May 13 2020
Scientists at New York Eye and Ear Infirmary of Mount Sinai (NYEE) have developed a new artificial intelligence (AI) algorithm that can precisely, quickly identify age-related macular degeneration (AMD), a principal cause of vision loss in the United States.
AMD is characterized by the deterioration of the central region of the retina known as the macula—the location of central vision—causing a blurry vision that can significantly worsen over time.
The new study published in the April/May issue of Translational Vision Science and Technology is the first to illustrate that AI technology might help doctors to estimate the risk of progression of AMD and its intensity, which can make patients seek earlier medical treatment and save their eyesight.
We are excited to have built a deep-learning form of AI that can be trained to match the performance of a human expert to accurately diagnose AMD grade and stage based on scanning retinal photographs, without using other information. This is an important step in identifying those at risk for late-stage AMD and may allow them to get quick referral to an eye specialist for timely, preventive treatment.
R. Theodore Smith, Study Lead Researcher, MD, PhD, Professor of Ophthalmology, Icahn School of Medicine, Mount Sinai
Smith continued, “This algorithm can easily be applied in the ophthalmology telemedicine landscape as the practice of medicine transforms under the impact of the COVID pandemic to embrace ‘medicine at a distance’. For example, our large ambulatory facilities can strategically place teleophthalmology kiosks with inexpensive cameras that take these retinal images to screen underserved populations for AMD.”
“The AI algorithm would instantly generate results, so patients get immediate diagnosis, and if they need additional care, they could have a same-day follow-up at a nearby ophthalmic center,” added Smith.
This may become an important and cost-effective tool for high-risk or low-income groups who may not have direct or frequent access to eye screening, as early detection is critical to preventing AMD. This will not only aid in quick diagnosis, but help to close gaps in health disparities.
R. Theodore Smith, Study Lead Researcher, MD, PhD, Professor of Ophthalmology, Icahn School of Medicine, Mount Sinai
At NYEE, researchers designed deep-learning AI screening and prediction models by making use of data from the Age Related Eye Disease Study, a long-time, wider study on AMD of more than 15 years, funded by the National Institutes of Health. Patients aged between 55 and 80 years were classified into categories for early, intermediate, advanced, late, or normal AMD.
For screening purposes, the researchers selected 116,875 color fundus photos (images capturing the eye’s inner surface) from 4,139 participants. They trained the algorithm to group them as “early,” “advanced,” “intermediate,” or “no” AMD along a 12-level severity scale to compare the results of human experts. In general, the algorithm achieved 98% precision while matching the decisions of experts.
Furthermore, the scientists at Mount Sinai combined the intensity scores with the patients’ sociodemographic clinical data (such as gender, age, and medical history like diet, diabetes or cardiac illness, and tobacco use) and other imaging data using a second algorithm to estimate AMD progression, particularly the risk of advancing to late AMD within a year or two.
The researchers trained and verified the predictive learning model on 901 patients who had disease progression within a year, 923 participants who had progression within two years, and 2,840 patients who did not show any progression within two years.
Furthermore, the AI model refined the risk of advancement to late AMD so that scientists were able to estimate the accurate type of progression of late AMD—either wet or dry. Dry AMD forms more gradually; layers of the macula get increasingly thin and lose their function. In the case of wet AMD, which is faster, abnormal blood vessels develop behind the retina and start leaking.
According to Smith, “The prediction program will produce a report that can help eye doctors counsel AMD patients on their risk for progression based on their retinal photographs and other lifestyle (diet and smoking) and demographic variables: age, gender, and medical history.”
Smith added, “The ophthalmologist can then recommend changes in modifiable factors in consultation with family and the primary care physician, and patients at high risk can be followed up with sooner.”
The proposed noninvasive technology thus proceeds in two steps: we first screen high volumes of patients in the community to find the at-risk patients with intermediate and advanced AMD for referral to an ophthalmologist, and second, we help the eye doctor manage these patients by predicting if they will develop late AMD in one to two years.
R. Theodore Smith, Study Lead Researcher, MD, PhD, Professor of Ophthalmology, Icahn School of Medicine, Mount Sinai
“This can allow screening to take place more efficiently and cost-effectively in primary care clinics, with detection of a much smaller at-risk group for referral to specialty care,” concluded Smith.
The algorithm has been tested by the NYEE for detection and staging of AMD in its eye clinics and has obtained positive results. The scientists believe that once these systems are ready for extensive use with automatic, affordable cameras at primary care facilities, patients will gain access to rapid, non-invasive screening for the blinding eye disease.