Improved Centerline Extraction in Fully Automated Coronary Ostium Localization and Centerline Extraction Framework...

Area of Science:

• Medical Imaging
• Cardiovascular Imaging
• Artificial Intelligence in Medicine

Background:

• Accurate coronary artery extraction is crucial for quantitative analysis of stenoses and atherosclerotic plaque in cardiac CT angiography (CCTA).
• Current methods may lack automation or require extensive manual intervention, limiting clinical applicability.

Purpose of the Study:

• To develop a fully automated workflow for coronary artery centerline extraction from CCTA image volumes.
• To improve the accuracy and efficiency of vessel analysis for clinical applications.

Main Methods:

• Utilized a regression U-Net to identify coronary artery ostium points within CCTA volumes.
• Employed an orientation and radius predictor CNN to track and extract vessel centerlines from ostium to endpoint.
• Developed a workflow dependent on convolutional networks for automated vessel extraction.

Main Results:

• Achieved an average of 96% accuracy in identifying ostium points within 5mm of their true location.
• Demonstrated high accuracy in coronary artery centerline extraction.
• The method requires fewer training parameters, enhancing suitability for clinical use.

Conclusions:

• The proposed automated workflow effectively extracts coronary artery centerlines from CCTA.
• This AI-driven approach offers high accuracy and efficiency, suitable for clinical integration and continuous learning in cardiovascular imaging.