AI spots a treatable weakness in ovarian cancer from ordinary tissue images
Some ovarian cancers have broken DNA repair machinery, a weakness that certain treatments can target. Finding it usually takes genomic testing. Researchers at King’s College London, led by Dr Heba Sailem, built an AI model called HRDPath that instead looks at high resolution microscope images of tumour tissue. In a study published in Nature Communications, it correctly flagged about 91% of tumours with the repair defect when it could look at several images from the same patient. The model also surfaced visual clues, such as rounder cell nuclei and densely packed cancer cells. The team hopes it could lead to a low cost, widely accessible test that helps doctors choose treatment.