Data

RSNA provides ground-truth, de-identified imaging datasets—validated by radiologists and sourced globally—to help you build and evaluate trustworthy AI.

Visit RSNA’s Medical Imaging Resource for AI (MIRA) to access datasets.

Visit the repository

Datasets you can trust

Collected worldwide Expert-verified (multi-reader) Completely anonymized (PHI removed) Bias-aware ingestion & transparent metadata

Help build quality data: join the AI Challenge

Contribute labeled, de-identified cases and benchmark models alongside the community through RSNA’s AI Challenge.

Learn more

Data Resource Papers

OPTIMAM Mammography Image Database (2020) – Data Resource
→
 EMory BrEast imaging Dataset (EMBED) (2023) – Data Resource
→
 ADMANI: Annotated Digital Mammograms (2022) – Data Resource
→
 Scottish Medical Imaging Archive (2023) – Data Resource
→
 RSNA Pulmonary Embolism CT Dataset (2021) – Data Resource
→
 RSNA Cervical Spine Fracture CT Dataset (2023) – Data Resource
→
 Augmentation of the RSNA PE CT Dataset (2023) – Data Resource
→
 Duke Lung Cancer Screening (DLCS) Dataset (2025) – Data Resource
→