Rewiring the retina

Senior Editor

The artificial retina will use a flexible, silicone-rubber-based implanted electrode such as this.

Although having artificial retinas in both eyes could give patients more visual information, it would double the cost and add another layer of complexity.

Researchers use techniques borrowed from IC manufacturing to make the silicone-rubber electrode array layer by layer.

Power and processed visual signals will be sent into the retinal implant via a hookup behind the patient's ear. System status and other data will be transmitted out of the implant using the same wireless connection.

Hundreds of thousands of people in the U.S. have been blinded by macular degeneration, retinitis pigmentosa, and other retinal diseases that damage the eyes' rods and cones. But while the rods and cones, the photoreceptors, might not be up to snuff in these people, the ganglion cells beneath them are often still working, patiently waiting to pass neural signals onto the brain via the optic nerve.

Doctors and engineers working on the government-sponsored Artificial Retina Project want to exploit those still-healthy ganglion cells, stimulating them directly, to return some degree of sight to those blinded by retinal disease. The Energy Department is providing $9 million in funding over three years for the project, and much of the R&D is being done at its national labs. The universities of Southern California and North Carolina State also play major roles, and Second Sight, a private company in Sylmar, Calif., plans to commercialize the final product.

The big picture

In principal, the artificial retina is fairly simple. The patient wears a pair of glasses with a camera mounted on it. The camera's video signals are digitized, processed, converted back to analog, and sent along a cable to an implanted IC, which distributes them to the appropriate electrodes in an array on the retina. Signals sent to the electrodes are strong enough to stimulate the ganglions, which send nerve signals to the optic nerve and then the brain. Patients perceive incoming signals as visual stimuli appearing as dots of light.

In clinical trials, Second Sight has temporarily placed 4 X 4-mm arrays of 16 electrodes in 20 volunteers and permanently implanted them in three others. The array covers roughly a 30° field of view in the patient's eye. Patients report seeing lines, letters, and shapes that correspond to which electrodes are being stimulated. Engineers at Second Sight hope to increase the array to 1,000 electrodes, about a 32 X 32 matrix, while keeping the same 4 X 4-mm size. "This should increase resolution and give us good quality images," says Robert Greenberg, president of Second Sight. "Our goal is to give patients enough vision to get around in unfamiliar surroundings."

Greenberg hopes to get the artificial retina FDA approved and on the market in two to five years. He envisions the device and surgery will cost about $30,000, the same as a cochlear implant, a similar device that bypasses damaged nerves in the ear and delivers sound to deaf people. In fact, Second Sight was founded by Alfred Mann, the same person who founded Advanced Bionics, the only U.S. company making cochlear implants. Second Sight benefits from this association in that it has access to Advanced Bionics experience in implanting electrodes. But even with this help, there are still significant hurdles to clear in refining the artificial retina.

Refining the hardware

Some of the hardware, especially that outside the patient, is relatively straightforward, thanks to the cochlear connection. For example, a computer worn on a belt or harness handles signal processing and sends signals and power through a transcutaneous, two-way inductive coupling located behind the patient's ear, just like the cochlear implant. A magnet placed beneath the skin holds the coupling in place and aligns exterior and interior antennae.

Currently, researchers are concentrating on getting patients to "see" single, black and white images, not streaming, full-color video, though that might not be far off. So signal processing involves edge enhancement, averaging to reduce the number of pixels (each electrode represents a pixel), and a host of other tricks.

At this stage, researchers are experimentally activating electrodes at different frequencies and sequences, changing signal strengths and conditioning algorithms. "Because all of this is new, there's a lot we don't know," says Greenberg. "So our first-generation device is extremely general in terms of hardware and software. It lets us explore signal and array parameters to see what they produce."

"We've also begun higher, more complex experiments with video, but we haven't released results yet," he says. "And we've been able to produce color, or the perception of color, though initial devices will be black and white."

Power - about an eighth of a watt - along with instructions, are transmitted through the coupling, while data from the implant is sent out. The company isn't saying exactly what data is sent out, but it does include status information on electrodes and implanted electronics.

There's another electronics package implanted behind the ear. "We do as much processing outside the body as possible, but there's still some tasks we have to do inside," says Greenberg. "For example, the implanted circuitry lets us shape the biphasic pulses and control the current drivers." Biphasic stimulation alternates between negative and positive stimulation voltages. Monophasic stimulation, either all positive or all negative, changes cell chemistry and damages cells.

A wire stretching under the skin connects the electronics to the array. At first, this wire was a problem. It kept breaking as the eye moved. Researchers strengthened the wire and added a bit more slack.

The array is made of polydimethylsiloxane (PDMS), a flexible and biocompatible silicone rubber. The thin, elastomeric material conforms to curved surfaces, such as the retina, and can be molded to more closely match the retina's curvature. The current array uses micromolded ribs along the perimeter to help withstand handling and implantation without folding or curling.

