A fast seed detection using local geometrical feature for automatic tracking of coronary arteries in CTA

Area of Science:

• Medical Imaging
• Cardiovascular Imaging
• Image Analysis

Background:

• Accurate segmentation and tracking of coronary arteries are crucial for diagnosing cardiovascular diseases.
• Current methods for coronary artery extraction from coronary computed tomographic angiography (CCTA) can be time-consuming and require manual intervention.
• Developing automated methods is essential for efficient clinical workflow and improved diagnostic accuracy.

Purpose of the Study:

• To propose a fast and efficient seed detection method for automatic tracking and extraction of coronary arteries in CCTA.
• To enhance the accuracy and computational efficiency of coronary artery analysis in CCTA datasets.
• To validate the proposed method using clinical CCTA data and receiver operating characteristic (ROC) curve analysis.

Main Methods:

• A novel local geometric feature, based on the similarity of consecutive cross-sections perpendicular to the vessel direction, is introduced.
• Hessian-based filtering is combined with this geometric feature for robust vessel region detection.
• Regions of Interest (ROIs) are extracted from axial slices containing main coronary arteries to improve computational efficiency.
• Seed points, representing vessel centroids, and their directions are identified for subsequent vessel tracking.
• A particle filtering-based tracking algorithm is employed for the final coronary artery extraction.

Main Results:

• The proposed method successfully detects seed points for coronary artery tracking in CCTA.
• Full automatic coronary artery extraction is achieved using the identified seed points and vessel directions.
• Validation on 19 clinical CCTA datasets demonstrates the method's effectiveness.
• ROC curve analysis confirms the advantages and superior performance of the proposed seed detection technique.

Conclusions:

• The developed fast seed detection method enables fully automatic coronary artery extraction from CCTA.
• The combination of Hessian-based filtering and a novel local geometric feature significantly improves detection accuracy and efficiency.
• This automated approach holds promise for streamlining cardiovascular imaging analysis and diagnosis.