TotalSegmentator v2
Command-line tool that segments 117 anatomical structures (v2) in any CT scan with nnU-Net models trained on more than 1,200 clinical CTs — the standard way to get organ masks for radiomics, dose planning and tumour-context features.
At a glance
What it does
pip install and one command: organs, bones, vessels and muscles labelled voxel-wise; separate task models add vertebrae, lung lobes, body composition and more.
Tasks, data types and cancers
Architecture
Training data
1,204 CT examinations (v1) from routine clinical practice at University Hospital Basel, covering a wide range of pathologies, scanners and protocols; v2 extends the classes and data.
Evaluation
| Benchmark / dataset | Metric | Value | External validation | Source |
|---|---|---|---|---|
| internal test set (v1 paper) | Dice | 0.943 mean over 104 structures | no | Source |
How to run
pip install TotalSegmentator
TotalSegmentator -i ct.nii.gz -o segmentations/ # 117 structures (v2)
TotalSegmentator -i ct.nii.gz -o seg/ --task body # extra task models: body, lung_vessels, ...Regulatory status and intended use
Regulatory status is quoted from the source linked above and can change. Research-use-only models must not be used for clinical decisions.
Limitations and bias
- Anatomy, not tumours: it does not segment lesions.
- Reduced accuracy on very low-dose or heavily artefacted scans.
Sources
This page is educational — it is not medical advice and does not replace consultation with an oncologist. Diagnostic and treatment decisions are made solely by specialist physicians.