Coronary Artery Dataset (CCA)

This dataset, named the CCA dataset, consists of 200 cases of CTA images depicting coronary artery disease. Out of these cases, 20 were allocated for training purposes, while the remaining 180 cases were reserved for testing. We have made our training dataset publicly available for further use.

The collected images are acquired with an isotropic resolution of 0.5 mm. Four radiologists participated in annotating the coronary artery internal diameter of 200 cases as ground truth, that each case was independently labelled by three radiologists and the remaining radiologist selected the best one among three annotations. The cases were shuffled randomly and organized into a queue. Once a radiologist labelled a case, they were assigned the subsequent case based on the random order of the queue. This process guaranteed that each case received labels from three distinct radiologists. Importantly, each radiologist remained unaware of the labels assigned by the other two radiologists, ensuring the masking of the labels during the assessment. After three radiologists have finished labeling, the remaining radiologist examines and selects the best one of three labels, completing the combining of the radiologists’ decisions.

If you use this dataset, please cite:

@misc{yang2023segmentation,
      title={Segmentation and Vascular Vectorization for Coronary Artery by Geometry-based Cascaded Neural Network}, 
      author={Xiaoyu Yang and Lijian Xu and Simon Yu and Qing Xia and Hongsheng Li and Shaoting Zhang},
      year={2023},
      eprint={2305.04208},
      archivePrefix={arXiv}
}