The VerSe2020 dataset includes 300 multidetector computed tomography (MDCT) image series of the spine. The data is published under the licence CC BY-SA 4.0 (see licence.txt and readme.txt).
The dataset is an extension to the Verse19 dataset, that can be found at https://osf.io/nqjyw/
**By downloading the data you agree to cite these papers in your consecutive 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 https://doi.org/10.1148/ryai.2020190138.
2. Liebl H, Schinz D, Sekuboyina A, ..., Kirschke JS. A computed tomography vertebral segmentation dataset with anatomical variations and multi-vendor scanner data SDATA-21-002892021.
doi: 10.1038/s41597-021-01060-0 (preliminary access at https://arxiv.org/abs/2103.06360)
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/j.media.2021.102166. Epub 2021 Jul 22. (preliminary access at https://arxiv.org/abs/2001.09193)
Please also pay attention to the license file here: https://osf.io/amh4f/ (Creative Commons Attribution-ShareAlike 4.0)
In VerSe 2020, there are also 30 scans contained from Glocker´s spine dataset, described here:
https://biomedia.doc.ic.ac.uk/data/spine/
These scans are indicated by "gl000" as name tag. We only use the CT data and created own annotations for this data. However, we ask you to also cite his work as indicated on his website.
Please respect our work, as we spent - next to more than two years of work for algorithmic development - more than 1000 hours for manually correcting the segmentations, in addition to the 2000h for verse19. Please cite us appropriately!
The 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)
The scans are stored as NIFTI files and include vertebra-level segmentations (NIFTI format) and labels (JSON format). The data is available in 2 different formats:
* Subject-based adaption of the Brain Imaging Data Structure (BIDS; https://bids.neuroimaging.io/). This format is in a similar way available for VerSe 2019: https://osf.io/nqjyw/ Please refer to this repository for all IDs < 499.
* MICCAI Challenge data structure. This format was used at MICCAI 2020 for the VerSe Challenge. https://verse2020.grand-challenge.org/. Please note, that (in addition to the naming) the format of the centroid files is different.
The data is published under the licence CC BY-SA 4.0 (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 TUM (Proposal 27/19 S-SR).
**Labeling rules**
We only label “free” vertebrae, i.e we do not label the sacrum or transitional vertebrae that are (partly) fused with the sacrum. Such fused vertebrae are referred to as Castellvi grade 3 and 4. In this regard, all "free" vertebrae (including an ankylosis due to degeneration), are called lumbar. We consider L1 to be the first vertebra without ribs or with rib remnants smaller than 4cm on both sides in a horizontal alignment (including heterotopic ossification of the transverse process). The last thoracic vertebra should have at least one rib longer than 4cm in a typical diagonal downward alignment. In ambiguous cases, the shape of vertebra and facet joints are considered. If T1 is not present in the scan (i.e. visible within the scan's field-of-view), the thoracic spine is considered to have 12 vertebrae.
[1]: https://osf.io/nqjyw/
