Artificial vision may seem like science fiction, and it's true that the kind you see in "Star Trek" or "Blade Runner" still is, but there are projects all over the world that are successfully giving back partial vision to to blind patients. There are, however, a number of obstacles: the size of the microelectrodes, the way of powering the device, the type of blindness the person has and other factors prevent current treatments from doing much more than letting patients see a few monochrome blobs.
That's enough to safely navigate a room or street (no small improvement), but what about recognizing faces and objects, or reading signs and symbols? New research by Dr Sheila Nirenberg at Cornell and Chethan Pandarinath at Stanford University claims to make such levels of acuity possible.
Their method doesn't rely on just making electrodes smaller or increasing the size of the image sensor. Instead, they looked at how the healthy retina communicates with the brain and tried to emulate that.
The retina is a complicated, multi-layered web of cells that are networked together and constantly communicating. Some forms of blindness result from a degeneration of the light-sensitive cells (rods and cones) while the rest of the neural circuitry remains in place. Loss of any entire cell type would cause blindness as well, but when this particular type happens, that means that the ganglion cells, which collect information from multiple rods and cones and collate it, are intact and could still potentially send signals to the brain.
It's as if two people were talking on the telephone: the conversation will end either if the line itself is disrupted, or if one of the people hangs up. In this type of blindness, the line is fine and the brain is still listening, but no one is talking on the other end. And as it turns out, the replacement signals sent by existing retinal implants have been extremely garbled. What the researchers did was to find out how to send a signal that is much more easily understood.
By studying ganglion cells closely, Nirenberg arrived at a sort of algorithm that describes how the ganglion cells expect to be fed information from the rods and cones. By taking the normal image signal and passing it through an "encoder" running this algorithm, their device can send that image to ganglion cells in such a way that a much clearer image is sent to the brain. You can see the differences in this diagram:
The technique, which they call "optogenic stimulation," works like this: the digital image, provided by a camera or image sensor in the eye, is sent to the encoder, which then sends the special encoded image to a microscopic projector. The projector shines onto the ganglion cells, which have received gene therapy so that they respond to light somewhat in the way the missing cells would have. And then the ganglion cells send that image along.
With it, they claim that 9,800 ganglion cells, properly treated and exposed with the device, will be able to "bring prosthetic capabilities into the realm of normal image representation." That is to say, a grid of 100-by-100 of them would give enough visual information that a person would have a serious semblance or real vision.
The experiments thus far, successful as they have been, were all performed on mouse retinas. But the researchers see no reason why it should not be attempted for humans as well; Nirenberg says that the gene therapy portion is the most important thing to test thoroughly, though similar techniques have already been used in the retina for other diseases.
Nirenberg and Pandarinath's paper, "Retinal prosthetic strategy with the capacity to restore normal vision," was published recently in the Proceedings of the National Academy of Sciences.
No comments:
Post a Comment