metadata
title: ReXGradient-160K Dataset
license: other
license_name: rexgradient
extra_gated_prompt: >
## ReXGradient-160K Non-Commercial Data Access and Use Agreement
By accessing the ReXGradient-160K Data Repository, as hosted by President and
Fellows of Harvard College on behalf of Harvard Medical School ("Harvard"),
you ("YOU" or "YOUR") agree to the following terms of this agreement
("Agreement"):
1. Acceptance of this Agreement
By downloading or otherwise accessing the data available from the
ReXGradient-160K Data Repository ("Data"), which Data is available for
non-commercial research and non-clinical purposes only, YOU (i) accept and
agree to the terms of this Agreement, and (ii) further acknowledge that YOU
may have an obligation to confirm that YOUR use of the Data under these terms
is consistent with YOUR obligations to YOUR institution.
2. Attribution
Data made available via the ReXGradient-160K Data Repository was provided to
Harvard by Gradient Health. YOU agree to recognize (i) the ReXGradient-160K
Data Repository and (ii) the contribution of Gradient Health as the source of
the Data in all written, visual, or oral public disclosures concerning YOUR
research using the Data, as appropriate in accordance with scholarly
standards.
3. Use of the Data
Use of the Data is subject to YOUR compliance with all of the terms and
conditions of this Agreement. No commercial use of the Data is permitted.
Representations and Covenants:
A. YOU represent and covenant that:
1. YOU are not bound by any pre-existing legal obligations or other applicable
laws that prevent YOU from downloading or using the Data;
2. YOU will access and use the Data in compliance with all applicable laws,
rules, and regulations, as well as all professional and ethical standards
applicable to scientific research and YOUR research project, including without
limitation, all applicable requirements pertaining to human subjects research
and animal research;
3. YOU agree not to use the Data in connection with the diagnosis or treatment
of human subjects.
4. YOU agree to establish appropriate administrative, technical, and physical
safeguards to prevent unauthorized use of or access to the Data and comply
with any other requirements relating to safeguarding of the Data that the
ReXGradient-160K Data Repository may require from time to time;
5. YOU will only use and download that portion of the Data that is necessary
for use in YOUR research project; downloading of the entire ReXGradient-160K
Data Repository or portions beyond what is needed for YOUR research project is
strictly prohibited. YOU will only share the Data with those authorized
employees, fellows, students and agents who have a need to access such Data
for purposes of working on YOUR research project, and whose obligations of use
are consistent with the terms of this Agreement.
6. YOU will retain control over the Data YOU access and download.
7. YOU will not, and will not permit any third party to, at any time, directly
or indirectly: (i) disclose, distribute, release, sublicense, sell, rent,
lease, loan, or otherwise grant access to the Data to any third party, except
other users expressly authorized by Harvard to access and use the Data; (ii)
publish or display any compilation or directory of the Data; or (iii)
knowingly use the Data in any manner or for any purpose that infringes,
misappropriates, or otherwise violates any intellectual property right or
other right of any person, or that violates any applicable law.
8. YOU will use the Data only for scientific research purposes and shall not
use the Data for any commercial or clinical activities;
9. YOU will regularly check the ReXGradient-160K Data Repository for any
updates to Data pertinent to YOUR research project, and will only use the
updated version of such Data, and will promptly destroy all outdated Data in
YOUR possession; and
10. YOU shall not use the Data except as authorized under this Agreement.
B. YOU further covenant that YOU will not:
1. obtain information from the Data that results in YOU or any third
party(ies) directly or indirectly identifying any research subjects with the
aid of other information acquired elsewhere;
2. produce connections or links among the information included in Harvard's
datasets (including information in the Data), or between the information
included in Harvard's datasets (including information in the Data) and other
third-party information that could be used to identify any individuals or
organizations, not limited to research subjects;
3. extract information from the Data that could aid YOU in gaining knowledge
about or obtaining any means of contacting any subjects already known to YOU;
or
4. use the Data, either alone or in concert with any other information, to
make any effort to identify or contact individuals who are or may be the
sources of Data.
4. Disclaimer
THE DATA IS PROVIDED "AS IS" AND "AS AVAILABLE" AND WITH ALL FAULTS AND
DEFECTS. HARVARD MAKES NO, AND HEREBY DISCLAIMS ALL, REPRESENTATIONS AND
WARRANTIES OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING, BUT NOT LIMITED TO,
WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE, OR THAT THE
USE OF THE DATA WILL NOT INFRINGE OR VIOLATE ANY PATENT, COPYRIGHT, TRADEMARK
OR OTHER THIRD PARTY RIGHT, AND ANY WARRANTIES IMPLIED BY ANY COURSE OF
PERFORMANCE OR DEALING.
