Restoring Sight to the Blind Illustration

Restoring Sight to the Blind


Welcome to the home page of the Thalamic Visual Prosthesis Project.

¤ Support our work. We rely on donations from individuals and foundations to support our important research to restore sight to the blind. If you are interested in making a tax-deductible gift, visit Any amount helps. Please inquire with your employer's HR department if they match charitable donations. And thank you.

¤ The fundamental idea we are pursuing is to provide restoration of sight to the blind. We hope to accomplish this by implanting multi-wire electrodes in the lateral geniculate nucleus (LGN), the part of the thalamus that relays signals from the retina in the eye to the primary visual cortex at the rear of the head. In leading causes of blindness, the eye ceases working as a light-sensitive organ, but the remainder of the visual system is largely intact. By sending signals from an external man-made sensor such as a digital camera into the brain through carefully implanted electrodes in the LGN, we hope to provide a crude approximation to normal vision and restoration of sight to the blind.

It is important to understand that we do not anticipate restoring vision that is in any way close to normal. Our best guess is that a visual prosthesis will provide the patient with an improvement in their quality of life, being able to navigate more easily through familiar and perhaps unfamiliar surroundings. We hope that it will allow the patient to distinguish and identify simple objects, perhaps even help recognize people. But, it is important to understand that these hopes are some time to come. There is a tremendous amount of work to be done before we have even the crudest initial experimental device temporarily implanted in a human.

¤ We have published a scientific paper describing our first high-profile results:

J. S. Pezaris and R. C. Reid, "Demonstration of artificial visual percepts generated through thalamic microstimulation," Proceedings of the National Academy of Science, 104(18):7670-7675, May 1, 2007 [PDF]

¤ Here are a few selected examples of the press coverage on the paper:

The Economist
The New Scientist
Associated Press (NYT)
Technoglogy Review

¤ The following illustration depicts a schematized version of an initial device and shows the basic elements of the design.

Pezaris Reid 2007 PNAS Illustration

This cartoon diagram shows how a visual prosthesis might work someday. The patient would wear a special set of glasses with a small digital camera mounted in the lens. The camera would have a wire that communicates to an external signal processor, worn in a pocket or on a belt. The signal processor would translate the image from the camera into the neural impulses and transmit them wirelessly to an implanted stimulator. The stimulator would drive the electrode, surgically placed in the brain, delivering images to the visual system. We expect the actual surgery to be comparatively minor, as it is based on the well-developed methods of Deep Brain Stimulation already used extensively to treat Parkinson's tremor.

Credit: J. S. Pezaris, adapted with permission from D. H. Hubel.

¤ The first of two small movies accompanying the article and press coverage helps understand what the animals are doing in this experiment.

This animation combines the animal's eye position, what it sees on the screen normally, and a simulation of the artificial percept. As part of the experiment, the animal's eye position is measured, and we show where it was looking on the computer screen by a blue dot. Targets that appear on the screen are white dots, and the animal has been trained to look at them, from a first one that appears in the center, to a normal second one that appears further away. In some instances the second point is not on the screen, but is created artificially through electrical stimulation. Although there is a yellow star depicting the electrical stimulation in this movie, in the experiment, nothing appeared on the screen in front of the animal. When the second point is artificial, the animal looks down and to the right, just as if it saw one of the normal points. By changing the placement of the electrode, we can change where the percept appears.

Credit: J. S. Pezaris.

¤ The second of the two movies helps us understand what prosthetic vision might appear like to the patient. This could be called an artist's rendition of the experience. There are numerous assumptions that underlay the simulation, many of which are likely incorrect; this movie should serve only as a guide.

On the left is a movie made with a small digital camera. On the right is the same movie, as it would appear to a patient with bilateral implants having 350 pixels per hemisphere. As the simulated patient moves their gaze around, as indicated by the red point, you see the pattern of pixels shifting across the image.

There are two things to notice in this movie. The first is that before the movie begins, the image on the right is not identifiable, but as soon as the animation starts to run, the brain does a marvelous job of integrating different facets of the image and the woman's face becomes clear. The second is that the right image contains about 700 pixels, while the left image contains 70,000; while not all of the fine details are resolvable, a remarkable amount of information can be conveyed in a relatively small number of pixels.

Credit: J. S. Pezaris.

¤ We have published a second scientific paper showing progress on the design parameters of a device:

J. S. Pezaris and R. C. Reid, "Simulations of electrode placement for a thalamic visual prosthesis," IEEE Transactions on Biomedical Engineering, 56(1):172-178, 2009 [PDF]

¤ We have published a scientific paper discussing possible modes of bringing signals into the brain, specifically using a thalamic visual prosthesis as an example of the larger field of computer-to-brain interfaces:

J. S. Pezaris and E. E. Eskandar, "Getting signals into the brain: Visual prosthetics through thalamic microstimulation," Neurosurgical Focus, 27(1):E6 2009 [PDF]

¤ Recently, there has been an effort to simulate how prosthetic vision will appear to the eventual recipient of an implant by using virtual reality technologies. We developed a simulation and used it to assess the visual acuity that would be available using a suite of different designs:

B. Bourkiza, M. Vurro, A. Jeffries, and J. S. Pezaris, "Visual Acuity of Simulated Thalamic Visual Prostheses in Normally Sighted Humans," PLOS ONE, 10.1371/journal.pone.0073592

¤ Continuing in that line of research, we adapted the simulation so that it could be used to test reading ability. Reading is one of the standard activities of daily living, and lends itself to easy measurement and analysis, as is detailed in our most recent publication:

M. Vurro, A. M. Crowell, and J. S. Pezaris, "Simulation of thalamic prosthetic vision: reading accuracy, speed, and acuity in sighted humans," Frontiers in Human Neuroscience, 10.3389/fnhum.2014.00816
Here is a movie from that study that shows one of our subjects reading the sentence "Ten different kinds / of flowers grow by / the side of the road" out loud using a simulated prosthesis. The subject's eye position is shown by the red circle, but was not visible to them during the experiment. The pattern darts about the screen as the subject looks from word to word. It might seem amazing that they were able to read at all; we have found a large gap between what someone watching one of these experiments understands and what the subject performing the experiment experiences. The simulated phosphene vision here has many more phosphenes (four thousand) than would be available from the proposed device, but we also tested lower resolution versions. Watch it a few times, and you'll start to see the words better, but keep in mind the person in this example had never used the simulation before.

Credit: J. S. Pezaris.

¤ In preparation for implanting a first-generation prototype in an animal model, we have been training monkeys to perform the same letter recognition task that we did with humans in the Bourkiza, et al. work above. While monkeys don't understand images of letters the way that humans do, they are very capable of distinguishing arbitrary visual shapes. We use letters as a convenient set of arbitrary shapes because they allow us to directly compare animal results with human results. During the training, we had the animals perform exactly the same task as before, in a simulation of artificial vision. Because monkeys learn this task much more slowly than humans, the training gave us an opportunity to study the learning of the task in very fine detail.

Killian NJ, Vurro M, Keith SB, Kyada M, Pezaris JS, "Perceptual learning in a non-human primate model of artificial vision," Scientific Reports, 10.1038/srep36329

¤ Finally, there is a longer video that describes the research, although it is somewhat dated at this point. If the movie does not play, it can be downloaded as a WMV file by clicking here.

Credit: J. S. Pezaris.

Copyright 2006-2016 J. S. Pezaris    |