---
license: cc-by-nc-sa-3.0
configs:
- config_name: axial
  data_files:
  - split: train
    path: axial/axial_train.csv
  - split: validation
    path: axial/axial_valid.csv
- config_name: coronal
  data_files:
  - split: train
    path: coronal/coronal_train.csv
  - split: validation
    path: coronal/coronal_valid.csv
- config_name: sagittal
  data_files:
  - split: train
    path: sagittal/sagittal_train.csv
  - split: validation
    path: sagittal/sagittal_valid.csv
task_categories:
- image-segmentation
- image-to-text
- visual-question-answering
language:
- en
tags:
- medical
pretty_name: OminiAbnorm-CT-14K
size_categories:
- 10K<n<100K
extra_gated_prompt: >
  ## Terms and Conditions for Using the OminiAbnorm-CT-14K Dataset


  **1. Acceptance of Terms**

  Accessing and using the OminiAbnorm-CT-14K dataset implies your agreement
  to these terms and conditions. If you disagree with any part, please refrain
  from using the dataset.


  **2. Permitted Use**

  - The dataset is intended solely for academic, research, and educational
  purposes.

  - Any commercial exploitation of the dataset without prior permission is
  strictly forbidden.

  - You must strictly comply with all requirements in the [Radiopaedia License](https://radiopaedia.org/licence). Especially, for machine learning projects, you must apply and pay the required fee through Radiopaedia's official application system.

  - You must adhere to all relevant laws, regulations, and research ethics,
  including data privacy and protection standards.


  **3. Data Protection and Privacy**

  - Acknowledge the presence of sensitive information within the dataset and
  commit to maintaining data confidentiality.

  - Direct attempts to re-identify individuals from the dataset are prohibited.

  - Ensure compliance with data protection laws such as GDPR and HIPAA.


  **4. Attribution**

  - Cite the dataset and acknowledge the providers in any publications resulting
  from its use.

  - Claims of ownership or exclusive rights over the dataset or derivatives are
  not permitted.


  **5. Redistribution**

  - Redistribution of the dataset or any portion thereof is not allowed.

  - Sharing derived data must respect the privacy and confidentiality terms set
  forth.


  **6. Disclaimer**

  The dataset is provided "as is" without warranty of any kind, either expressed
  or implied, including but not limited to the accuracy or completeness of the
  data.


  **7. Limitation of Liability**

  Under no circumstances will the dataset providers be liable for any claims or
  damages resulting from your use of the dataset.


  **8. Access Revocation**

  Violation of these terms may result in the termination of your access to the
  dataset.


  **9. Amendments**

  The terms and conditions may be updated at any time; continued use of the
  dataset signifies acceptance of the new terms.


  **10. Governing Law**

  These terms are governed by the laws of the location of the dataset providers,
  excluding conflict of law rules.


  **Consent:**

  Accessing and using the OminiAbnorm-CT-14K dataset signifies your
  acknowledgment and agreement to these terms and conditions.
extra_gated_fields:
  Name: text
  Institution: text
  Email: text
  I have read and agree with Terms and Conditions for using the dataset: checkbox
---

OminiAbnorm-CT is the first large-scale dataset designed for abnormality grounding and description on multi-plane whole-body CT imaging. 
It comprises 14.5K CT images from Radiopedia, covering axial, coronal, and sagittal planes and diverse anatomical regions. 
All images and the paired reports have been rigorously reviewed by experienced radiologists. 
We invite 4 radiologists (with at least 7 years’ experience) to manually annotate around 19K abnormal findings on the images in the format of either bounding boxes or
segmentation masks. 
All regional annotations are linked to the corresponding report descriptions and further categorized according to a hierarchical [taxonomy](https://huggingface.co/datasets/zzh99/OminiAbnorm-CT-14K/blob/main/Taxonomy(EN)1018.json).
The taxonomy was developed in collaboration with 7 radiologists from 3 centers (each with 10-16 years of experience), and covers 404 representative abnormal findings organized across 40 major anatomical regions and 82 sub-regions.

<div align="center">
   <img src="./Data.png">
</div>

For more details, please refer to our [paper](https://www.arxiv.org/abs/2506.03238) and [github repo](https://github.com/zhaoziheng/OminiAbnorm-CT).

Below are some data samples:

<div align="center">
   <img src="./Demo.png">
</div>

Note that：
- Each grounding annotation is a connected region.
For cases where the same abnormality may appear in multiple locations within an image 
(for example, a patient may have multiple infected areas in the lungs), 
we use `abnormality_group` to associate these spatially disconnected annotations together.
- Each abnormality may have up to 5 coexisting categories.

Images in this dataset are contributed by radiologists worldwide. 
We provide `link` for each of them for proper attribution to the original authors and institutions.
We deeply appreciate all the radiologists for sharing these valuable resources to advance medical research.

⚠️ Please strictly adhere to the [Radiopedia License](https://radiopaedia.org/licence?lang=us) when using this dataset, including but not limited to:
- Attribute the content correctly.
- Do not use it for any commercial or profitable purpose without explicit permission.
- [Apply](https://radiopaedia.org/licence) and pay the required fee ($250 USD) when using it for machine learning.

### Citation
```
@misc{zhao2025rethinkingwholebodyctimage,
      title={Rethinking Whole-Body CT Image Interpretation: An Abnormality-Centric Approach}, 
      author={Ziheng Zhao and Lisong Dai and Ya Zhang and Yanfeng Wang and Weidi Xie},
      year={2025},
      eprint={2506.03238},
      archivePrefix={arXiv},
      primaryClass={eess.IV},
      url={https://arxiv.org/abs/2506.03238}, 
}
```