HARBOR: Health Asset Registry for Bioimaging, Orchestration, & Reuse

HARBOR is a national platform designed to enable medical imaging algorithm developers to source, process, and train on large volumes of high-quality, deidentified radiology and pathology images. Every hospital in America holds large archives of X-rays, CTs, MRIs, and pathology images, yet researchers struggle to get the high-quality, representative data they need. Available datasets are fragmented, costly, poorly curated, and secured behind hospital firewalls.  

HARBOR is investing in federated learning technologies that allow researchers to reach deidentified data that stays secured inside hospital systems, which protects patient privacy and leaves hospitals in control. As a platform for innovation in computational modeling, HARBOR will allow collaboration between physician scientists, engineers, product developers. An ecosystem of service providers will allow teams to focus less on duplicative annotations and more on engineering effective diagnosis solutions for patients.

HARBOR is working directly with U.S. Food and Drug Administration’s Office of Science and Engineering Laboratories to incorporate regulatory science tools into the platform that check data quality, build training and testing cohorts, and evaluate model performance. These tools will contribute to faster product development, bringing better care to patients sooner.  

Back to Award Directory