Segmentation of the Liver 2007 (SLIVER07)

Note: This dataset is subject to a customized license. Original CT images are not redistributed here; please refer to the official download link (below).

Supplementary segmentation annotations from our project are provided here. Users are expected to respect the intended use and restrictions of the original dataset license.

The mapping between our segmentation files and the original images can be found below.

License

Customized license:

The following rules apply to those who register a team, download data and submit results:

The original data sets and associated segmentation data downloaded here, or any data derived from these data sets, will not be given nor distributed under any circumstances to persons not belonging to the registered team. The data on this site is to be used exclusively for the purpose of developing segmentation algorithms of the liver. Use of the data for any other purposes requires explicit permission from the maintainers of this site, listed at the bottom of this page. Commercial use of segmentation algorithms that use the supplied data (test or training cases) as training material is forbidden. In scientific publications (journal publications, conference papers, technical reports, presentations at conferences and meetings) that use the data from this website, you must cite the following paper:

Citation

Paper BibTeX:

@article{heimann2009comparison,
  title={Comparison and evaluation of methods for liver segmentation from CT datasets},
  author={Heimann, Tobias and Van Ginneken, Bram and Styner, Martin A and Arzhaeva, Yulia and Aurich, Volker and Bauer, Christian and Beck, Andreas and Becker, Christoph and Beichel, Reinhard and Bekes, Gy{\"o}rgy and others},
  journal={IEEE transactions on medical imaging},
  volume={28},
  number={8},
  pages={1251--1265},
  year={2009},
  publisher={IEEE}
}

Dataset description

SLIVER07 is a benchmark challenge to evaluate liver segmentation methods on contrast-enhanced CT scans during the central venous phase. It originated from the MICCAI 2007 Grand Challenge, where 16 teams (10 automatic, 6 interactive) evaluated their methods on a shared dataset. Algorithms included statistical shape models, atlas registration, level-sets, graph-cuts, and rule-based systems.

Challenge homepage: https://sliver07.grand-challenge.org/

Number of CT volumes: 30

Contrast: Contrast-enhanced (central venous phase)

CT body coverage: Chest

Does the dataset include any ground truth annotations?: Yes

Original GT annotation targets: Liver and internal structures (e.g., vessels)

Number of annotated CT volumes: 20

Annotator: Human

Acquisition centers: Multiple clinical partners

Pathology/Disease: Various liver pathologies, including tumors, metastases, and cysts

Original dataset download link: https://zenodo.org/records/2597908

Original dataset format: .mhd / .raw

ID mapping: Our segmentation files use the naming convention liver-orig**_train/test.
The original Zenodo images use the naming convention liver-orig**.raw.
Each segmentation file is linked to the image with the same ID in the train or test folder.