Automatic 3D coronary artery segmentation based on local region active contour model

Area of Science:

• Medical Imaging
• Cardiovascular Imaging
• Image Analysis

Background:

• Coronary artery segmentation in computed tomography angiography (CTA) is crucial for diagnosing and treating heart diseases.
• Manual analysis of CTA data is time-consuming and requires specialized radiologist expertise.
• There is a need for automated methods to efficiently segment coronary arteries from CTA datasets.

Purpose of the Study:

• To develop and validate an automated method for segmenting coronary arteries in 3D CTA images.
• To overcome the limitations of manual segmentation, including time consumption and reliance on expert knowledge.

Main Methods:

• Anisotropic diffusion filtering for noise reduction while preserving vessel boundaries.
• A two-step coronary skeleton extraction process (thick and thin vessels) using clustering, vesselness filtering, region growing, and height ridge traversal.
• Incorporation of vesselness measure into a local region active contour model guided by vessel geometry.
• Generation of initial contours from the coronary artery skeleton for segmentation.

Main Results:

• The automated method achieved an average precision of 86.64%, recall of 91.26%, and Dice Similarity Coefficient (DSC) of 79.13% on chest CTA images.
• Demonstrated effective segmentation of coronary arteries, including challenging thin vessel extraction.
• Validated the method's performance in accurately segmenting complex coronary artery structures.

Conclusions:

• The developed automated method successfully segments coronary arteries from CTA data without manual seed selection or initial contour definition.
• The approach facilitates efficient extraction of coronary artery skeletons and subsequent precise segmentation.
• This automated technique offers a promising solution for improving the speed and accuracy of coronary artery analysis in clinical practice.