Aug 26 2007
Frontiers in Optics 2007 (FiO), the 91st Annual Meeting of the Optical Society of America, will be held from Sept. 16-20 in San Jose, Calif., alongside Laser Science XXIII, the annual meeting of the American Physical Society Division of Laser Science.
Restoring Sight, One Pixel at a Time
Researchers at the University of Southern California (USC) are developing a tiny camera for prosthetic systems that can be implanted directly into the human eye and connected to the retina, the part of the eye that converts visual information into electric signals that travel to the brain. Such an implantable camera would represent an important milestone in the ultimate goal of providing limited vision to those rendered blind by certain diseases, via a fully implantable retinal prosthetic device.
In both retinitis pigmentosa and age-related macular degeneration – two of the most common causes of vision loss – the photoreceptor layer of the retina is destroyed, but the inner layers remain largely intact, still capable of responding to incoming signals and transmitting output signals to the brain's visual cortex via the optic nerve. The discovery several years ago that direct electrical stimulation of retinal nerve cells in blind test subjects produced some sense of vision led to the development of the first retinal prosthesis.
Current retinal prostheses are designed to be used with an external (extraocular) camera mounted in a pair of glasses – awkward because subjects must move their heads in order to scan the environment. The miniaturized prototype being developed by the USC team would be directly implantable and would allow for natural eye and head movements.
In order to optimize the design constraints for their ultra-miniature camera, the group performed a series of studies to determine the minimum requirements for vision-related tasks like object recognition, face recognition, navigation, and mobility. They found that surprisingly few pixels were required to achieve good results for many of those tasks: approximately 625 pixels in total, compared to more than a million for a typical computer display. They also found that in many cases blurring images – both before and after they were converted into pixels –resulted in significantly improved object recognition and tracking – even better for moving objects than for static ones.
Taken together, these findings have made it possible to substantially relax the once extremely stringent design requirements of key components of the intraocular camera, thereby reducing the prototype intraocular camera's size and weight from an object the size of a Tylenol tablet down to an object that is now about one-third the size of a Tic-Tac. Early prototypes have been highly successful in initial tests, although human FDA trials are still at least two years out. (Paper FThT1, "Intraocular Camera for Retinal Prostheses: Design Constraints Based on Visual Psychophysics")