Alfred Pasieka/Photo Researchers, Inc

Ablend of medical, electronic, and engineering break-throughs has improved the quality of life for many people faced with physical handicaps. New prosthetic limbs can move and flex like the original equipment and even respond to the wearer's own mental commands and cochlear implants let the deaf hear. Blindness, however, has proven less amenable to technological solutions. Yet that, too, may soon change. An assortment of engineers confronting that challenge now expect that, within a few years, those who have lost their sight to degenerative retinal disease will again see the light.

These bionic eyes won't restore normal sight, their inventors are quick to emphasize. Patients' visual perceptions will more closely resemble simple white patterns on a gray canvas, or perhaps coarse, colored pixels on a portion of the mind's video screen. "But to someone who can't see a difference between day and night, any ability to...


Competitors pursuing the first approach offer variations on a theme first established by biophysicist William Dobelle, who began researching electronic eyes in 1968. The patient wears glasses on which a video camera is mounted. The camera's signals are sent to a cell-phone-size processor clipped to the patient's belt or tucked in a pocket, and a cable transmits the images to a receiver implanted in the patient's eye.


Courtesy of the Imaging Staff, National Eye Institute

These images of diseased human retinas show some of the ways vision can become impaired. (A) Angiod Streaks, or hemorrhages in the macular and subretinal neovascularization; (B, C) Gaucher's Disease, an inherited metabolic condition demonstrating retinal deposits; (D) Glaucoma with Pan Retinal Photocoagulation (PRP) scars; (E) Toxoplasmosis-induced hemorrhage with lipid from a subretinal neovascular membrane; (F) X-Linked Retinoschesis, showing a split in the retinal layers.

Dobelle's prototypical "Artificial Vision System," available only at his clinic in Portugal, surgically implants electrodes in the patient's visual cortex, and trails connecting wires out through a grommet in the patient's skull to a portable, VCR-size computer. Images from the patient's eyeglasses-mounted camera are digitized in the computer and sent to the brain through the wires as a pattern of white dots.

Though Dobelle pioneered the field, his approach now is considered antiquated. "The FDA probably would never approve it because of the invasiveness of the procedure and the risk of infection," Dudley explains.

The largest alternative project under way in the United States involves a device that sends images not to the brain but to a grid of electrodes affixed to the diseased retina. Codeveloped by the University of Southern California's Doheny Eye Institute, the private start-up firm Second Sight in Sylmar, Calif., and the US Department of Energy, this artificial eye tucks a small packet of electronics inside the eye socket, then threads a filament of cable through a slit made in a region of the eye that lacks a retina, so vision won't be affected.


Courtesy of the USC Doheny Eye Institute

The intraocular retinal prosthesis consists of 16 electrodes arrayed in a 4x4 grid. These electrodes are stimulated by digitized images transmitted to the device from a camera mounted on a pair of glasses. The electrodes, in turn, stimulate the patient's remaining retinal neurons.

This signal wire ends in a polymer wafer about 6-mm square that contains as many as 100 electrodes. Each electrode emits an electrical pulse that penetrates the retina and travels through the optic nerve to the brain, where the pulse is perceived as a small white spot. The constellation of spots enables the patient to see visual patterns as a "connect-the-dots" picture.

Tests of the design using an early version of the device bearing only 16 electrodes showed promise. "It turned out to be more useful than we expected," admits physician Robert Greenberg, Second Sight's president and CEO. The engineers wondered if patients would be able to interpret a scene with relatively few points of reference. "But patients found that they can move their heads to scan an object and then mentally integrate the different views into an image that they can identify," says Greenberg.

That early, bulkier prototype has given way to a trimmed-down commercial version that will enter trials by 2006. "We expect FDA approval within five years," Greenberg says.


Advanced Medical Electronics (AME) of Minneapolis thinks it can make those connect-the-dots images even sharper by blending images from the visual and far-infrared spectra in one display. "Far-infrared radiation from an object is primarily a matter of heat," explains CEO Gary Havey. "It's almost impervious to shading or light conditions. As a result, far less resolution is needed to identify contrast between an object and its background when the objects in view have different temperatures." For example, a blind person using the system could detect the whereabouts of a black cat in a darkened room, something that other devices could have trouble doing.

