UCL Institute of Ophthalmology


Predicting Geographic Atrophy with DARC AI Biomarker

4 August 2021

A new study shows that DARC technology can identify eyes with rapidly enlarging atrophy up to 36 months in advance

Image showing DARC technology

The new study led by UCL IoO’s Professor Cordeiro and published in Progress in Retinal Eye Research describes a group of patients with pre-existing Geographic Atrophy. The patients were assessed with DARC only once and were then followed up to 36 months later with regular OCT imaging. There were two main results: 

  • DARC predict retinal areas that would  lead to GA in the future.
  • The level of DARC activity was able to predict the extent and rapid growth of large areas of new atrophy.

Age-related macular degeneration or AMD is the commonest cause of blindness in over-55s affecting 700,000 in the UK, 11 million in US and estimated to double in the next 25 years. Although the wet form of AMD is treatable, 90% is of the “dry” type, of which Geographic Atrophy is the advanced, currently untreatable form. A recognised challenge is to identify Geographic Atrophy early enough to avoid any vision loss. But this is hampered due to lack of meaningful disease activity measures and predictors of disease. It has also meant that new potential therapies for Geographic Atrophy need long and expensive trials. 

This study for the first time shows that DARC technology can fill this gap, helping advance development of new treatments for this catastrophically, blinding disease. 

DARC is a method that allows us to see the sick and dying cells on the retina as “WHITE SPOTS” when you take a picture of the retina. Clinical trials have shown that these DARC spots indicate disease activity accurately. This has been done using artificial intelligence algorithms to rigorously count and assess the DARC spots. Using DARC AI counts, DARC has recently been shown to predict glaucoma progression and new lesions of wet AMD.

The biomarker DARC has been developed by Professor M Francesca Cordeiro’s groups at UC: Institute of Ophthalmology and Imperial College through Wellcome Trust funding, and has recently been combined with  artificial intelligence through Novai Ltd. 

Professor Cordeiro says:

These results are very promising as they show DARC could be used as a biomarker when combined with the AI-aided algorithm in this blinding disease where we urgently need new treatments. Importantly, what that means was DARC was able to predict new Geographic Atrophy up to 36 months in advance of it occurring and this is huge - it means that DARC activity can guide the clinician into treating more intensively those patients at high risk of rapidly expanding GA. 

The biomarker DARC, developed by Professor Francesca Cordeiro’s groups at UCL Institute of Ophthalmology and Imperial College through Wellcome Trust funding, has recently been combined with  artificial intelligence through Novai Ltd. 

DARC technology is now being commercialised through Novai Ltd, a newly formed company actively looking for investment. 



  • DARC test prediction: Retinal images taken from a patient with Geographic Atrophy, showing that DARC was able to predict new areas of atrophy growth up to 36 months later.