Main content

Files | Discussion Wiki | Discussion | Discussion
default Loading...



Loading wiki pages...

Wiki Version:
The original dataset *VerSe 2019* ([][1]) - includes 160 CT image series of 141 patients with segmentation masks of 1725 vertebrae. - is split into training (80), validation (40), and test set (40). - was prepared for a vertebral labeling and segmentation challenge hosted at the 2019 International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI). Please read the readme.txt and licence.txt: (Creative Commons Attribution-ShareAlike 2.0). Data Structure ------------------- This dataset is available in two different ways: 1. **Image series based (MICCAI):** 160 image series of 141 patients are divided into a training (n=80), validation (n=40), and test (n=40) set as originally published for the MICCAI challenge. 2. **Subject based:** 141 patients holding 160 image series are divided into a training (n=67), validation (n=37), and test (n=37) set. Subject identifiers equal image series identifiers ('verseXXX'). For patient with 2 or 3 image series, new subject identifiers ('verse4XX') were introduced. This is an adaption of the Brain Imaging Data Structure (BIDS; We recommend to use the subject-based format, as this is completely consistent in VerSe 2019 and VerSe 2020. Please note, that the VerSe 2019 cases that were re-used in the VerSe 2020 challenge have to be downloaded from the subject-based VerSe 2019 repository and are not again included in the subject-based VerSe 2020 repository. Additionally, we offer python scripts to easily work with the subject-based data-structure here: Citation --------- Please respect our work, as we spent >2 years for algorithmic development and >2000 working hours for manual corrections of segmentation masks. **By downloading this data you agreed to cite these papers in your work:** 1. Löffler M, Sekuboyina A, Jakob A, Grau AL, Scharr A, Husseini ME, Herbell M, Zimmer C, Baum T, Kirschke JS. A Vertebral Segmentation Dataset with Fracture Grading. Radiology: Artificial Intelligence, 2020 2. Liebl H, Schinz D, Sekuboyina A, ..., Kirschke JS. A computed tomography vertebral segmentation dataset with anatomical variations and multi-vendor scanner data Sci Data. 2021 Oct 28;8(1):284. doi: 10.1038/s41597-021-01060-0. 3. Sekuboyina A, Bayat AH, Husseini ME, Löffler M, Menze BM, ..., Kirschke JS. VerSe: A Vertebrae labelling and segmentation benchmark for multi-detector CT images. Med Image Anal. 2021 Oct;73:102166. doi: 10.1016/ Epub 2021 Jul 22. preliminary access at An overwiew of the data is provided in reference 1. The methods to generate the initial segmentations that were manually corrected afterwards are detailed in reference 2 and 3. This work has been supported by the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (grant agreement No 637164 — iBack — ERC-2014-STG). Data Format ----------- Scans and segementation masks are stored in NIFTI format ( Coordinates of vertebral body centroids per vertebral level are stored in JSON format. Please refer to the Component Wiki Pages for further details. Licence and Ethics ------------------ The data is published under the licence CC BY-SA 2.0 (see licence.txt). When using the data you must cite the three papers mentioned above. Ethical approval to publish this data has been obtained from the local ethics committee at the Technical University of Munich (Proposal 27/19 S-SR). Subsequent Project ------------------ The follow-up of this project is available here: [1]:
OSF does not support the use of Internet Explorer. For optimal performance, please switch to another browser.
This website relies on cookies to help provide a better user experience. By clicking Accept or continuing to use the site, you agree. For more information, see our Privacy Policy and information on cookie use.

Start managing your projects on the OSF today.

Free and easy to use, the Open Science Framework supports the entire research lifecycle: planning, execution, reporting, archiving, and discovery.