Utilizing neuroscience, gene therapy, and optogenetics, a pair of researchers from Cornell University have created a bionic prosthetic eye that can restore almost-normal vision to animals blinded by destroyed retinas.
We have discussed bionic eyes at length, but for the most part these have been dumb prosthetics — chips that wire themselves into the ganglion cells behind the retina, which are the interface between the retina and optic nerve. These chips receive optical stimuli (via a CMOS sensor, for example), which they transmit as electrical signals to the ganglion cells. These prosthetic eyes can produce a low-resolution grayscale field that the brain can then interpret — which is probably better than being completely blind — but they don’t actually restore sight.
The Cornell prosthetic eye however, developed by Sheila Nirenberg and Chethan Pandarinath, is a much closer analog to a real eye. Its construction and implementation is rather complex, so bear with me.
Comparison of various prosthetic eye/retina technologies
First, gene therapy is used to deliver special proteins to the patient’s damaged retina (i.e. caused by degenerative diseases, such as macular degeneration or diabetic retinopathy). By using optogenetics, these proteins have been modified so that they’re sensitive to light — they’re not quite rods and cones, but they’re along the same lines.
The next step is the clever/unique bit. For years now, Nirenberg has been working on decoding the signals sent by the retina to the brain. A year ago, she cracked this code. At the time, she had only cracked the code used by the mouse retina, but now she’s cracked the monkey code too — and a monkey’s retina is very similar to ours.
That’s not the breakthrough here, though: Nirenberg and Pandarinath have now taken the mouse retina code and developed a working prosthetic, completely restoring a mouse’s vision.
The prosthetic contains a camera pointed forward, a Texas Instruments OMAP 3530 SoC (system-on-a-chip), and a tiny DLP pico projector. The SoC converts the camera’s output into encoded data that the mouse’s brain can understand, and then the projector is used to beam that data to the optogenetic proteins that were earlier placed in the retina using gene therapy. The optogenetic proteins then transmit the encoded signal to the brain, via the ganglion cells and optic nerve. Voila: restored (grayscale) vision.
Prosthetic bionic eye, encoded vs. plain old optogenetic methods
In the image above you can see just how effective Nirenberg and Pandarinath’s prosthetic eye is. The top row is what a normal mouse eye would see (just before it gets eaten, seemingly), and the second row shows the images produced by the prosthetic eye. The bottom right corner shows the image your eye would see if it had only received the optogenetic gene therapy, and none of the fancy camera/neural-encoding tricks.
A bionic prosthetic eye setupWith the monkey retina code worked out, the next step is to build a pair of spectacles with all of the prosthetic equipment built in (pictured right). Before human testing can occur, the gene therapy step needs to be approved — but because similar gene therapies have been approved before, the researchers seem confident that human trials could begin in 1 or 2 years.
For more information, watch the TED video below, hit up the research paper (which is free for now, but may not remain so), or read our previous story about decoding the optic nerve’s signals.
I personally find this to be cool as hell. The article itself has image comparisons that are worth looking at. The improvement over what we've had for several years now is truly impressive. Not just a fuzzy blur, it's actually something where the face is distinguishable.
