BEAMSTER: Brain mEtAstases segMentation for STEreotactic Radiotherapy, A Retrospective MRI Dataset with Expert Segmentations.
Authors
Affiliations (6)
Affiliations (6)
- Department of Deputy Director for Science and Research, University Hospital & Faculty of Medicine, Ostrava, 708 52, Czech Republic. [email protected].
- Department of Biomedical Engineering, Faculty of Electrical Engineering and Communication, Brno University of Technology, Brno, 616 00, Czech Republic. [email protected].
- Department of Neurosurgery, University Hospital & Faculty of Medicine, Ostrava, 708 52, Czech Republic. [email protected].
- Department of Deputy Director for Science and Research, University Hospital & Faculty of Medicine, Ostrava, 708 52, Czech Republic.
- Department of Oncology, University Hospital & Faculty of Medicine, Ostrava, 708 52, Czech Republic.
- Department of Biomedical Engineering, Faculty of Electrical Engineering and Communication, Brno University of Technology, Brno, 616 00, Czech Republic.
Abstract
We present a retrospective dataset of contrast-enhanced T1-weighted magnetic resonance imaging scans from 140 patients with brain metastases who underwent stereotactic radiotherapy. In total, 260 metastatic lesions were annotated by an experienced radiation oncologist using a radiotherapy planning system, and the resulting binary segmentation masks were converted into NIfTI format. All data were de-identified prior to release, and facial features were removed from imaging data using a defacing procedure to ensure patient privacy. The dataset represents a valuable resource for the development and validation of computer-aided detection and segmentation methods, including those based on machine learning. A particular strength is the inclusion of small metastatic lesions, which are clinically relevant yet challenging to detect and segment automatically. We anticipate that this collection will facilitate reproducible research and support the development of robust algorithms for clinical translation in radiotherapy planning.