Ken Tobin, left, and Tom Karnowski, right, were on the team that developed a retinal scanning analysis software for detecting diabetic retinopathy and other potentially blinding diseases. Credit: Ron Walli, ORNL/U.S. Dept. of Energy.
ORNL researchers developed a retinal scanning analysis software in the early 2000’s that was an early precursor to the modern technology used in telemedicine to recognize eye diseases such as diabetic retinopathy. The latest guidelines recommend that people with diabetes be screened at least once a year for diabetic retinopathy, which is the leading cause of blindness in working-age adults. However, low-income and rural patients may struggle to access vision specialists.
ORNL’s technology, which was developed to support screenings by primary care physicians or pharmacies, utilized machine learning methods to compare images from a library of scans representing various eye diseases. It was licensed by a company that now produces retinal scanning equipment. Although machine learning methods and software have since advanced further, the innovation demonstrated an early medical application for them to benefit medically underserved populations that suffer disproportionate levels of chronic health conditions such as diabetes.
- In 2021, an estimated 9.6 million Americans had diabetic retinopathy
- 95 percent of vision loss due to diabetic retinopathy can be prevented with early detection
- As many as 30 percent of people with diabetes have diabetic retinopathy
- About 58 percent of diabetic adults follow the annual screening recommendation
- 70 percent screening rate is a Healthy People 2030 objective