Artificial intelligence: The future of eye screening
CERA researchers have developed a cutting-edge AI screening tool to catch diabetic retinopathy and other common eye diseases.
Diabetic retinopathy affects over 30% of patients with diabetes. It damages the tiny blood vessels in the retina, and can lead to blindness if left untreated.
Unfortunately, while early diagnosis and treatment can prevent almost all vision loss from diabetic retinopathy, around 50% of people don’t know they have it until it’s reached an advanced stage.
That’s why regular eye screening is essential. Early detection of diabetic retinopathy could be greatly improved if screening could be done in more GP and endocrinology clinics, says Professor Mingguang He, Principal Investigator in Ophthalmic Epidemiology at CERA.
“But the challenge is that many GPs don’t have the technical confidence to examine the retina – the back of the eye,” he says.
The power of artificial intelligence
Artificial intelligence (AI) could help change this. Professor He has developed an innovative AI screening tool, designed to give GPs and endocrinologists a quick and simple way to detect signs of eye disease.
Now, he’s working with research fellow Dr Jane Scheetz to test the clinical integration of the AI screening tool in GP, endocrinology and Indigenous health settings.
“The AI system has been trained using 70,000 images of the retina, and is now highly accurate in determining whether someone has changes caused by diabetes – more accurate than a human,” says Dr Scheetz.
“The primary aim is to screen for diabetic retinopathy, but the algorithm can also look for glaucoma, age-related macular degeneration and cataract – so it can screen for the four most common blinding eye diseases at once.”
The screening takes just seconds. After taking a photo of the back of a patient’s eye, the AI system scans for signs of disease, and prints out a report identifying if the patient should be referred to a specialist for further assessment and treatment.
“Over the last 18 months we’ve been trialling the system in real world settings – like the Endocrinology Clinic at St Vincent’s Hospital, Box Hill Hospital, Derbarl Yerrigan Health Services in Perth and now in an Aboriginal community in South Australia,” Dr Scheetz says.
Making eye screening more accessible
This work has inspired Professor He to design a fully automated, operator-free version of the screening tool as the exciting next phase.
“In the future, screening of eye diseases could be as easy as taking a portrait in a photo booth,” he says.
The hope is that this technology will increase accessibility to diabetic retinopathy screening for all Australians, leading to earlier diagnosis and treatment. In particular, it could make a huge difference in regional and remote areas where eye health services are lacking.
“Artificial intelligence has the potential to close the gap in eye care services, considerably increase early diagnosis of the four most common blinding eye diseases and reduce the burden of vision loss in the Australian communities that need it the most.”