WITHOUT LIMITING THE FOREGOING, HARVARD MAKES NO WARRANTY AS TO THE ACCURACY,
COMPLETENESS, RELIABILITY, ORIGINALITY, AVAILABILITY, OR SECURITY OF THE DATA,
THAT THE DATA WILL MEET YOUR REQUIREMENTS OR PROVIDE PARTICULAR BENEFITS, OR
THAT THE DATA AND ANY RELATED FILES WILL BE FREE FROM DEFECTS, ERRORS, VIRUSES
OR OTHER HARMFUL COMPONENTS. YOUR USE OF THE DATA IS SOLELY AT YOUR OWN RISK.
5. Limitation of Liability
IN NO EVENT SHALL HARVARD BE LIABLE UNDER CONTRACT, TORT, STRICT LIABILITY,
NEGLIGENCE OR ANY OTHER LEGAL THEORY WITH RESPECT TO THE DATA (I) FOR ANY
DIRECT DAMAGES, OR (II) FOR ANY LOST PROFITS OR SPECIAL, INDIRECT, INCIDENTAL,
PUNITIVE, OR CONSEQUENTIAL DAMAGES OF ANY KIND WHATSOEVER.
6. Liability
Except to the extent prohibited by law, YOU assume all liability for damages
which may arise from YOUR use, storage, download, disclosure, or disposal of
the Data. Harvard will not be liable to YOU or any other person or entity for
any loss, claim, liability, or demand made by YOU, or made against YOU by any
other party, or otherwise relating to or arising in connection with the use of
the Data by YOU, except to the extent permitted by law when caused by the
gross negligence or willful misconduct of Harvard. No indemnification for any
loss, claim, damage, or liability is intended or provided by either party
under this Agreement.
7. Governing Law; Venue
This Agreement shall be governed by and interpreted in accordance with the
laws of the Commonwealth of Massachusetts (excluding the conflict of laws and
rules thereof). All disputes under this Agreement will be resolved in the
applicable state or federal courts of Massachusetts. YOU consent to the
jurisdiction of such courts and waive any jurisdictional or venue defenses
otherwise available.
8. Integration and Severability
This Agreement represents the entire agreement between YOU and Harvard with
respect to the subject matter hereof, and supersedes all prior or
contemporaneous representations or understandings (whether oral, written or
electronic) between YOU and Harvard with respect to the subject matter hereof.
If any provision of this Agreement is or becomes invalid or is ruled invalid
by any court of competent jurisdiction or is deemed unenforceable, it is the
intention of the parties that the remainder of this Agreement shall not be
affected.
9. Reporting Requirement
Should YOU (i) inadvertently receive identifiable information or otherwise
identify a subject, or (ii) become aware of any use or disclosure of the Data
not provided for or permitted by this Agreement, YOU shall immediately notify
Harvard via email to
[[email protected]](mailto:[email protected]) with
**"ReXGradient-160K" in the subject line**, and follow Harvard's reasonable
written instructions, which may include return or destruction of Data.
10. Ownership
Harvard shall retain ownership of any rights it may have in the Data, and YOU
do not obtain any rights in the Data other than as set forth in this
Agreement.
11. Use of Harvard Name
Except as expressly provided in this Agreement, YOU shall not use or register
Harvard's name (alone or as part of another name) or any logos, seals,
insignia or other words, names, symbols or devices that identify Harvard,
including any Harvard school, unit, division or affiliate ("Harvard Names")
for any purpose in connection with this Agreement except with the prior
written approval of, and in accordance with restrictions required by, Harvard.
The foregoing notwithstanding, YOU may respond to legitimate business
inquiries with factual information regarding the existence and purpose of the
relationship that is the subject of this Agreement, without written permission
from Harvard. Without limiting the foregoing, YOU shall cease all use of
Harvard Names permitted under this Agreement on the termination or expiration
of this Agreement except as otherwise approved by Harvard.
Except as expressly provided in Section II. Attribution of this Agreement, YOU
shall not use Gradient Health's name, trademarks, or other logos in any
publicity, advertising, or news release without the prior written approval of
an authorized representative of Gradient Health.
12. Term and Termination
YOU may use the Data for the duration of YOUR research project as allowed
under the permitted uses in Article III of this Agreement. Either party may
terminate this agreement for any reason at any time, however, YOUR obligations
that by their nature are intended to persist following termination shall
remain in effect even after termination. Upon termination, YOU shall
immediately cease use of the Data and destroy Data in YOUR possession and, if
requested by Harvard, certify to Harvard as to its destruction.
