vsd-lower-extremities-seg

This dataset contains 30 CT scans of the lower body. Each scan comes with:

• a label map of 63 individual bones, from the sacrum to the toes
• a surface mesh (STL) of each bone
• subject metadata

The subjects are 14 women and 16 men, aged 19–95.

Preview of lower-extremity segmentations in this dataset.

Source

This dataset combines two releases by M. C. M. Fischer:

What we used	Source
CT images, subject metadata	VSDFullBodyBoneReconstruction (Zenodo)
Bone meshes (MATLAB files)	VSDFullBodyBoneModels v3.0 (Zenodo)

The first release also includes segmentations and meshes, but they are hollow shells, so we did not use them. The filled label maps were created from the closed meshes of the second release.

How the dataset was made

The scripts are in dataset_creation/, numbered in the order they were run:

1. 1_convert_mat_meshes_to_stl.m (MATLAB) saves every bone in the MATLAB mesh files as a separate STL file.
2. 2_split_fused_foot_bones_rhino.py (Rhino 3D) splits the fused tarsal, metatarsal and phalanx meshes into individual bones. This step is semi-manual: you click each bone to export it. 2_split_fused_foot_bones_rhino.mp4 shows the process.
3. 3_create_lower_extremity_ct.py (Python) builds the dataset:
   • merges subjects that were scanned in two parts (pelvis–thighs and shanks–feet) into one CT image
   • flips scans that were stored upside down
   • converts the STL meshes into a filled label map on the CT grid
   • copies the meshes and metadata

The scripts contain local paths to the source data, so you need to adjust them before running them. The Python requirements are in dataset_creation/requirements.txt.

Contents

vsd-lower-extremities-seg/
├── lower_extremity_ct/
│   ├── dataset.json          # label names and numbers
│   ├── imagesTr/             # 001_0000.nii.gz … 030_0000.nii.gz   CT in Hounsfield units
│   └── labelsTr/             # 001.nii.gz … 030.nii.gz             labels 0–63
├── meshes/
│   └── 001/ … 030/           # one STL per bone, in the same coordinates as the CT
├── metadata/
│   └── 001.json … 030.json   # subject and scan information
├── dataset_creation/         # scripts used to create the dataset (see above)
└── assets/preview.png

Labels

Bone	Left	Right
Sacrum (single bone in the midline)	1	–
Hip	2	3
Femur	4	5
Patella	6	7
Tibia, fibula and all foot bones	8–62 (even)	9–63 (odd)

• Labels 8–63 are the 56 labels of total-feet-seg, shifted by +7.
• The full list is in lower_extremity_ct/dataset.json.

Meshes

Each folder meshes/<case>/ contains:

• One STL per labeled bone. The file name is the label name, for example Talus_L.stl.
• The original fused foot meshes. These are Tarsals_*.stl, Metatarsals_*.stl and Phalanges_*.stl.

Metadata

Each file metadata/<case>.json is the original record of that subject from the VSD database, plus one extra field, source_subject. The most useful fields are:

• source_subject: the original subject ID, for example 002 or z001.
• subjectSnapshot: sex, age (in days), height and weight.
• rawImage: slice thickness and tube voltage (kVp), for some cases only.

The records contain links to smir.ch (for example selfUrl and downloadUrl). That website no longer hosts the VSD database, so do not follow these links.

Download

# everything (~8.7 GB)
hf download BoneHub/vsd-lower-extremities-seg --repo-type dataset --local-dir ./vsd-lower-extremities-seg

# only the meshes
hf download BoneHub/vsd-lower-extremities-seg --repo-type dataset --include "meshes/*" --local-dir ./vsd-lower-extremities-seg

Training with nnU-Net

lower_extremity_ct/ is already in nnU-Net format, but nnU-Net expects dataset folders to be named DatasetXXX_Name.

1. Rename the folder. Move lower_extremity_ct/ into $nnUNet_raw under a name with an unused ID:

   mv ./vsd-lower-extremities-seg/lower_extremity_ct "$nnUNet_raw/Dataset003_vsd-lower-extremities-seg"
   

2. Preprocess and train. Use the same ID with -d and in the train command:

   nnUNetv2_plan_and_preprocess -d 3 -c 3d_fullres --verify_dataset_integrity
   nnUNetv2_train 3 3d_fullres all -tr nnUNetTrainerNoMirroring
   

   Use nnUNetTrainerNoMirroring, so that left and right labels are not swapped during training.

Known issues

Two cases are missing some bones. They have no label and no mesh:

• Case 026: the distal phalanges of toes 2, 3 and 4 on the right foot (labels 45, 51, 57).
• Case 030: all phalanges (labels 36–63).

Related

• BoneHub/vsd-feet-seg: the same subjects, cropped to the feet. It uses the same case numbers.
• BoneHub/total-feet-seg: the foot segmentation model trained on that dataset.

License

This dataset is released under CC BY-NC-SA 4.0, the same license as the source bone models.

Citation

If you use this dataset, please cite both this dataset (doi:10.57967/hf/10471) and the source publication:

@misc{seyedhamidrezaalaviVsdlowerextremitiesseg,
  title = {vsd-lower-extremities-seg},
  author = {Seyed Hamidreza Alavi and Malte Asseln},
  year = {2026},
  publisher = {Hugging Face},
  doi = {10.57967/HF/10470},
  url = {https://huggingface.co/datasets/BoneHub/vsd-lower-extremities-seg},
  urldate = {2026-09-16}
}

@article{fischerDatabaseSegmentationsSurface2023,
  title = {Database of Segmentations and Surface Models of Bones of the Entire Lower Body Created from Cadaver {{CT}} Scans},
  author = {Fischer, M.C.M.},
  year = 2023,
  journal = {Scientific Data},
  volume = {10},
  number = {1},
  pages = {763},
  publisher = {Nature Publishing Group},
  issn = {2052-4463},
  doi = {10.1038/s41597-023-02669-z},
}