As with dermatology, ophthalmology is another specialty where the examination of the eye, including the fundus is a huge part of the screening and diagnosis for a variety of eye conditions. As such, if there is a way to create high-quality digital pictures of the fundus (which there is!) then AI algorithms can be used to identify and monitor some of these conditions.
The FDA has approved the first autonomous AI system to screen for diabetic retinopathy. This system, IDX-DR by Digital Diagnostics, has shown to be very accurate in detection of diabetic retinopathy on fundus photographs. When combined with a device to take photo of the fundus, the algorithm can detect more than mild diabetic retinopathy. The algorithm analyzes images taken with a retinal camera and uploaded to a cloud server. Within minutes the software provides doctors with a binary result, either indicating that more than mild diabetic retinopathy is present and that the patient should be referred to an eye care professional, or that the screen is negative and should be repeated in 12 months. The software is notable in that it was the first AI-based diagnostic system to be authorized by the FDA for commercialization in the US that can provide a screening decision without the need for clinician interpretation.
Google also has developed a system that can auto-detect diabetic retinopathy from a fundus photograph. In 2 validation sets of 9963 images and 1748 images, at the operating point selected for high specificity, the algorithm had 90.3% and 87.0% sensitivity and 98.1% and 98.5% specificity for detecting referable diabetic retinopathy. The study’s conclusion was that a deep learning algorithms had high sensitivity and specificity for detecting diabetic retinopathy and macular edema in retinal fundus photographs.
All of this shows that AI can be a front line screener for diseases that have a visual diagnosis component such as dermatology, ophthalmology, radiology, etc and augmenting the decision making of physicians during the diagnostic process. This can lead to increased accuracy as well as more efficiencies.