Poetically described as the windows to the soul, our eyes can also reveal the condition of our hearts. Small changes in the fine blood vessels supplying the retina, the layer of light-sensitive cells in our eyes, are linked to characteristics of the heart associated with cardiovascular disease. These subtle signs can be revealed with retinal scans (pictured), a technique routinely performed by opticians to diagnose diseases of the eyes. Using machine learning techniques, researchers devised an artificial intelligence system capable of estimating heart properties and predicting the risk of heart attacks based on retinal scans: after training with over 5000 patient images, the system could identify patients likely to experience a heart attack with around 70% accuracy. With further refinements, this tool could mean that, in future, checking your vision will also flag up elevated risk of serious heart conditions, improving the chances of earlier detection and more effective treatments.
Written by Emmaunelle Briolat
Research by Andres Diaz-Pinto and colleagues, Centre for Computational Imaging and Simulation Technologies in Biomedicine, School of Computing, University of Leeds, Leeds, UK
Image copyright held by the UK Biobank
Research published in Nature Machine Intelligence, January 2022
You can also follow BPoD on Instagram, Twitter and Facebook
