Medical Segmentation Decathlon
Ten segmentation tasks: brain tumor, hippocampus, lung tumor, liver tumor, pancreas tumor, colon cancer, hepatic vessels, spleen, prostate, and left atrium
Open the source page ↗ See it in the directory
At a glance
- What it is
- Ten segmentation tasks: brain tumor, hippocampus, lung tumor, liver tumor, pancreas tumor, colon cancer, hepatic vessels, spleen, prostate, and left atrium
- Area
- Neuro, Chest, Body, Cardiac
- Modality
- CT, MR
- Format
- NIfTI
- Size
- Ten tasks, from 30 to 750 volumes each; test labels are not released
- Normals included
- No
- License
- CC BY-SA 4.0
- Commercial use
- Allowed by the license, with its attribution terms
- Access
- Open download, no account
- Status
- Verified: source page reviewed by Hounsfield Lab
- Added
- 2026-10-19
- Last checked
- 2026-11-11
- Source
- registry.opendata.aws/msd/
How to access
Use the AWS Open Data page. The bucket s3://msd-for-monai can be listed and downloaded with the AWS CLI and --no-sign-request, so no AWS account is needed. MONAI also ships a loader for these tasks. Task descriptions are on medicaldecathlon.com.
Limitations
Built for a 2018 segmentation challenge, so each task targets one structure or lesion and the cohorts are small for some tasks (left atrium has 30 volumes). Test-set labels are withheld. The brain tumor task is drawn from BraTS 2016–2017 and the lung task from TCIA, so check overlap before combining with those sources. Share-alike applies if you redistribute adapted data.
Citation required?
Yes. Cite Simpson et al., A large annotated medical image dataset for the development and evaluation of segmentation algorithms (arXiv:1902.09063), and Antonelli et al., Nature Communications 2022.
Related open-source tools
- HD-BET: Deep-learning brain extraction (skull stripping) for MRI
- FastSurfer: Deep-learning whole-brain segmentation and cortical surface reconstruction, a fast alternative to FreeSurfer recon-all
- FreeSurfer: Neuroimaging analysis and visualization suite
- BrainChop: In-browser 3D MRI rendering and segmentation
- Neuro-VL: Neuroimaging vision-language model (Neuro-VL-8B-Thinking, Qwen3-8B language model with an HLIP visual encoder) that reads complete MRI and CT exams, converses about findings, drafts reports with reasoning traces and compares serial exams; trained on nearly 1 million clinical neuroimaging studies from 450,000 patients; released with the NeuroBench benchmark
- dcm2niix: Converts DICOM to NIfTI with JSON sidecars; the standard converter in many brain MRI and BIDS pipelines
- Pillar-0 (finetune code + HF weights): Open 3D radiology foundation model; finetuning framework, weights on Hugging Face
More neuro datasets
- BrainMetShare: MR · Stanford AIMI research use agreement (non-commercial; commercial license sold separately)
- Burdenko-GBM-Progression: MR, CT · NIH Controlled Data Access Policy (images); CC BY 4.0 (clinical and genomic CSV)
- CFB-GBM: MR, CT · CC BY 4.0
- CPAISD: CT · CC BY 4.0
- CQ500: CT · CC BY-NC-SA 4.0
- CT-ICH: CT · PhysioNet Restricted Health Data License 1.5.0
Hounsfield Lab links to the source and does not host images. The license and access summary reflects the source page on the last-checked date; terms can change, so read the source before you download. Not legal advice.