The RSNA has expanded its Annotated Library of AI Systems (ATLAS) platform to include more than 230 model cards and dataset cards across 31 imaging subspecialties.
Launched in November 2025, ATLAS is a free platform where AI researchers, developers, and healthcare providers can share and search detailed descriptions of imaging AI models and datasets using a standardized format. Cards are tagged using RadLex and RSNA content codes, and each published card receives a Digital Object Identifier (DOI) for traceability and citation.
The platform also supports meeting regulatory requirements including those of the U.S. Food and Drug Administration (FDA), RSNA said.
The platform includes an AI Card Creator tool with a template for building submissions, as well as an AI extractor tool that can pull information from existing documents to pre-fill a card.
Cards are indexed using Radiology Ontology of AI Datasets, Models and Projects (ROADMAP), which provides a controlled terminology for AI model and dataset metadata developed by RSNA's Radiology AI Data Standards Committee.
