README.md

metadata

license: mit
task_categories:
  - video-classification
language:
  - en
tags:
  - angiography
  - cardiology
  - X-ray
  - multi-view
  - video
  - coronary
  - dominance
  - medical
  - imaging
  - stenosis
  - occlusion
  - artery
  - uncertainty
  - outliers
pretty_name: coronary_dominamnce
size_categories:
  - 10B<n<100B

The dataset containes invasive coronary angiograms for the coronary dominance classification task, an essential aspect in assessing the severity of coronary artery disease. The dataset holds 1,574 studies, including X-ray multi-view videos from two different interventional angiography systems. Each study has the following tags: bad quality, artifact, high uncertainty, and occlusion. Those tags help to classify dominance classification more accurately and allow to utilize the dataset for uncertainty estimation and outlier detection.

More information about coronary dominance classification using neural networks in https://doi.org/10.48550/arXiv.2309.06958.

Some angiographic studies from the dataset are from CardioSYNTAX dataset of coronary agiograms for the SYNTAX score prediction in https://doi.org/10.48550/arXiv.2407.19894

CITATION Please cite:

@misc{ponomarchuk2024endtoendsyntaxscoreprediction,
      title={End-to-end SYNTAX score prediction: benchmark and methods}, 
      author={Alexander Ponomarchuk and Ivan Kruzhilov and Galina Zubkova and Artem Shadrin and Ruslan Utegenov and Ivan Bessonov and Pavel Blinov},
      year={2024},
      eprint={2407.19894},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2407.19894}, 
}