KalbodyarAll packs

All packs › CT atlas › Sources and licenses

Sources and licences of the CT atlas

Which data it is built from and under what conditions it is used

Review

No label has been reviewed by a radiologist. The base labels come from the datasets and the rest were created semi-automatically (density threshold, region growing, geometric division); the educational texts, questions and interpretation of the clinical cases have not been reviewed by a specialist either. This atlas is educational and not for diagnostic use. If you see an error, let us know with the “⚑ Report error” button on each page.

Data

TotalSegmentator dataset v2.0.1 — CC BY 4.0

All regions except the hand and foot, the whole trunk and the clinical cases. Wasserthal J, Breit H-C, Meyer MT, et al. “TotalSegmentator: Robust Segmentation of 104 Anatomic Structures in CT Images”. Radiology: Artificial Intelligence 2023;5(5):e230024. Dataset v2.0.1, doi:10.5281/zenodo.10047292. License: CC BY 4.0. The base labels come from this dataset; this project created the tibia, fibula, patella, radius, ulna, cerebrospinal fluid spaces and muscle groups.

DICOM DECT of patients with gout (hands) — CC BY 4.0

Atlas of the hand (left) and foot (right). Udodov V, Zamyshevskaya M, Zavadskaya V, Kim TV. “DICOM DECT of patients with gout (hands)”. Zenodo, 2024. doi:10.5281/zenodo.14283402 (cases “11 h” and “36 h”). License: CC BY 4.0. In this project the bones were separated with manual points and geodesic/watershed competition; the patients have gout, and pathological changes may be visible.

TotalSegmentator open tasks — Apache-2.0

Supplementary labels and the locations of the findings in the clinical cases come from the open TotalSegmentator models: lung_vessels, liver_segments, liver_vessels, vertebrae_body, ventricle_parts, head_muscles, head_glands_cavities, oculomotor_muscles, craniofacial_structures, headneck_bones_vessels, headneck_muscles, pleural_pericard_effusion, lung_nodules. License: Apache License 2.0. This atlas contains no labels from the TotalSegmentator models with non-commercial licenses.

VerSe 2019 — CC BY-SA 4.0

Clinical case: lumbar spine was created from VerSe 2019 (osf.io/nqjyw, sub-verse078): the CT image and vertebra masks come from this dataset (the masks were corrected by specialists of that project); the choice of vertebra, the grading and the educational text are from this project.

License: CC BY-SA 4.0. Images and labels created from this data (the CT volume and the labels of that page) are also released under CC BY-SA, not under Kalbodyar's general terms; anyone may use them with attribution and under this same license. That page also displays this license. The rest of the atlas does not use this data. The VerSe papers that must be cited:

  1. Löffler M. et al. Radiology: AI 2020, doi:10.1148/ryai.2020190138 — “A Vertebral Segmentation Dataset with Fracture Grading”. Radiology: Artificial Intelligence 2020;2(4):e190138.
  2. Liebl H. et al. Sci Data 2021, doi:10.1038/s41597-021-01060-0 — “A computed tomography vertebral segmentation dataset with anatomical variations and multi-vendor scanner data”. Scientific Data 2021;8:284.
  3. Sekuboyina A. et al. Med Image Anal 2021, doi:10.1016/j.media.2021.102166 — “VerSe: A Vertebrae labelling and segmentation benchmark for multi-detector CT images”. Medical Image Analysis 2021;73:102166.

What this project added

Resampling and cropping of the image, semi-automatic labels, landmarks and joints, Persian and English names, educational texts, questions, clinical cases and the body map. Images are used without any patient information. These parts are also published under the same licence condition (attribution) as the data of each region.