AI and computer vision can work together to save human vision thanks to a new automated system that can identify more than 50 eye diseases with better accuracy than doctors. Deep learning software is able to detect dozens of common eye diseases from 3D scans and recommend treatment options.
How AI Uses 3D Scans to Detect Eye Diseases
Deep learning uses algorithms to find common patterns in data. Distinct machine learning algorithms first create maps of optical coherence tomography scans. Those maps are assessed by even more algorithms that can indicate the likelihood of disease as a percentage.
For the system, the data consists of 3D scans of patients’ eyes made using optical coherence tomography (OCT). The scans take about 10 minutes. During the scan, near-infrared light is bounced off the interior surfaces of the eye. A 3D image of tissue is built to assess the eye’s health. OCT scans help to identify eye disease early and save a patient’s sight.
The AI computer vision software is trained with thousands of OCT scans from thousands of patients. Included in the data set are the diagnoses from human doctors. The AI software first learns how to identify the anatomical elements of the eye, and then recommends clinical action based on the signs of disease shown by the scans. The software made the same recommendations as a panel of eight doctors 94% of the time.
Advances in AI are pushing us closer to a reality where software is no longer applied and interpreted by a doctor, but instead makes decisions on behalf of humans. The FDA recently approved the first AI-powered program that diagnoses disease without human oversight. To avoid error, systems use multiple algorithms, and instead of providing a single explanation, offer several. The system also shows how it labeled the parts of the eye.
Benefits of Detecting Eye Disease with AI
Hundreds of millions of people around the world live with some form of sight loss, of which eye disease is the biggest cause. OCT scans are excellent at spotting eye disease, but the bottleneck is having the data interpreted. Algorithms could help triage patients by directing doctors to those most in need of care, making the technology incredibly beneficial.
Diagnosing and treating eye conditions early gives doctors the best chance of saving a person’s eyesight. With further research, the technology could lead to greater consistency and quality of care for patients with eye problems.
Learn more about the technologies being leveraged to advance patient care by visiting our Vision in Life Scienceseducational section.