Automated carotid artery intima layer regional segmentation

Area of Science:

• Medical imaging
• Cardiovascular diagnostics
• Ultrasound technology

Background:

• Carotid artery wall evaluation is crucial for cardiovascular risk stratification and diagnosing pathologies.
• Accurate segmentation of the intima layer is essential for precise measurements.

Purpose of the Study:

• To introduce a novel, user-independent algorithm, carotid artery intima layer regional segmentation (CAILRS), for automated carotid artery intima layer segmentation.
• To extract lumen-intima and media-adventitia borders from the far wall of the carotid artery using ultrasound images.

Main Methods:

• CAILRS employs mean shift classification applied to the far wall of the carotid ultrasound artery.
• The algorithm's performance was validated by comparing its border extraction to expert manual tracings.
• Benchmarking included comparison with a semi-automatic first-order absolute moment edge operator (FOAM) and prior automated methods (CALEX, CULEX, CALSFOAM, CAUDLES-EF).

Main Results:

• CAILRS demonstrated a mean intima-media thickness (IMT) bias of -0.035 ± 0.186 mm compared to FOAM's -0.016 ± 0.258 mm.
• The algorithm showed uniform behavior across a database of 300 carotid ultrasound images.
• CAILRS achieved a figure of merit of 95.6%, outperforming previous automated techniques.

Conclusions:

• CAILRS offers a robust and user-independent solution for automated carotid artery intima layer segmentation.
• The algorithm provides accurate border extraction, contributing to improved cardiovascular risk assessment.
• CAILRS represents a significant advancement over existing automated methods in carotid ultrasound analysis.