VerSe – Vertebrae Labelling and Segmentation Benchmark

License

CC BY-SA 4.0
Creative Commons Attribution-ShareAlike 4.0 International License

Citation

Paper BibTeX:

@article{sekuboyina2021verse,
  title={VerSe: a vertebrae labelling and segmentation benchmark for multi-detector CT images},
  author={Sekuboyina, Anjany and Husseini, Malek E and Bayat, Amirhossein and L{\"o}ffler, Maximilian and Liebl, Hans and Li, Hongwei and Tetteh, Giles and Kuka{\v{c}}ka, Jan and Payer, Christian and {\v{S}}tern, Darko and others},
  journal={Medical image analysis},
  volume={73},
  pages={102166},
  year={2021},
  publisher={Elsevier}
}

Dataset description

The VerSe benchmark, introduced at MICCAI 2019 and 2020, provides multi-detector CT scans for vertebrae labelling and segmentation. It includes 374 scans with over 4,500 vertebrae annotated using a human–machine hybrid approach, enabling the development and evaluation of algorithms across diverse anatomy and acquisition protocols.

Challenge homepage: https://verse2020.grand-challenge.org/

Number of CT volumes: 374

CT Type: Multi-detector CT (MDCT)

CT body coverage: Spine (various fields of view)

Does the dataset include any ground truth annotations?: Yes

Original GT annotation targets: Vertebrae C1–L5, transitional T13 and L6

Number of annotated CT volumes: 374

Annotator: Automated algorithm + manual refinement

Acquisition centers: -

Pathology/Disease: Vertebral fractures, metallic implants, and foreign materials

Original dataset download link:

https://github.com/anjany/verse