NIHR researchers create A.I. tool to detect retinopathy of prematurity in new-born babies

28 April 2023

A team of researchers led by NIHR Moorfields BRC investigator Dr Konstantinos Balaskas has published a study showing how artificial intelligence (A.I.) can be used to identify infants at risk of retinopathy of prematurity (ROP). This condition, which commonly arises following premature birth, can lead to permanent sight-loss if left untreated. 

The deep-learning model developed by the team could be of particular use in countries that have limited neonatal and ophthalmology services. It can diagnose the condition as accurately as specialised ophthalmologists using a simple eye scan, which can be taken by a nurse, making referral for treatment much simpler.

Newborn baby in a cot, black and white image. Credit: Photo by charlesdeluvio on Unsplash

The study – published last week in The Lancet Digital Health – was conducted by an international team of scientists and supported by the NIHR Moorfields Biomedical Research Centre at Moorfields Eye Hospital NHS Foundation Trust and UCL Institute of Ophthalmology. 

ROP is the leading cause of childhood blindness in wealthy and middle-income countries. As survival rates for premature birth improve globally, its prevalence is likely to increase. The condition already affects up to 30% of premature babies born in sub-Saharan Africa, and while treatment is relatively straightforward, detection rates are low due to the lack of specialist care. This new method of automating ROP diagnosis could help to increase access to treatment around the world and prevent sight loss in thousands of new-born babies. 

The ability of the A.I. tool to work across different ethnicities is particularly significant.  Although currently optimised for the UK population, it has been tested using data from diverse populations in the US, Brazil and Egypt. It has been designed so it can be further optimised for additional populations using data from specific groups or countries. As the tool is ‘code-free’, it can refined by people with little or no coding experience. 

Commenting on the results, first author Dr Siegfried Wagner said:

“We are now further validating our tool in multiple hospitals in the UK, and are seeking to learn how people interact with the A.I.’s outputs to understand how we could incorporate the tool into real-world clinical settings.”

Two retinal photographs side by side. On left: normal image; on right: region affected by ROP highlighted by A.I. model

About retinopathy of prematurity (ROP)

ROP is a condition that mainly affects premature babies. It occurs when abnormal blood vessels grow in the retina – the layer of nerve cells at the back of the eye that convert light into electrical signals to the brain. These vessels can leak or bleed, damaging the retina and potentially leading to retinal detachment. Across the world, as many as 50,000 children are living with sight loss due to ROP. 

Milder forms of ROP simply need monitoring, but more severe cases require prompt treatment to prevent permanent sight loss. This includes laser treatment, injections of anti-VEGF drugs and surgery (when the retina is fully or partially detached). 

Find out more

Media coverage