AI is starting to diagnose diseases like any human doctor, and the latest example is AI learning to identify dozens of common eye diseases from 3D scans. Researchers from Google’s subsidiary DeepMind, University College London, and Moorfields Eye Hospital deploys deep learning to create software that identifies eye diseases (about 50), and to suggest treatment to patients.
Even though the software is not ready for clinical use, Mustafa Suleyman, head of DeepMind Health, terms this development as “groundbreaking” and said in a press statement that the project was “Incredibly exciting.” In a few years, when deployed in the healthcare industry, it could “transform the diagnosis, treatment, and management of patients with sight-threatening eye conditions… around the world.”
The deep learning software was fed with nearly 15,000 OCT (Optical Coherence Tomography) scans from some 7500 patients. The algorithms use this data to identify common patterns in the OCT scans of underdiagnosis patients, and then recommend clinical action based on the various signs of diseases that the scans show. The recommendations, however, matched diagnoses by doctors almost every time.
Dr. Pearse Keane, the ophthalmologist at Moorsfield, who was involved in the project said, “The number of eye scans we’re performing is growing at a pace much faster than human experts are able to interpret them…causing delays in the diagnosis and treatment… AI gives us the best chance of saving people’s sight.”