Using AI to identify retinopathy of prematurity

A new study published in Lancet Digital Health has described a deep learning artificial intelligence (AI) model capable of identifying retinopathy of prematurity (ROP) from retinal images.

The research, which was led by University College London and Moorfields Eye Hospital, involved training a deep learning model on 7414 images of the eyes of 1370 newborn babies.

The babies had been admitted to Homerton Hospital, London and assessed for ROP by ophthalmologists.

The model was then assessed on its ability to identify ROP in a further 200 images and compared with the assessments of senior ophthalmologists.

Further validation was carried out by applying the tool to datasets from the US, Brazil and Egypt.

The AI tool was found to be as effective as senior ophthalmologists in discriminating between healthy eyes and those with ROP.

Lead author Dr Konstantinos Balaskas highlighted that ROP is the leading cause of childhood blindness in middle income countries and the US.

He added that as many as 30% of premature babies in sub-Saharan Africa have some degree of ROP.

“As it becomes more common, many areas do not have enough trained ophthalmologists to screen all at-risk children; we hope that our technique to automate diagnostics of ROP will improve access to care in underserved areas and prevent blindness in thousands of newborns worldwide,” Balaskas highlighted.

Main image: an eye with retinopathy of prematurity (left) and how the AI tool identifies the diseased area (right).