Automatic segmentation, detection and quantification of coronary artery stenoses on CTA.

Area of Science:

• Cardiovascular Imaging
• Medical Image Analysis
• Computational Pathology

Background:

• Accurate detection and quantification of coronary artery stenoses are crucial for effective treatment planning in patients with suspected coronary artery disease.
• Computed tomography coronary angiography (CTCA) is a key imaging modality for assessing coronary artery disease.

Purpose of the Study:

• To develop and evaluate an automated method for detecting and quantifying coronary artery stenoses from CTCA data.
• To assess the performance of the proposed method against established quantitative coronary angiography (QCA) and manual CTCA assessments.

Main Methods:

• Centerline extraction using a two-point minimum cost path and refinement.
• Lumen segmentation employing graph cuts initialized by centerlines.
• Estimation of healthy lumen diameter using robust kernel regression.
• Stenosis detection and quantification via comparison of estimated and actual lumen diameter profiles.

Main Results:

• The method achieved a detection sensitivity of 29% and PPV of 24% versus QCA, and 21% sensitivity and 23% PPV versus manual CTCA assessment.
• Stenosis degree estimation showed an average absolute difference of 31% and RMS difference of 39.3% compared to QCA.
• Lumen segmentation achieved Dice scores of 68% for healthy and 65% for diseased segments.
• The method ranked second for stenosis quantification and lumen segmentation in the evaluation framework.

Conclusions:

• The developed automated method shows potential for coronary artery stenosis detection and quantification in CTCA.
• Further improvements are needed to enhance detection sensitivity and quantification accuracy for clinical application.
• The method demonstrates competitive performance in lumen segmentation and quantification compared to existing approaches.