so that data from the eye’s 130 million photoreceptors can be sent on the 1.2 million axons in the optic nerve.) The Stanford researchers’ most recent work, reported at the IEEE International Electron Devices Meeting in December, improves on their system by making it fit the curvature of the eye, which keeps all the pixels in focus. The team used a MEMS fabrication process that left the pixels mechanically connected by 300-nm-thick silicon flexures. The flexures allow the array to curve along with the natural shape of the retina but provide enough resistance (roughly 100 megohms) to render the array’s elements electrically isolated.
As advanced as the retinal prosthesis is, the vision it provides would be limited. According to Daniel Palanker, a Stanford professor of ophthalmology who worked on the chip, a device with 100 µm pixels corresponds to a visual acuity of about 20/200. (That figure, which is the threshold beyond which a person is considered legally blind, means that the person would have to be within 20 feet [6.1 meters] of an object to see it with the level of clarity that a fully sighted person experiences from 200 feet away [61 meters].) In the best-case scenario, a photovoltaic prosthesis is limited to a pixel size of about 50 µm, corresponding to visual acuity of 20/100. That should suffice, says Palanker, for face recognition and for reading large fonts.
Work in the field of retinal prostheses is at various stages of development. Of the dozen projects currently active, the furthest along is USC’s Argus II. In the late 1990s, the USC researchers formed a Sylmar, Calif.–based company called Second Sight, which has since commercialized the implant. In the first generation of the device, a camera on a pair of glasses wirelessly sends video signals to a receiver implanted inside the head. The signal is then sent by wire to 16 electrodes attached to the retina. Five patients who were once totally blind still use that version of the device. In the second generation of the device (Argus II), the video data is transmitted wirelessly to a coil that surrounds the eye’s iris (but is not visible to onlookers), which in turn routes the data to a chip attached to the side of the eyeball. The chip processes the data and sends pulses via a thin cable to electrodes implanted on the retina. At least 13 people have been outfitted with the new version, which contains 60 electrodes.
The MIT-Harvard team’s system, which is still being tested on animals but may move to human clinical testing in three years uses similar RF transmission technology, but the electrode array is placed in subretinal space where the Stanford researchers place their photovoltaic chip.
The Stanford group is not the only one to attempt photovoltaic conversion. Optobionics Corp., of Glen Ellyn, Ill., got as far as Phase 2 clinical trials using a 5000-microphotodiodes chip. But the firm went bankrupt before trials could continue. According to Palanker, the ambient light Optobionics relied on is insufficient to drive an implant.
Palanker says he has ”no clue” when a version of the Stanford system approved for clinical use will be available. The group is perfecting the system in animal experiments and developing image-processing software for the implant.