Researchers proposed a Siamese neural network to simulate the diagnosis process for primary open-angle glaucoma (POAG) in a new study published in Ophthalmology Science. A leading cause of irreversible blindness in the U.S. and the most common form of glaucoma among African-Americans and Hispanics, POAG requires early treatment and intervention to prevent vision loss. One form of diagnosis is through expertly graded optic disc photography. Senior author Dr. Yifan Peng and colleagues have developed an automated model, POAGNet, which uses artificial intelligence to identify POAG from optic disc photographs. The findings showed that POAGNet achieved high diagnosis accuracy and is superior to the current state-of-the-art method. Researchers believe this new deep learning model has great potential for enhancing POAG diagnosis following extensive validation across multiple and diverse image data sets.
- Highlights