Overview

Schema Version

https://atlas.rsna.org/schemas/2025-11/dataset.json

Name

UCSF-BMSR: University of California San Francisco Brain Metastases Stereotactic Radiosurgery MRI Dataset

Link

https://imagingdatasets.ucsf.edu/dataset/1

Indexing

Keywords: Brain Metastases, Public Datasets, Artificial Intelligence, MRI, Stereotactic radiosurgery, Segmentation

Content: MR, NR, OI

RadLex: RID28639, RID10312, RID5231, RID16807, RID10782, RID49850

Author(s)

Jeffrey D. Rudie

Rachit Saluja

David A. Weiss

Pierre Nedelec

Evan Calabrese

John B. Colby

Benjamin Laguna

John Mongan

Steve Braunstein

Christopher P. Hess

Andreas M. Rauschecker

Leo P. Sugrue

Javier E. Villanueva-Meyer

Organization(s)

University of California San Francisco

Center for Intelligent Imaging, Department of Radiology and Biomedical Imaging, UCSF

Department of Radiation Oncology, UCSF

License

Text: Noncommercial license

URL: https://imagingdatasets.ucsf.edu/dataset/1

Funding

Authors declared no funding for this work.

Ethical review

Data collection approved by the UCSF Institutional Review Board with a waiver for consent.

Date

Published: 2024-02-21

References

[1] Rudie JD; Saluja R; Weiss DA; Nedelec P; Calabrese E; Colby JB; Laguna B; Mongan J; Braunstein S; Hess CP; Rauschecker AM; Sugrue LP; Villanueva-Meyer JE. "The University of California San Francisco Brain Metastases Stereotactic Radiosurgery (UCSF-BMSR) MRI Dataset". Radiology: Artificial Intelligence. 2024-02-21. doi:10.1148/ryai.230126. PMID: 38381038. PMCID: PMC10982817.

Dataset

Motivation

Provide a large, expert-annotated multimodal MRI dataset of brain metastases to advance AI detection and segmentation methods.

Sampling

Retrospective identification of stereotactic radiosurgery planning MRIs performed between 2017-01-01 and 2020-02-29 at UCSF.

Partitioning scheme

Training set (461 MRI examinations) publicly available; test set (99 individuals/images) to be released after completion of the 2024 MICCAI challenge.

Relationships between instances

Multiple MRI examinations per patient (560 exams from 412 patients); 99 cases have two sets of independent annotations combined for final reference standard.

External data

324 preoperative training images were included as an external dataset in the 2023 ASNR-MICCAI BraTS challenge (syn51156910).

Confidentiality

All images skull stripped to protect personal health information; no additional normalization or atlas registration performed prior to modeling.

Re-identification

Faces removed via skull stripping; risk reduced accordingly.

Sensitive data

Medical imaging and associated clinical information.