Researchers have recently shown that together, AI (artificial intelligence) software and a smartphone camera can offer a convenient screening system for diabetic retinopathy (DR). Coming from the University of Michigan Kellogg Eye Center, this work used high definition retinal images from a smartphone-mounted device to determine whether a patient needs to be referred to an ophthalmologist.
DR is a fairly common condition, affecting nearly one-third of diabetic adults over the age of 40 as per the CDC. Early symptoms can be subtle and overlooked, but the disease can cause permanent vision loss if not treated in a timely manner. With 4.2 million adults living with DR and 655,000 having vision-threatening DR, physicians are putting emphasis on finding simplified screening procedures to detect the disease earlier.
“The key to preventing DR-related vision loss is early detection through regular screening,” said the study’s lead author Yannis Paulus, M.D., and Kellogg vitreoretinal surgeon. “We think the key to that is bringing portable, easy-to-administer, reliable retinal screening to primary care doctors’ offices and health clinics.”
Paulus and his colleagues at Kellogg developed the tool that transforms a normal smartphone camera into a retinal imaging device. The project, known as CellScope Retina, received funding from the University of Michigan’s Translational Research and Commercialization for Life Sciences Innovation Hub in 2016. This program facilitates the development of innovative concepts that have great potential in improving human health.
This latest research makes use RetinaScope, the newest model of the team’s retinal-imaging device. This smartphone-paired deice offers an inexpensive and portable means of screening for DR that requires no professional training.
“Traditional retinal cameras are expensive, large, immovable and require special training to operate, whereas RetinaScope is a smartphone-based platform that is cheap, handheld and easy to use with no required training,” Paulus explained.
Not only does RetinaScope cut costs and improve accessibility to retinal-imaging technology, but its AI software greatly reduces the time it takes to analyze these images as well. The software platform used by the researchers is called EyeArt, and is developed by the California-based AI company Eyenuk.
“It can take two to seven days for an ophthalmologist to interpret the images,” Paulus said. “To make screening truly accessible, we need to provide on-the-spot feedback, taking the photo and interpreting it while the patient is there to schedule an eye appointment if necessary.” He went on to state that the AI platform “can enhance and review images and provide automated grading of lesions present in DR, indicating which lesions require referral to an ophthalmologist for follow-up.”
In their study, the researchers collected data from 69 diabetic adults, including results from previously recorded dilated slit-lamp fundus examinations, a technique used to detect optic abnormalities.
RetinaScope was used to image the patients’ retinas after pupillary dilation, and these images were then analyzed by EyeArt AI-software. This platform then decided whether or not the patient needed to see an ophthalmologist for a follow-up due to DR.
These same images were also analyzed by two professionals trained in detecting DR to test the efficacy of RetinaScope. The team found that slit-lamp evaluation confirmed referral-warranted diabetic retinopathy (RWDR) in 76.8 percent of patients, and that the automated system did so with 73.3 percent specificity. The professionals were found to have much lower specificities, finding RWDR at 40 percent and 46.7 percent.
“This is the first time AI used on a smartphone-based platform has been shown to be effective when compared to the gold standard of clinical evaluation,” said Paulus. The team is optimistic about these results, and is striving to improve the system further and achieve FDA clearance.
— Michigan Medicine (@umichmedicine) April 29, 2019
Source: University of Michigan Health