AME's system, comprising the familiar glasses with mounted camera feeding data to a clip-on processor, is possible thanks to something called a microbolometer. A bolometer is a device that detects far-infrared radiation, but the usual versions are bulky and operate at cryogenic temperatures. The invention of small, room-temperature models sparked AME's corporate imagination while it was using the new devices to create sensory-replacement technologies for the US Army. Now the firm is laboring with scientists at the University of Minnesota and Johns Hopkins University to craft an effective way for software to blend and organize information from the two spectra and to present it intelligibly to the human visual cortex.

The resulting product will be marketed to companies such as IIP Technologies of Bonn, Germany, which takes an approach similar to Second Sight's but with a key difference: It replaces the eye's own lens with an artificial one. In a two-hour operation, the patient's intraocular lens is removed and IIP's light-sensing device, about twice the size of a match head, is implanted in its place. The sensor has an infrared diode in the center, which faces out from the front of the eye much like the eye's own pupil, only smaller. This diode receives a picture of the patient's surroundings in a series of infrared pulses from an external processor.

Training Your Seeing Eye

In technological seeing-eye systems, software that processes a camera's images needs to be customized to present the digital data in a pattern that matches the idiosyncratic way in which each patient's brain processes visual images.

That's usually done over a few weeks after the patient has recovered from the implant surgery. In the system developed by IIP Technologies of Bonn, Germany, for instance, a computer sends a series of random images to the implant, and the patient is asked to identify the one that most resembles, say, a circle. The software uses the patient's response to choose a new series of images, all of which are intended to resemble a circle, and the cycle repeats in an iterative process. The fine-tuning continues through a succession of basic shapes. After the patient gets accustomed to the new way of seeing, the software can be fine-tuned again any time the patient wishes.

The signals entering the sensor travel down a polymer filament about 3 cm long and 10 μm thick, embedded with electrical transmission wires. The rear of the filament flares to a small disk containing electronic controls and an array of electrodes. This flared end is tacked to the retina, and the electrodes translate the processor's infrared impulses into electrical signals.

To create the system, the company had to learn how to transmit enough visual information to create images while sending only minute amounts of electrical power to the implant. "There are a lot of regulatory guidelines about how much electromagnetic energy you can send into human tissue," Becker points out. Staying within those strictures was a series of painstaking gains of "five percent more efficiency here, 10 there, optimizing the design of the electrical coils, some clever tricks in preprocessing data to compress it," Becker says. "What we have is the sum of those efforts."

But that wasn't the only challenge. Mounting electronics anywhere in the body "is like throwing a television set in the ocean and hoping it still works," Greenberg notes. The fluids inside the eye, as in the rest of the body, are rife with minerals and corrosive salts. Although packing electronic devices inside the eye can be more efficient and less cumbersome to the patient, it adds the challenge of coating implants, or making them from special materials, to resist the body's inevitably caustic assault. Indeed, some scientists testing implants inside animals' eyes are finding that their devices begin eroding after only a few years.

That's unlikely to pose a health hazard, as the implants contain too little toxic material to poison humans. Instead, says Dudley, "It's more likely that eventually the implants will just stop working." But that's not deterring researchers bent on the second approach to artificial vision, which packs the entire device inside the eyeball itself.


VisionCare, with offices in Saratoga, Calif., and a laboratory in Israel, minimizes the corrosion problem by eliminating electronics and, in its place, using what it calls an Implantable Miniature Telescope (IMT). In a straightforward, hour-long operation, the eye's own lens is replaced with a tiny quartz tube containing a series of wide-angle lenses inside. The front portion of the tube is open to collect light; the tube's rear walls are solid to trap and focus the gathered light. The tube does not touch the retina but is fixed to the front of the eye by two spindly arms that are tucked inside the eye's lens capsule.

