Eye2Gene: predicting genetic diagnosis from eye scans
Retina UK is excited to announce a long-awaited publication describing how artificial intelligence (AI) can be used to predict the genetic cause of an inherited retinal condition from a retinal scan alone.
Research funded by Retina UK helped pave the way for this groundbreaking innovation.
Authored by Nikolas Pontikos and colleagues from across UCL Institute of Ophthalmology and NIHR Moorfields Biomedical Research Centre, the paper is entitled “Eye2Gene: next generation phenotyping of inherited retinal diseases using multimodal imaging” and is published today (18 June 2025) in Nature Machine Intelligence.
Identifying the specific gene that causes a person’s sight loss can make a significant difference to affected individuals and families. It enables better understanding of their condition and prognosis, guides important life decisions such as family planning and potentially opens up access to existing and emerging treatments. However, genetic testing via blood sample analysis can take many months or even years, results can sometimes be confusing, and the test is not always readily offered in every clinic.
In the new publication, the researchers describe their development of an AI algorithm that can effectively be used to predict the genetic cause of an inherited retinal disease (IRD) directly from a retinal scan for up to 63 causative genes, a task that even leading human experts struggle with.
The algorithm, known as Eye2Gene, was developed using one of the world’s largest datasets of individuals with an IRD genetic diagnosis, curated over 30 years at Moorfields Eye Hospital and consisting of almost 2,500 patient records. The research team benchmarked the algorithm against human experts and validated its performance in IRD datasets from other hospitals across the world.
Eventual implementation of Eye2Gene in the clinic could vastly increase the speed and efficiency of accurate diagnosis.
Retina UK is proud to have supported the UK IRD Consortium which initially funded the Eye2Gene Principal Investigator Associate Professor Nikolas Pontikos in 2017. This support represented the foundation for this work and the creation of the collaboration from which Eye2Gene benefited.