Creating a Brain MRI Similarity Search Engine to Improve Patient Care

Across the United States, nearly 40 million MRIs are performed each year, yet almost none are used to improve future patient care. Dr. Rauschecker’s project uses novel computational methods to leverage these massive datasets for improving the care of patients with neurological diseases, systematically organizing large sets of brain MRIs into a multidimensional feature space that is directly searchable by image content and interpretable and informative to physicians.