🧠 BraTS 2021

English简体中文

RSNA–ASNR–MICCAI brain tumor benchmark · pre-operative MRI

Official dataset source Related challenge

📚 Publications and records

arXiv 2107.02314 TCIA DOI 10.7937/jc8x-9874 IEEE TMI 2015

Quick facts
Input T1, T1ce, T2, FLAIR
Task Glioma subregion segmentation; the benchmark also included radiogenomic classification
Cases 2,040 total: 1,251 train, 219 validation, 570 test
Institutional origin Lead organizer: CBICA, University of Pennsylvania; challenge partners: RSNA, ASNR, and MICCAI. Data came from multiple institutions, including CBICA/UPenn, UCSF, Washington University in St. Louis, Heidelberg University, and Tata Memorial Centre. Official organizer and contributor information.
Official sources BraTS 2021 challenge · Current Synapse access notice

🗂️ Official split and label availability

Cohort Cases Reference labels
Training 1,251 Publicly available with the training release
Validation 219 Challenge labels were withheld for leaderboard evaluation
Test 570 Hidden challenge test; labels were not part of the ordinary public challenge download

The current Synapse notice says the old BraTS 2018–2021 datasets are no longer hosted there. It directs users to the registered-user BraTS-GLI release, which reuses the BraTS 2020/2021 data after acceptance of post-challenge terms.

⚠️ Important label-version note

The current BraTS-GLI release uses the same underlying data as BraTS 2020/2021 with one documented change: the enhancing-tumor label is 3 instead of the original 4. Keep the release name and label mapping with any downloaded masks or results.

Data and annotations

Each case uses four structural MRI sequences. The standard BraTS targets are whole tumor, tumor core, and enhancing tumor; the training references are expert tumor subregion annotations. The benchmark’s label convention should be taken from the specific release rather than inferred from a filename.

🔒 Access and terms

A Synapse account and acceptance of the post-challenge terms are required for the current GLI route. Follow the current Synapse terms before downloading or redistributing data.

RSNA–ASNR–MICCAI BraTS 2021 evaluated glioma subregion segmentation on multi-institutional pre-operative MRI and MGMT promoter methylation classification.

📚 Publications and citation

Benchmark paper: Baid et al. “The RSNA-ASNR-MICCAI BraTS 2021 Benchmark on Brain Tumor Segmentation and Radiogenomic Classification.” arXiv preprint, 2021. arXiv:2107.02314.

Related foundational reference: Menze et al. “The Multimodal Brain Tumor Image Segmentation Benchmark (BRATS).” IEEE Transactions on Medical Imaging, vol. 34, no. 10, pp. 1993–2024, 2015. DOI: 10.1109/TMI.2014.2377694.

The arXiv entry is the linked benchmark citation; no journal or proceedings venue is claimed for that preprint here.


🕘 Sources last checked: 27 October 2026 · Back to the dataset catalog