Leukocoria, also known as white puppilary reflex, is a symptom of a number of diseases, including retinoblastoma, Coats’ disease, and congenital cataracts. When light enters the eye at certain angles in people with leukocoria, a white reflection from the retina can be seen. While easy to spot with ophthalmic equipment, the condition, and the underlying disease, often goes unnoticed for much too long. Retinoblastoma, for example, is detected via identification of leukocoria during general physicals in fewer than 10% of cases.
Researchers at Baylor University have now developed an app that can scan through the photos of parents’ smartphones to automatically spot signs of leukocoria in their children.
Called CRADLE (ComputeR Assisted Detector LEukocoria) the app has been trained to identify white spots within everyday photos, and the effectiveness of the app, including its sensitivity and specificity, was evaluated by scanning through over 50,000 photos of kids, including of those that were captured before their diagnoses.
The app successfully identified leukocoria in 80% of the children who had related diseases, and it did so in photos that were captured an average of 1.3 years before diagnosis. This is pretty remarkable, and will certainly help to spot a number of diseases, including cancer, earlier than ever before. The technology could help to save thousands of children from the potentially terrible consequences of undiagnosed disease.
Study in Science Advances: Autonomous early detection of eye disease in childhood photographs
Via: Baylor University