- Google’s sister company, DeepMind, has developed an artificial intelligence system that can identify 50 eye diseases.
- DeepMind takes pride that the AI can detect an eye condition from retinal scans with 94% accuracy.
- With the growing rate of people needing an expert advice about their eye health, AI can help in early diagnoses.
Alphabet’s London-based DeepMind issued a progress report on Monday about the study of using artificial intelligence to diagnose eye conditions.
As published by Nature Medicine, the study reports that DeepMind, in collaboration with London’s Moorfields Eye Hospital, has prepared its algorithms to identify over 50 vision-threatening conditions accurately as specialists do. It can recommend the best course of action for patients and can identify who needs urgent care by looking at 3D retinal scans.
For two years, numerous historic and anonymized eye scans were used to identify diseases leading to sight loss. The AI system was able to correctly diagnose an eye condition with an accuracy of 94 percent.
AI’s Role in Health Care
In June, Babylon Health reported that the AI they developed can perform better than humans when given the same test required of would-be general practitioners in Britain. In March, researchers discovered that machine learning can categorize heart anatomy on an ultrasound scan better than echocardiographers.
AI is now being used in Europe to help emergency call dispatchers to recognize heart attack situations.
With the complexity of eye disease detection and the prevailing eye problems of the aging population, AI gets a chance to assist.
In a statement, Pearse Keane, a consultant ophthalmologist at Moorfields, said: “The number of eye scans we’re performing is growing at a pace much faster than human experts are able to interpret them.”
“There is a risk that this may cause delays in the diagnosis and treatment of sight-threatening diseases, which can be devastating for patients,” Keane added.
AI can help in early diagnoses and consequently, early treatment could be given to patients. Thus, eye problems will be lessened if not controlled.
“It gives us the best chance of saving people’s sight,” Keane said.
Though DeepMind’s AI was trained using one particular type of eye scanner, it’s compatible with any model and can still be used in the future with no hardware limits and even when a scanner is upgraded.
The AI can give details to doctors on how it arrived at a certain conclusion. This will help doctors to study whether it made the right assessment before giving the treatment.
AI should go through clinical trials and get regulatory approval first before it can be used in hospitals to diagnose actual patients.
“As optometrists are often the first port of call for people with the symptoms of eye disease, we are very excited about the potential that AI has to assist them in helping patients,” said Martin Cordiner, head of research at the UK’s College of Optometrists.