Harvard may limit, suspend or terminate YOUR access to Data at any time if
Harvard believes YOU have violated the terms of this Agreement or otherwise
acted negligently with respect to the Data or that continued use of or access
to the Data by YOU presents significant risk to the individuals who are the
subjects in the Data.
13. Miscellaneous
YOU may not assign, transfer or delegate any of YOUR rights and obligations
hereunder without consent of the other party. No agency, partnership, joint
venture, or employment relationship is created as a result of the Agreement.
Neither Party shall have authority to make any statements, representations or
commitments of any kind on behalf of the other Party, or to take any action
which shall be binding on the other Party.
No modification or waiver of any provision of this Agreement shall be valid
unless in writing and executed by duly-authorized representatives of both
parties. A failure by one of the parties to this Agreement to assert its
rights hereunder shall not be deemed a waiver of such rights. No such failure
or waiver in writing by any one of the parties hereto with respect to any
rights shall extend to or affect any subsequent breach or impair any right
consequent thereon.
ReXGradient-160K Non-Commercial Data Access and Use Agreement v.1, April 7,
2025
extra_gated_fields:
Name: text
Institution: text
Email: text
By_checking_this_box:
type: checkbox
description: >-
By checking this box, YOU acknowledge that YOU have read and are
effectively signing this Non-Commercial Data Access and Use Agreement
[https://huggingface.co/datasets/rajpurkarlab/ReXGradient-160K] and
agreeing to its terms and conditions
configs:
- config_name: metadata
data_files:
- split: train
path: metadata/train_metadata.csv
- split: validation
path: metadata/valid_metadata.csv
- split: test
path: metadata/test_metadata.csv
Overview
ReXGradient-160K is the largest publicly available multi-site chest X-ray dataset, containing 273,004 unique chest X-ray images from 160,000 radiological studies, collected from 109,487 unique patients across 3 U.S. health systems (79 medical sites). This comprehensive dataset includes multiple images per study and detailed radiology reports, making it particularly valuable for the development and evaluation of AI systems for medical imaging and automated report generation models.
Dataset Composition
The dataset is divided into three splits:
- Training: 140,000 studies, 238,968 images, 95,716 patients
- Validation: 10,000 studies, 17,007 images, 6,964 patients
- Public Test: 10,000 studies, 17,029 images, 6,807 patients
An additional private test set (10,000 studies) is reserved for model evaluation on the ReXrank benchmark.
Image Characteristics
All images in the dataset were originally in DICOM format and have been converted to 16-bit PNG format using a standardized preprocessing pipeline, and downsampled to 1/4 of original dimensions in accordance with data use agreement
Report Structure
Each radiological report is structured into four key sections:
- Indication: Provides relevant patient background and reason for examination
- Comparison: References to previous studies for comparison
- Findings: Detailed radiological observations
- Impression: Summary of key conclusions and recommendations
Citation
If you use this dataset in your research, please cite:
@inproceedings{zhang2025rexgradient,
title={ReXGradient-160K: A Large-Scale Publicly Available Dataset of Chest Radiographs with Free-text Reports},
author={Zhang, Xiaoman and Acosta, Julián N. and Miller, Josh and Huang, Ouwen and Rajpurkar, Pranav},
booktitle={arXiv:2505.00228v1},
year={2025}
}
@article{zhang2024rexrank,
title={ReXrank: A Public Leaderboard for AI-Powered Radiology Report Generation},
author={Zhang, Xiaoman and Zhou, Hong-Yu and Yang, Xiaoli and Banerjee, Oishi and Acosta, Juli{\'a}n N and Miller, Josh and Huang, Ouwen and Rajpurkar, Pranav},
journal={AAAI Bridge Program AIMedHealth},
year={2025}
}
License and Terms of Use
This dataset is subject to the terms and conditions outlined in the Non-Commercial Data Access and Use Agreement at the top of this page.
Acknowledgments
This dataset was provided to Harvard by Gradient Health. Users are required to recognize (i) the ReXGradient-160K Data Repository and (ii) the contribution of Gradient Health as the source of the Data in all publications or presentations.
This work was supported by the Biswas Family Foundation's Transformative Computational Biology Grant in Collaboration with the Milken Institute. We extend our sincerest thanks for making this work possible.
Contact
For questions regarding the dataset, please contact: [email protected] with "ReXGradient-160K" in the subject line.