Error loading MacroEngine script (file: )

Saving Eyesight with Early Detection

By Tim Schnettler

Diabetic retinopathy is the most common diabetic eye disease, but its symptoms often do not manifest until it is too late. Texas A&M researchers are working to make early detection easier.

Sight is a precious gift. But it is one that an estimated 20,000 people lose each year to an unexpected source, diabetes. Diabetic retinopathy is the leading cause of adult blindness in Americans between the ages of 25 and 74. Early detection could prevent up to 90 percent of those cases, according to the AgrAbility Project.

Diabetic retinopathy is caused by too much blood sugar, resulting in blocked or damaged capillaries and lesions in the eye; blurred vision; and, sometimes, blindness.

Detecting the symptoms of diabetic retinopathy in the early stages has long been a problem because of the amount of labor and the number of errors associated with screening for the disease. Researchers in Texas A&M's Department of Computer Science and Engineering are using sophisticated computer analysis techniques to make early detection easier.

"Diabetic retinopathy is a hidden problem, and chronic patients don't have obvious symptoms before it becomes too late," says Steve Liu, a professor in computer science. "There are issues with how you detect and identify these symptoms earlier so you can begin treatment to slow down the development of this disease.

To do this, Liu and his colleagues have developed a computer-based diabetic retinopathy detection system that uses computer algorithms to recognize the well-defined symptoms of the disease, especially the symptoms that appear in the earliest stages.

This system, the Texas Advanced DR detection system (TADRS), involves two key pieces of equipment: an off-the-shelf retina camera and software that has been developed by Liu and other researchers to read the data from the picture of a patient's retina.

The camera that is used is a nonmydriatic retina camera that can take pictures of the retina without having to dilate the pupil. Not having to dilate the pupil allows someone with less training to do the screening, providing greater access for more patients.

"Dilation would give a better-quality image," Liu says. "But because we want to do this, hopefully in primary care - a first-line-of-care location - without dilation you don't require extra training for the nurse."

The software Liu developed takes the data and picture of the retina from the camera, analyzes it, and recognizes trouble spots that could be signs of diabetic retinopathy.

"We have worked with ophthalmologists, and every one of them points to one thing that is critical - microaneurysm," Liu says.

A microaneurysm is an enlargement of blood vessels in the retina at weak spots caused by high levels of glucose in the blood. The balloon-like enlargements gradually force their way through the retina's tissue layers and show themselves as circles or blobs. Being able to detect them is crucial for diagnosis.

"This is the most reliable indicator," Liu says. "We have been told to focus on this as [one of the] major detection criteria. We developed the computer procedures to detect this."

Currently, the collected data is sent to screening software servers in Liu's lab on campus. Ultimately, the goal is to place the detection software on-site so that once a patient is screened, health care providers can decide immediately whether to forward a case to a higher level of care.

"The image of the retina will be read by the computer software, and if it passed a certain bar of concern, it would be sent to the clinicians, who are the experts and the most reliable source," Liu says. "The objective is to not have the data read by a professional unless the case becomes serious enough.

"The majority of the people coming to visit do not have a problem. If everybody's results were read by a professional, it would be a waste of time and money."

In addition to cutting down on the time necessary to review the scans and determine whether further action is necessary, the process allows diabetics to have their retinal photos screened more regularly and more conveniently.

Liu's research will also address one of the major barriers to diagnosis: cost.

"If you look at the population distribution of diabetes, many of the individuals are in under-cared-for sectors," Liu says. "People's eyes could be photographed, but it is a costly, slow and labor-intensive process. This can be eased by this technology."

The technology can help screen for diabetic retinopathy, but Liu points out that it is not a tool for diagnosing diabetes; instead, it gives clinicians an idea of how the vascular network is being affected by chronic situations, one of which is diabetic retinopathy.

TADRS is currently being used in clinical trials in 12 Texas cities - Alice, Beaumont, Bryan, Conroe, Corpus Christi, Georgetown, Laredo, Nacogdoches, Paris, Richmond, Sinton and Waco - as well as internationally in Mexico City, Saltillo, and Monterrey, Mexico. Liu says the response has been extremely positive, a sign of hope for addressing the leading cause of adult blindness in Americans.

"It is a problem that is too important to be put aside." Liu says.

Dr. Steve Liu
Dr. Jyn-Charn (Steve) Liu
Professor
Computer Science & Engineering
979.845.8739