A challenge in the project was to devise a method of attaching the array to the retina. "The retina has the consistency of wet tissue paper," notes Greenberg. "It is difficult to attach anything to it. We went through 20 different designs that we tested in animals before coming up with a solution, a metal tack about 1-mm across. It keeps the array close to the retina without damaging it."

"We've also been lucky in that there are no pain sensors behind the eye," says Greenberg. "So patients feel no discomfort after the array is in place."

At Livermore Labs, engineers at the Center for Microtechnology are working on "metallizing" PDMS, selectively applying metal lines onto the array to act as wires and electrodes then applying an insulating coat of PDMS. The lines are about 100 microns wide and 0.25 microns thick and are made using IC batch-manufacturing techniques. Batch fabrication should bring down the cost, once arrays are made in volumes, and they give researchers a method of cramming 1,000 electrodes into the space now occupied by 16.

One of the breakthroughs at Livermore was developing a way to make the proprietary metal traces or wires stretch. "This is critical for flexible devices designed to conform to the shape of the retina," says Peter Krulevitch, leader of the Livermore team developing the array. "Right now, the wires withstand approximately 7% strain before breaking. We are working to increase the amount of strain it takes before failing."

Making it biocompatible

Another Second Sight goal is to make the implant last the life of the patient. That's one reason power gets sent in from outside; it means no battery to change or recharge. But the eye is a corrosive environment and the amount of stimulation planned adds another level of stress on components. "It's like taking your TV, throwing it in the ocean, and hoping it works," says Greenberg.

Fortunately, Second Sight can call on colleagues at Advanced Bionics and other companies. Collectively, they have over 30 years designing implanted insulin pumps and pacemakers. "So while we've got a lot of technologies for protecting implants, they were developed for much larger devices. We've had to invent new methods that will let us do the same thing, but in the small space of the eye."

One method of making devices biocompatible is to coat them with ultrananocrystalline diamond (UNCD), a material being developed at Argonne labs. The Argonne team, Dieter Gruen, John Carlisle, and Orlando Auicello, makes the pure carbon material using plasma-chemical-vapor deposition (CVD). But unlike other CVD methods of making diamond film that use 1% methane and 99% hydrogen as the carbon source, Argonne researchers use 1% methane and 99% inert gases. That means the the only hydrogen comes from breaking down the methane molecule (CH4). Argonne's film has grains 3 to 5 nm in diameter, compared to the 30 to 100-nm grains resulting from other methods.

The UNCD coating is extremely inert, thus making it biocompatible. It could be used to coat future implanted electronics, chip-based electrode arrays, or to make electrodes themselves.

So far, researchers have had no problem with corrosion or encapsulation of implanted devices by tissues. "We've designed the implant to last a lifetime," says Greenberg, "And we've had no problems so far. The biggest challenge will be increasing the resolution of the image by adding more electrodes to the same sized array."

The Project still has a ways to go and must overcome both engineering challenges and regulatory hurdles before Second Sight can bring its artificial retina to the market.

Project roster

Organizations involved in the Artificial Retina Program include:

Argonne National Lab is testing components for biocompatibility and devising the packaging for the implanted electronics.
Lawrence Livermore National Lab is developing the electrode array.
Los Alamos National Lab is exploring optical-imaging techniques and signal conditioning for the implanted array.
Oak Ridge National Lab is managing the project, testing electrode arrays, and developing ocular sensors.
Sandia National Lab is also developing the electrode array.
North Carolina State University is leading research into powering the implant and communicating with it.
University of Southern California will provide medical direction and clinical services related to surgical implantation, testing, and medical follow-up.
Second Sight is producing prototypes that are being tested.

What the patient needs

Not every blind person will benefit from an artificial retina. A major requirement is that they have a viable optic nerve and that they could see at least for a couple of years. They need an optic nerve because it serves as a conduit for sending visual signals to the brain. Not long ago, conventional wisdom held that if a person did not use a major nerve, such as the auditory or optic nerve, it atrophied and became useless. Fortunately, this has been disproved. A team at Johns Hopkins, for example, has shown that 30% of the cells in the optic nerve are still active even in patients blind for many years. And one of Second Sight's volunteers had been blind for 50 years but could still discern some perceptions from the artificial retina.

Even with a good optic nerve, however, patients need to have had sight, even for a few years. It lets the brain's cortex develop neural pathways for handling and making sense of visual information. "We tested the array in a patient who had never seen before," recalls Robert Greenberg, president of the company commercializing the artificial retina. "He had a very emotional response. He described it as a retinal storm. It was as if we turned on some sixth sense he had never experienced before. We had to stop the experiment and since then, we've never tested a similar patient. But it's something we'd like to come back to eventually from a neurological perspective. It presents some interesting questions."