Macular degeneration leaves a blank spot at the center of a person's visual field, while peripheral vision remains more or less intact. The little telescope is a sophisticated magnifier that capitalizes on the vision that remains by enlarging the central part of the visual field that the patient can't see by up to three times. The telescope then shines that enlarged detail onto portions of the retina that still work. The enlargement shows the patient more detail than he otherwise could see.

"For example, if you suffer from macular degeneration and look at someone's face, you might see only the person's chin and hair," says Chet Kumar, VisionCare's director of business and market development. "But if that face becomes three times larger, perhaps then you can see the eyes and mouth and only the nose is obliterated by that blank spot at the center of your vision. Then suddenly you can recognize who's in front of you."

Because the IMT claims the peripheral vision in the eye it occupies, the device is implanted in only one eye of a person who has macular degeneration in both. The untreated eye is left to supply the wider view. Then, after surgery, patients undergo several weeks of graded training as they learn to see near objects with the treated eye while simultaneously using the untreated eye for peripheral sight. "In effect, they learn to see and integrate two views at the same time," Kumar notes.

After a clinical trial of 14 patients, VisionCare won regulatory approval to test another 200; the results will be finalized in late 2005. "We hope for the FDA's approval for our device in three to five years," Kumar says.


The IMT measures a mere 6.4 × 4.8 mm. Yet it looks downright huge next to the "artificial silicon retina" developed by ophthalmologist Alan Chow and his brother Vincent, an engineer. The implant is a silicon foil slightly wider than a pinhead, 25 μm thick (thinner than a human hair) and dotted with 5,000 photo-responsive cells. Under a microscope, it resembles a housefly's eye. The idea is as simple as the device. Natural light entering the eye falls on the photocells, which, in response, emit an electrical charge that mimics the natural stimulation no longer provided by the dying retinal cells.


Courtesy of James P. Gilman, CRA

This IMT (Implantable Miniature Telescope) device, which is implanted behind the iris during an outpatient procedure, renders central visual images onto healthier, peripheral portions of the retina.

In a 90-minute outpatient procedure, a surgeon makes a slit in the retina at the back of the eye and slips in the silicon disk so it lies against the layer of dying photoreceptor cells. To avoid damaging the macula, where light-sensitive cells are most densely packed, the chip is set slightly off to the side, but still close enough so that the electrical impulses produced by the retina overlying the chip are as strong as possible when they reach those tightly packed receptors.

All 10 patients in an initial clinical trial have regained some degree of vision and reported no serious difficulties with the chip. But surprisingly, patients claim that their field of vision continues to expand even after the surgery, indicating that photoreceptors near, but not actually contacting, the silicon photochip seem to be coming back to life as well. The Chow brothers theorize that the stimulated cells produce biochemicals that reactivate other cells nearby.

But this conjecture is controversial. Greenberg notes that some subjects in Second Sight's clinical trial claimed improved vision even with the implant switched off, "but when we objectively tested those patients, they could only detect light with the device turned on," he says, indicating that the claimed effects may have been psychosomatic.

The Chows also face another, more mundane challenge. Keeping all of the technology inside the eye "is beautifully simple," admits IIP's Holger Becker. "But the problem is that biological photoreceptors are about three orders of magnitude more acute than the best photocells that humans can build. That leads to the problem that artificial systems contained in the eye can't produce enough energy to achieve the resolution that the natural eye is capable of. So to achieve the sharpest resolution, you really need an external power supply."

Whether the externally assisted or wholly internal approach to artificial sight will prevail, or whether both will coexist, won't be known for more than a decade. Although each developer expects to field a market-ready artificial eye before 2010, Dudley counsels a prudent skepticism. "Given the need for extensive testing, we're at least 10 years away from a commercial device," he cautions, "and we can't predict exactly how useful these devices will be for vision. Patients won't be able to drive and probably won't be able to see the shades of color on a flower. But if you're blind, being able to tell where the door is may be enough of an improvement."

Bennett Daviss daviss@sover.net is a freelance writer in Walpole, NH.

Interested in reading more?

Magaizne Cover

Become a Member of

Receive full access to digital editions of The Scientist, as well as TS Digest, feature stories, more than 35 years of archives, and much more!
Already a member?