MR-RATE: A Vision-Language Foundation Model and Dataset for Magnetic Resonance Imaging

This is the MR-RATE-atlas repository, part of the MR-RATE dataset release. It contains atlas-registered MRI volumes in which all imaging sequences within each study have been spatially normalized to a standard atlas-space. For full dataset details, native-space MRI volumes, radiology reports, metadata, and data splits, please refer to the MR-RATE repository. To explore, download, and work with the dataset see Dataset Organization & Getting Started.

[2026-04-19] Some studies initially missing due to processing job failures have been uploaded. See the backfill guide for details.

Atlas-Registered MRI Volumes

Within each study in native-space, a T1-weighted MRI volume is selected as the center modality and is rigidly registered to the MNI152 (ICBM 2009c Nonlinear Symmetric) atlas using ANTs. All other MRI volumes acquired in the same study referred to as moving modalities, and the brain mask and defacing mask of center modality are then transformed to atlas-space by composing the native-to-center transforms (from MR-RATE-coreg) and center-to-atlas registration transforms, bringing all sequences of a given study into a standardized coordinate space that enables group-level analyses and cross-patient comparisons.

This repository provides:

• Atlas-registered center modalities (in NIfTI format): All T1-weighted center modality MRI volumes, and corresponding brain masks and defacing masks warped to atlas-space
• Atlas-registered moving modalities (in NIfTI format): All moving modality MRI volumes transformed to atlas-space
• Registration transforms (in .mat format): The ANTs registration transform files mapping each center modality to the atlas, enabling users to apply them in their custom workflows

Study folders are zipped to comply with Hugging Face's per-repository file count limits.

Citing Us

When using this dataset, please consider citing the following related papers:

Coming soon

Ethical Approval

This study was approved by the Clinical Research Ethics Committee at Istanbul Medipol University (E-10840098-772.02-6841, 27/10/2023). All MRI volumes, metadata, and radiology reports were fully anonymized prior to analysis to protect patient privacy.

License

We are committed to fostering innovation and collaboration in the research community. To this end, all elements of the MR-RATE dataset are released under a Creative Commons Attribution–NonCommercial–ShareAlike (CC BY-NC-SA) license.

This licensing framework ensures that our contributions can be freely used for non-commercial research purposes, while also encouraging contributions and modifications, provided that the original work is properly cited and any derivative works are shared under similar terms.

For commercial inquiries related to MR-RATE, please contact: contact@forithmus.com.

Acknowledgements

This project is conducted by Forithmus and the University of Zurich, in collaboration with NVIDIA and Istanbul Medipol University.

We are grateful to NVIDIA for their support, which made this work possible. We also sincerely thank Istanbul Medipol University for their support and for providing the data used in this project. High-performance computing resources were provided by NVIDIA and the University of Zurich ScienceCluster. We would also like to thank the following individuals from NVIDIA for their contributions to the development of MR-RATE: Marc Edgar, Daguang Xu, Dong Yang, Yucheng Tang, Can Zhao, Andriy Myronenko, and Pengfei Guo.

This collaboration represents an important step toward the long-term mission to make high-quality medical intelligence accessible worldwide.