Bengaluru: A homegrown artificial intelligence (AI) tool that can analyse retinal images of premature babies at their bedside and flag suspected cases for specialist review has received regulatory approval, potentially expanding access to screening for retinopathy of prematurity (ROP), a condition that can cause partial or complete blindness.Developed by Bengaluru-based medical technology company Remidio in collaboration with Karnataka Internet-Assisted Diagnosis of Retinopathy of Prematurity (KIDROP), a public-private partnership between the state govt and Narayana Nethralaya, the tool can analyse retinal images and identify babies who may need further evaluation. Funded by Biotechnology Industry Research Assistance Council (BIRAC), govt of India, it received approval from Central Drugs Standard Control Organisation (CDSCO) last week.What is ROP?ROP affects the retina, the light-sensitive layer at the back of the eye, in premature babies whose retinal blood vessels are not fully developed at birth. These vessels continue to grow after birth, ideally completing their development within four to six weeks. However, when this growth is disrupted, it can lead to abnormal blood vessel development and ROP.“The condition is classified into five stages based on its severity. To identify the stage, specialists have to examine the inside of the baby’s eyes. This is not easy because these babies are often extremely small and admitted to neonatal intensive care units (NICUs),” said Dr Anand Vinekar, founder and programme director of KIDROP and principal investigator of the ROP AI project. He was speaking at a World Prematurity Day event organised by Indian Retinopathy of Prematurity Society (iROP) and Narayana Nethralaya Tuesday.For retinal imaging, babies need to be carefully wrapped, and anaesthetic drops administered into their eyes, he said.Early detection and timely treatment can achieve success rates of more than 95%, according to experts. Dr Vinekar said an estimated 30% of premature babies in India develop ROP, of whom 2% to 10% may require treatment. However, with fewer than 300 ROP specialists in the country, ensuring timely screening and treatment remains a challenge.“This is where technology can bridge the gap. Trained and certified technicians can visit these locations, capture retinal images using specialised cameras and screen babies on time. Those requiring specialist attention or treatment can then be referred to appropriate facilities,” Dr Vinekar said.Dushyantsinh Anopsinh Jadeja, senior vice-president, AI and Software, Remidio, said the tool has been designed to function offline, making it suitable for remote locations where internet and mobile connectivity may be unreliable.Trained technicians capture retinal images using specialised cameras, while the AI software analyses them to help identify suspected cases and guide referrals and follow-up. According to research submitted to the Centre, the AI model has demonstrated 100% sensitivity in identifying babies with Type 1 ROP who require treatment.
