VerSe 2019: 160 spine CTs with vertebra-level label maps and centroids

pl.load("verse19_spine") loads this dataset. The Large Scale Vertebrae Segmentation Challenge 2019 (training, validation and test splits, 80 + 40 + 40 scans): spine CT in NIfTI with a 28-class vertebra label map and a JSON of vertebra centroids per scan. Three zips on the BoneScreen S3 bucket read scan by scan with HTTP Range. VerSe 2020 is the same layout under dataset-verse20*.zip.

import pardesline as pl

ds = pl.load("verse19_spine")
print(ds.describe())  # roles, frames, units, labels, licence, citation
s = ds[0]
obj = s.ct  # fetched now, cached under ~/.cache/pardesline/datasets/verse19_spine

Provenance and licence

Source	remote archive read by HTTP Range: https://s3.bonescreen.de/public/VerSe-complete/dataset-verse19training.zip, https://s3.bonescreen.de/public/VerSe-complete/dataset-verse19validation.zip, https://s3.bonescreen.de/public/VerSe-complete/dataset-verse19test.zip
Scheme	archive
Size of the whole source	12.0 GB
Licence	CC-BY-SA-4.0
Commercial use	allowed by the licence
Redistribution	allowed
Licence note	CC BY-SA 4.0 as stated in the VerSe repository (checked 2026-10-14).
Terms	https://github.com/anjany/verse
Homepage	https://github.com/anjany/verse
Cite	Sekuboyina A., Husseini M.E., Bayat A., et al. VerSe: A Vertebrae labelling and segmentation benchmark for multi-detector CT images. Med Image Anal 73:102166, 2021.
Samples grouped by	id

Roles (what each sample contains)

Role	Type	Frame	Units	Labels / layout	Pattern or file
ct	volume (CT)	from the file header	from the file header		dataset-verse19{split}/rawdata/{id}/{id}_ct.nii.gz Spine CT (NIfTI); field split = training / validation / test
mask	labels	from the file header	from the file header	1=C1, 2=C2, 3=C3, 4=C4, 5=C5, 6=C6, 7=C7, 8=T1 … (28 labels)	dataset-verse19{split}/derivatives/{id}/{id}_seg-vert_msk.nii.gz Vertebra label map (VerSe convention: 1-7 cervical, 8-19 thoracic, 20-25 lumbar, 26 sacrum, 27 cocygis, 28 T13)
centroids	file	from the file header	from the file header		dataset-verse19{split}/derivatives/{id}/{id}_seg-subreg_ctd.json Vertebra centroids and direction (JSON)

Standardized layout

Files are fetched on first access into the dataset cache and stay in their original format; ds.export(folder) writes every sample as standard files (PLY for points and meshes, MetaImage for volumes and label maps) with a .pardesline.json sidecar carrying frame, units, labels, licence and provenance, plus a manifest for the folder. Label maps become exact surfaces with to_mesh(label=k), DICOM series become one volume, raw arrays are decoded through the layout above.

ds = pl.load("verse19_spine").take(2)
ds.export("./standardized")        # PLY / MetaImage + sidecars + manifest.json

Algorithms and models that run on it

• labels roles: mask.to_mesh(label=k), mask.to_numpy(), metrics: dice, hd95 (evaluation-metrics-3d) ; catalog modules: pyvista_volume_processing, simpleitk_registration, simpleitk_segmentation
• volume roles: volume.align(moving), volume.segment(lower, upper), volume.to_mesh(isovalue=...), volume.to_sitk() ; catalog modules: pyvista_volume_processing, simpleitk_registration, simpleitk_segmentation
• Supervised segmentation: prepare_nnunet_dataset.py writes this dataset in the nnU-Net v2 raw layout; PointNet trains on the label surfaces.
• Self-supervised pretraining: 3D JEPA reads the volumes directly.

Related

• Datasets catalog · How to download a medical imaging dataset in Python · Add a dataset
• Tags: medical, ct, spine, vertebrae, segmentation, nifti

_{Generated from src/pardesline/data/manifests/verse19_spine.json on 2026-10-14.}