Robust simultaneous detection of coronary borders in complex images

Area of Science:

• Cardiology
• Medical Imaging
• Biomedical Engineering

Background:

• Visual estimation of coronary obstruction severity from angiograms is unreliable due to poor reproducibility and accuracy.
• Existing automated methods often fail to accurately identify vessel borders, limiting their clinical adoption.
• Complex angiographic features like poor contrast and overlapping structures pose challenges for automated analysis.

Purpose of the Study:

• To develop and validate a robust method for simultaneous detection of left and right coronary artery borders.
• To compare the reliability of the new simultaneous border detection method against conventional automated border detection.
• To assess the accuracy of stenosis diameter measurements obtained by the new method.

Main Methods:

• Development of a novel algorithm for simultaneous detection of left and right coronary artery borders.
• Testing the simultaneous border detection method on 130 complex coronary angiograms where conventional methods were expected to fail.
• Comparison of failure rates and stenosis diameter measurements between simultaneous and conventional automated border detection methods.

Main Results:

• Conventional automated border detection failed in 50% of complex images.
• The new simultaneous border detection method demonstrated significantly higher robustness, failing in only 12% of images (p<.001).
• Simultaneous border detection yielded stenosis diameters that correlated significantly better with observer-derived measurements (p<0.001).

Conclusions:

• Simultaneous detection of left and right coronary borders is a highly robust automated method.
• This technique shows substantial promise for improving the accuracy and clinical utility of quantitative coronary angiography.
• The method effectively handles complex angiographic features, addressing limitations of current automated approaches.