Data for AI
Advance your research with the most reliable data in medical imaging. RSNA curates ground-truth datasets that are diverse, fully anonymized and triple-labeled by radiologists—ensuring transparency and trust in every pixel.
Discover AI Models & Datasets with RSNA ATLAS
RSNA ATLAS model card and dataset hub is a searchable catalog that makes it easy for researchers and developers to find and compare AI models and datasets.
Explore ATLAS
Models and datasets are presented in a standardized card format that highlights their key attributes, so you can quickly find and evaluate AI resources.
Publish an ATLAS data card, ensuring your AI model or dataset is discoverable, easy to evaluate and ready to be used.
ATLAS provides transparent, standardized documentation to support sharing and evaluation, potentially including regulatory review.
RSNA’s MIRA Data Repository
RSNA’s data repository, Medical Imaging Resource for AI (MIRA), supports AI-focused research in medical imaging with high-quality datasets.
Explore MIRA’s Datasets
Collected for RSNA AI challenge competitions, annotated by subspecialist radiologists.
MIRAdatasets reflect a diverse range of imaging sources to mitigate bias.
MIRA datasets are completely deidentified using the RSNA Anonymizer tool to protect patient privacy.
Join Our AI Challenges
RSNA’s AI challenges accelerate the development of AI tools radiologists can use to enhance the efficiency and accuracy of diagnoses. Each year, RSNA invites you to collaborate with us and other radiological organizations from around the world to train AI systems, enable benchmarking, improve patient outcomes and power additional research.
Explore the Latest ChallengeBuild & Train with Trusted Data
Train AI models using expert-verified, clinically trusted data.
Benchmark Your Model
Test your model’s accuracy and readiness for clinical use.
Improve Diagnostic Precision
Transform patient care by solving real-world diagnostic challenges.
Power Additional Research
Accelerate future breakthroughs by expanding RSNA’s MIRA repository.
Data Resource Papers
Explore the performance of models developed during past AI challenges.
Medical Imaging and Data Resource Center (MIDRC)
MIDRC is a multi-institutional, NIH-funded data commons designed to accelerate AI innovation in medical imaging. Hosted by the University of Chicago and co-led by RSNA, MIDRC’s open-access repository of medical imaging data houses more than 500,000 imaging studies.
Intake
Curated from diverse sources, including RSNA’s COVID-19 database, data is anonymized and validated before being structured for use.
Annotation & Labeling
MIDRC’s infrastructure can be used to annotate images and metadata in a semi-automated process to support AI model training.
Access & Discovery
Researchers can explore, filter, and retrieve imaging data across repositories.
We’re Here to Support Your Research Efforts
We encourage you to contact us for guidance. Email our team at informatics@rsna.org