Google better than experts in diagnosing eye disease

Researcher’s at Google’s DeepMind got together with Moorfields Eye Hospital NHS Foundation Trust to create a new Artificial Intelligence (AI) technique to review eye imaging to make recommendations for care.

The team combined two different neural networks with one network analyzing the OCT scan to map the tissue and symptoms of disease it sees and the second network examines the map to give the diagnosis and the appropriate referral. Researchers trained the network by putting together 14, 884 OCT scan volumes from 7,621 patients who had been referred to Moorfields Hospital with symptoms indicating macular pathology. They then tested the system on a different dataset and compared the network’s performance to that of eight experts reviewing the same dataset.

What they discovered when they compared the framework’s decisions to the experts’ decisions was that the framework was 94 percent accurate. It matched the two best retina specialists and performed notably higher than the other two retina specialists and all four optometrists when the doctors only used OCT scans to diagnose the patients. In addition, the AI made no clinically serious wrong decisions.

With the technology being device-independent, it can be applied to different types of eye scanners and can be applied around the world, which would greatly increase the number of patients who could benefit.*

Looking for VisiVite eye supplements? They can be found at https://www.visivite.com/shopeyevitamins.html

Speak Your Mind

*