ST. LOUIS — Researchers at Washington University School of Medicine in St. Louis, in collaboration with the University of Washington and Genentech, Inc., developed an experimental artificial intelligence system named OCTCube-M. A recent study found that OCTCube-M more accurately identified eight retinal diseases compared with older artificial intelligence models.
The technology uses three artificial intelligence models designed to interpret three-dimensional images of the eye's retina. Findings describing the system in its research stage were published in Nature Biomedical Engineering. OCTCube-M was also more accurate in predicting the progression rate of geographic atrophy, a severe form of age-related macular degeneration.
Professor of Ophthalmology and Visual Sciences Aaron Lee said, "Today's eye scans provide physicians an unprecedented, highly detailed view of the inside of the eye, revealing structures and subtle changes that would otherwise go undetected. But we still lack the tools to help physicians process the volume of generated images. Our AI system has the potential to empower physicians to make faster diagnoses, tailor treatment more precisely and design clinical trials that bring new therapies to patients faster." Lee added, "By better predicting how fast disease will worsen, we can run smaller, more efficient studies. That could lower costs, shorten the time it takes to test new therapies, reduce the number of people exposed to treatments that don't work and help effective drugs reach patients sooner."
Researchers trained OCTCube-M using over 26,000 three-dimensional optical coherence tomography images, comprising 1.62 million individual retinal slices. OCTCube-M identified six of the eight retinal diseases with a four to six percentage point higher accuracy than a model trained on two-dimensional images. This increase in accuracy translates to identifying 43 to 60 additional cases of eye disease per 1,000 individuals.
The eight diseases identified by the model primarily affect the retina and optic nerve, are leading causes of vision loss, and are linked to diabetes, hypertension, and cardiovascular disease. The model can also predict health outcomes such as heart attack, stroke, and kidney failure based solely on retinal imaging. Lee said, "The model has the potential to turn a simple eye exam into a powerful tool for helping to detect illness beyond the eye. It opens the door to earlier detection, more precise monitoring and potentially better outcomes for patients who might otherwise go undiagnosed until their disease is far more advanced."
The researchers adapted the artificial intelligence model by incorporating data from infrared retinal imaging and fundus autofluorescence imaging. The model trained on all three imaging types predicted the growth rate of geographic atrophy with nearly 50 percent greater accuracy than the current leading model.