The U.S. National Institutes of Health has published DeepLeision, a dataset of images of lesions to foster the development of algorithms that can spot lesions in computed tomography (CT) scans. The dataset includes 32,000 CT images from across the body, such as liver tumors, and from over 4,000 patients. Developing algorithms that can spot lesions in CT scans can help radiologists spot them more quickly and measure them more accurately, allowing radiologists to devote more time to studying them.
Image: James Heilman, MD.