The World Heath Organization estimates that upwards of 3 million people suffer from blindness caused by degeneration of the photoreceptors in the retina, typically from diseases such as age-related macular degeneration (AMD) or retinitis pigmentosa. That number is expected to rise to more than 5 million by 2020, driving the development of retinal prostheses. Technologies currently under development include photodiode-array implants (Optobionics; Wheaton, IL) and external imagers wirelessly connected to implanted stimulation devices (Second Sight LLC; Valencia, CA). In both cases, only a limited number of electrodes are implanted at the retinal wall to interface with ganglion cells and induce a form of sight. Now, using microchannel glass arrays, researchers at the Johns Hopkins University (Baltimore, MD) and the Naval Research Laboratory (NRL; Washington, D.C.) have developed a massively parallel electronic/tissue interface that simultaneously provides millions of electrical stimulation channels.
Microchannel array fitted with microwires and a multiplexer chip interfaces with the retina to stimulate the optic nerve.
The hybrid test implant developed by the group integrates a microchip multiplexer with a glass matrix containing microwire electrodes. The microchannel glass-array matrix contains millions of microwires, each approximately 5 µm in diameter. Implanted between the multiplexer and the retina, the matrix stimulates the ganglia cells; it is axons leading away from the ganglia that form the optic nerve.
NRL's Dean Scribner said the microchannel glass array grew out of work sponsored by the U.S. Defense Advanced Research Projects Agency (DARPA) aimed toward using living cells interfaced with silicon chips as a way to detect chemical and biological agents in real time. "Neural tissue is very sensitive to minute levels of toxins and they react very quickly, but somehow you have to be able to talk to the tissue. You need a neural tissue interface," he says.
The microchannel glass array starts with an acid-etchable glass core inserted into an inert glass tube. Thousands of these combinations are bundled together, heated and then drawn, similar to an optical-fiber preform, then the process is repeated. The resultant glass boule is then cross-sectioned into a wafer several millimeters in diameter and a couple millimeters thick (approximately the same size as the multiplexer chip). After polishing the glass flat and acid-etching it to remove the glass cores, the group uses electrodeposition techniques to create microwires in the holes.
The wafer is repolished flat on one side, while the other is curved to conform to the curved inside of the retina (see figure). A second, selective acid wash removes a few microns of glass so that the microwires protrude slightly from both sides of the array. Finally, the multiplexer is bump-bonded to the flat side of the glass array. Because of the small diameter and close packing of the microwires, several microwires connect to each unit cell on the multiplexer chip, which helps to ensure redundancy.
Scribner says the multiplexer chip allows scanned images to be delivered to the retinal cells simultaneously. Simultaneous stimulation from the multiplexer in all 80 x 40-unit cells is necessary because the human eye is used to receiving optical signals in parallel, not scanned like a TV. During experiments in the laboratory, the image data collected by an external CCD camera was fed through a PC and across a cable to the multiplexer chip, where each cell stores the pixel value. Then a biphasic pulse was sent through each cell simultaneously, which was then modulated to reflect the pixel value. Using biphasic pulses, rather than just solid positive or negative electrical pulses, eliminates disassociation of the electrodes (similar to electroplating) and the formation of hydrogen and oxygen bubbles in the tissue through electrolysis.
"It's very interesting work," says Robert Greenberg, Second Sight's president and CEO. "The challenging part is the interface to the patientthe electronics are relatively straightforward. If the array turns out to have advantages, then that's the important part of the project."
Future work will focus on integrating photodiodes on the multiplexer chip. This implies a backside-illuminated design using standard thin-chip CMOS with power input via electro-magnetic coupling to external source. Heat dissipation does not appear to be a problem because the device can operate using approximately 10 mW of power.