CoronaryDominance: Angiogram dataset for coronary dominance classification

Area of Science:

• Cardiology
• Medical Imaging
• Machine Learning

Background:

• Coronary artery disease (CAD) severity assessment is crucial for treatment planning.
• SYNTAX score estimation relies on accurate coronary dominance classification.
• Existing datasets may lack diversity or specific classifications for robust model training.

Purpose of the Study:

• To introduce a comprehensive dataset for coronary dominance classification.
• To facilitate research in coronary artery disease severity assessment using invasive coronary angiograms.
• To enable the development of models for accurate SYNTAX score estimation.

Main Methods:

• A dataset of 1,574 invasive coronary angiogram studies was curated.
• Studies are X-ray multi-view videos classified as right or left dominant.
• The dataset includes three parts: main (with 5 categories including RCA occlusion), real distribution, and domain shift.

Main Results:

• The dataset provides detailed classifications for 1,574 angiographic studies.
• Includes categories for normal, poor quality, artifact, high uncertainty, and right coronary artery (RCA) occlusion.
• The dataset is structured into main, real distribution, and domain shift subsets for varied testing.

Conclusions:

• This dataset advances coronary dominance classification research.
• It supports improved accuracy in SYNTAX score estimation and CAD severity assessment.
• The diverse subsets facilitate robust model evaluation and generalization.