Novel artificial intelligence technology improves eye disease diagnosis

Non-invasive eye scans provide a detailed, three-dimensional view of the eye’s interior to help diagnose vision-threatening conditions, but the manual review process is time-consuming and prone to human error.
Researchers at Washington University School of Medicine in St. Louis, in collaboration with the University of Washington and Genentech, developed an AI system called OCTCube-M, which can accurately interpret retinal scans to identify eight different retinal diseases more precisely.
The AI system, OCTCube-M, not only enhances the detection of eye diseases such as age-related macular degeneration but also predicts the progression rate of geographic atrophy, the severe form of macular degeneration.
The study demonstrates that the AI model trained on 3D images, in combination with two other imaging techniques, significantly improves disease prognosis and could potentially streamline clinical trials for new therapies.
Besides eye diseases, the model can predict systemic health risks such as heart attacks, strokes, and kidney failure, based on retinal imaging, potentially transforming eye exams into broader health diagnostics tools.
The researchers aim to further improve OCTCube-M by training it with larger datasets that include more diverse patient populations, diseases, and imaging types.