From kidney stones, to heart trouble, to cancer, artificial intelligence stands to revolutionize the field of medical diagnostics by spotting signs of disease earlier than us humans are capable of. Scientists in the UK have been putting these capabilities to the test for detecting glaucoma, with the results of the trial indicating AI can pick up on progression of the disease 18 months earlier than current methods.
As the leading cause of irreversible blindness around the world, glaucoma affects more than 60 million people and is driven by the death of cells in the retina. Specialists rely on a range of techniques to diagnose glaucoma, including vision tests and measurements of the cornea and eye pressure, but it is cell death that the researchers hope to leverage for a more unequivocal picture of the disease.
Called Detection of Apoptosing Retinal Cells (DARC), the test uses a fluorescent dye that is injected into the bloodstream and makes its way to the eye, where it binds to retinal cells and lights up the ones in this process of death, known as apoptosis. These dying cells appear as bright white dots when viewed via eye exams, and by incorporating AI into the method, the researchers aimed to remove the element of human error when diagnosing the disease. Where specialists viewing the same eye scans may disagree on the severity or progression of the disease, the AI-powered algorithm, developed by scientists at University College London and Imperial College London, is hoped to offer more definitive conclusions. The team started by training it on retinal scans of healthy subjects following the injection of the dye, and then put it to the test on glaucoma patients.
This involved analyzing scans of 60 patients, 20 with glaucoma and 40 as healthy controls, during a Phase II clinical trial of DARC that used the AI to assess the health of their eyes. The subjects where then assessed 18 months later, with the AI-powered analysis proving an accurate predictor of progressive damage caused by glaucoma. Every patient with a count of white spots over a certain threshold were shown to have progressive glaucoma at the 18-month follow-up point. This was still 18 months before the condition was able to be detected by the existing gold standard retinal imaging technology, indicating that DARC shows promise as a biomarker when combined with AI. The scientists are hopeful that with further work, this new technology can become a common tool for clinicians.
“Being able to diagnose glaucoma at an earlier stage, and predict its course of progression, could help people to maintain their sight, as treatment is most successful if provided at an early stage of the disease,” says first author of the study Dr Eduardo Normando. “After further research in longitudinal studies, we hope that our test could have widespread clinical applications for glaucoma and other conditions.”