The Eyes Have It
Researchers from USC Viterbi, the College and the Keck School head $6 million low vision project.By Diane Ainsworth
November 1, 2007
An interdisciplinary team of biomedical researchers from the USC Viterbi School of Engineering, USC College and the Keck School of Medicine of USC has received a $6 million Bioengineering Research Partnership grant from the National Institutes of Health to begin designing visual aids for millions of older adults who suffer from significant vision loss.
The USC team, led by Norberto Grzywacz, professor of biomedical engineering at the USC Viterbi School and director of the USC Center for Vision Science and Technology, will join other researchers from Harvard Medical School and the University of Houston School of Optometry to address low vision problems caused by neural pathologies, such as macular degeneration and other diseases affecting the retina.
Many of these vision problems are prevalent in older adults and cannot be fully corrected with ordinary lenses, medical treatment or surgery.
“Aging, injuries and diseases can all cause low vision, but the leading causes among older adults and the elderly are impairments such as age-related macular degeneration,” said Grzywacz, principal investigator on the five-year project, who also directs the USC Neuroscience Graduate Program. “Our dream is to build devices like intelligent glasses or intelligent television displays that can improve these people’s lives.”
The project brings together experts on the brain, the eye and cutting-edge technology that may help bridge the gap for older adults facing blindness.
Faculty from the College psychology department working on the NIH study include Bosco Tjan, assistant professor of psychology, who focuses on visual perception in people with normal and impaired vision; Bartlett Mel, an expert on image processing in the brain who holds appointments in psychology at the College and in biomedical engineering at Viterbi; Irving Biederman, holder of the Harold Dornsife Chair in Neurosciences in USC College and professor of computer science at USC Viterbi, who is an expert in object, face and scene recognition; and Zhong-Lin Lu, holder of the William M. Keck Chair in Cognitive Neuroscience in USC College and professor of biomedical engineering at USC Viterbi, who specializes in motion perception and perceptual learning.
Other USC faculty involved with the project are macular degeneration expert Mark Humayun, an engineer and physician who holds joint appointments at the Keck School’s ophthalmology department and USC Viterbi, and computer scientist Gerard Medioni of the USC Viterbi School.
Age-related macular degeneration (AMD) is a condition that usually develops in older adults and the elderly, and gradually destroys the central vision of the eye and an individual’s ability to see fine detail. AMD patients quite often lose their ability to read, recognize faces and drive.
With an aging population, the incidence of AMD is on the rise. According to Grzywacz, approximately 3.5 percent of people in industrialized countries over the age of 75 have AMD.
“That percentage rises to a staggering 18.5 percent for people over 85 years old,” he said. “AMD is also responsible for about 50 percent of all cases of registered blindness in industrialized countries.”
The statistics are just as daunting in the United States. According to the National Eye Institute, approximately 1.7 million Americans have some form of AMD.
Humayun predicts that by 2020, that number will climb to nearly 3 million. At the same time, an additional 8 million people will have clinical signs of AMD.
The NIH project will concentrate on designing visual displays that will help these people, who have lost their central vision and must rely on peripheral vision to see.
“We plan to use some of the techniques of computer vision and computational neuroscience to build visual displays that will enhance certain parts of an image enough so that a person with AMD will be able to digest the visual information better,” Grzywacz said. “We aren’t concerned with the optics of the eye in low vision — that can be corrected with glasses or surgery. Rather, our preoccupation is with the nervous system and the way in which the brain processes information.”
The nervous system has been damaged in people with low vision, lacking some kinds of neurons that process information, he said. In AMD, for example, people lack central photoreceptors, the neurons that transduce light energy into electrochemical signals, the means of brain communication. Grzywacz hopes to design visual displays that will compensate for that neural loss.
As part of the work, engineering faculty at USC’s Center for Vision Science and Technology will improve visual displays in two ways. First, they will enhance contrast in scenes and suppress background “noise,” or irrelevant details that might confuse a visually-impaired person. Second, they will design displays that brighten the contours of objects in a scene, creating a “cartoon” of outlines that would be more recognizable by someone with poor eyesight.
Medioni will lead the first task of building displays with region-specific contrast enhancement. His goal is to modify and extend an already existing system that was developed for photography, emphasizing visibility rather than aesthetics.
In standard photography, a camera adjusts its light sensitivity to the overall luminosity in a scene, so some objects will come out much darker than the brightest object in the picture. In the contrast-enhanced display, all regions of the scene would appear equally well lit, regardless of how dark or light they actually are.
“When you lose your central vision, you lose the ability to use the fovea, which is responsible for the perception of sharp details,” Grywacz said. “This is a region with a high concentration of cone photoreceptors.
“Once that region is damaged, an individual can only see in the peripheral region of the retina, which has far fewer cone photoreceptors and can only deliver information of low resolution to the brain,” he said. “The near periphery is inferior to the fovea and gets confused when too many details appear in the scene; scientists call this the ‘masking’ and ‘crowding’ effects.”
To combat that reduced ability to see sharp details, a second team of researchers will use automatic techniques to outline and simplify the main objects in a scene, as in a cartoon, to increase their visibility and salience. These simplified objects may be more easily recognizable by subjects with low vision, said Mel, who will lead the effort.
In the third phase of the study, a team of psychologists and clinicians from USC College, the Keck School of Medicine, Harvard Medical School and University of Houston School of Optometry will develop and administer tests to measure the effectiveness of these new visual display systems. This team has expertise in blindness and low vision and in the development of visual aids.
The researchers will administer a battery of high-level vision tests to probe psychophysically visual functions, such as recognition of objects, faces and scenes. Other, lower-level tests will help the team determine whether subjects have better visual acuity, contrast sensitivity, visual fields, color vision and reading ability.
The patients also will be trained in techniques of perceptual learning to make the best use of the devices; statistical learning tools built into the devices will adapt automatically to the needs of the patients. Tjan will lead that effort, along with Susana Chung from the University of Houston and Eli Peli from Harvard Medical School.
“There are two significant challenges in human testing,” Tjan said. “First, we must be able to test parts of the systems before the systems are built, because we need test results to guide system development. Second, the tasks used in these tests have to be rich and realistic enough to be relevant to people’s daily needs, but at the same time simple enough to be quantifiable and interpretable to affect system designs.
“We’ve made significant progress in designing these tests.”