Baylor University researchers have developed a free smartphone app that uses machine learning to flag eye disease from personal photos, as much as a year ahead of clinical diagnosis.
The White Eye Detector sifts through casual portraits on a phone for the appearance of a white reflection in the pupil—a symptom of leukocoria, which could indicate the presence of retinoblastoma or other eye disorders.
The team applied the app to nearly 53,000 photos of 40 children, half with healthy eyes and half diagnosed with a leukocoria-related disorder; ages varied from birth to several years old, allowing the detector to determine the age at which the disease developed.
The app identified leukocoria in photos taken an average of 1.3 years before diagnosis in 16 out of 20 children with eye disease.
The detector also spotted leukocoria in subjects with retinoblastoma more than nine months before diagnosis, on average.
From IEEE Spectrum
View Full Article
Abstracts Copyright © 2019 SmithBucklin, Washington, DC, USA