An affordable, smartphone-based camera can reportedly help doctors identify premature infants in need of treatment for retinopathy of prematurity (ROP), according to a new study funded by the National Eye Institute (NEI). The study also demonstrated that an artificial intelligence (AI)-based system was effective in assessing those same smartphone images to accurately flag babies requiring ROP care.
Timely treatment of ROP in newborns can prevent retinal damage and permanent vision loss or blindness. The findings could potentially improve access to ROP screening and treatment for thousands of babies worldwide via telemedicine approaches to care.
ROP is often diagnosed by ophthalmologists using specialized, wide-field cameras that can view a large portion of the retina at the back of babies’ eyes. However, each of these cameras can cost up to $150,000 and require trained technicians to take and evaluate the images. Narrow-view cameras, such as those used in this study, are significantly less expensive (at $500 to $1500), but they produce images that show a smaller portion of the retina.
“We found that even though smartphone images captured less information than a wide-field camera, both ROP clinicians reading the images remotely and the AI algorithm were able to accurately identify all babies with severe ROP,” said Peter Campbell, M.D., Oregon Health Sciences University, Portland, the lead author of the study. “If the results of this study can be replicated in other telemedicine programs, it may be possible to rapidly improve access to care in regions where traditional ROP cameras are not available or affordable.”
The study enrolled 156 infants in an ROP telemedicine program at Aravind Eye Hospital in India. The babies were screened for ROP using a conventional wide-field camera, as well as one of two smartphone-based imaging devices: the Make-In-India Retcam or the Keeler Monocular Indirect Ophthalmoscope. The AI was not as capable at diagnosing mild forms of ROP as clinicians, indicating improvements may still be made.