metadata
license: cc-by-sa-4.0
🏆 Medical Segmentation Decathlon Dataset
📝 Overview
The Medical Segmentation Decathlon (MSD) is a comprehensive benchmark dataset for validating algorithms in 3D medical image segmentation. It includes 10 distinct tasks, each with unique challenges like small data sizes, unbalanced labels, varying object scales, multi-class labels, and multimodal imaging.
🔗 Dataset Access
- 🌐 Website: Medical Decathlon
- 📂 Google Drive: MSD Google Drive
🧩 Task Descriptions and Download Links
🧠 Task 01: Brain Tumours
- Target: Gliomas segmentation (necrotic/active tumour and oedema)
- Modality: Multimodal MRI (FLAIR, T1w, T1gd, T2w)
- Size: 750 4D volumes (484 Training + 266 Testing)
- Source: BRATS 2016 & 2017 datasets
- Challenge: Complex, heterogeneously-located targets
- 📥 Download: Task01_BrainTumour.tar
❤️ Task 02: Heart
- Target: Left Atrium
- Modality: Mono-modal MRI
- Size: 30 3D volumes (20 Training + 10 Testing)
- Source: King’s College London
- Challenge: Small dataset with high variability
- 📥 Download: Task02_Heart.tar
🫁 Task 03: Liver
- Target: Liver and tumour
- Modality: Portal venous phase CT
- Size: 201 3D volumes (131 Training + 70 Testing)
- Source: IRCAD Hôpitaux Universitaires
- Challenge: Unbalanced labels with large (liver) and small (tumour) targets
- 📥 Download: Task03_Liver.tar
🧬 Task 04: Hippocampus
- Target: Hippocampus head and body
- Modality: Mono-modal MRI
- Size: 394 3D volumes (263 Training + 131 Testing)
- Source: Vanderbilt University Medical Center
- Challenge: High-precision segmentation of small neighboring structures
- 📥 Download: Task04_Hippocampus.tar
🧑⚕️ Task 05: Prostate
- Target: Prostate central gland and peripheral zone
- Modality: Multimodal MRI (T2, ADC)
- Size: 48 4D volumes (32 Training + 16 Testing)
- Source: Radboud University Medical Centre
- Challenge: Segmenting two adjacent regions with large variations
- 📥 Download: Task05_Prostate.tar
🌬️ Task 06: Lung
- Target: Lung and tumours
- Modality: CT
- Size: 96 3D volumes (64 Training + 32 Testing)
- Source: The Cancer Imaging Archive
- Challenge: Small target (cancer) in a large image
- 📥 Download: Task06_Lung.tar
🍃 Task 07: Pancreas
- Target: Pancreas and tumour
- Modality: Portal venous phase CT
- Size: 420 3D volumes (282 Training + 139 Testing)
- Source: Memorial Sloan Kettering Cancer Center
- Challenge: Unbalanced labels with large, medium, and small structures
- 📥 Download: Task07_Pancreas.tar
🩸 Task 08: Hepatic Vessels
- Target: Hepatic vessels and tumour
- Modality: CT
- Size: 443 3D volumes (303 Training + 140 Testing)
- Source: Memorial Sloan Kettering Cancer Center
- Challenge: Small, tubular structures near a heterogeneous tumour
- 📥 Download: Task08_HepaticVessel.tar
🌿 Task 09: Spleen
- Target: Spleen
- Modality: CT
- Size: 61 3D volumes (41 Training + 20 Testing)
- Source: Memorial Sloan Kettering Cancer Center
- Challenge: Large foreground size variation
- 📥 Download: Task09_Spleen.tar
🩹 Task 10: Colon
- Target: Colon Cancer Primaries
- Modality: CT
- Size: 190 3D volumes (126 Training + 64 Testing)
- Source: Memorial Sloan Kettering Cancer Center
- Challenge: Heterogeneous appearance
- 📥 Download: Task10_Colon.tar
📜 License
The data is available under a permissive CC-BY-SA 4.0 license, allowing sharing, distribution, and further development.
🖋️ Citation
Please cite the following paper when using this dataset:
📏 Assessment Metrics
Performance is evaluated using:
- 🥇 Dice Score (DSC)
- 📐 Normalized Surface Distance (NSD)
📬 Contact
For questions, please reach out to the organizers at: [email protected]