Abstract
Quantification of coronary artery disease (CAD) from cardiac computed tomography angiography (CTA) is important both structurally (lumen area stenosis) and functionally (combined with computational fluid dynamics to determine fractional flow reserve) for assessment of ischemic stenosis and to guide treatment. Hence, it is important to have CTA image processing technique for segmentation and reconstruction of coronary arteries. In this study, we developed segmentation and reconstruction techniques, based on fast marching and Runge–Kutta methods for centerline extraction, and surface mesh generation. The accuracy of the reconstructed models was validated with direct intravascular ultrasound (IVUS) measurements in 1950 cross sections within 4 arteries. High correlation was found between CTA and IVUS measurements for lumen areas ( $$r=0.993$$ , $$p<0.001$$ ). Receiver-operating characteristic (ROC) curves showed excellent accuracies for detection of different cutoff values of cross-lumen area (5 $$\text {mm}^2$$ , 6 $$\text {mm}^2$$ , 7 $$\text {mm}^2$$ and 8 $$\text {mm}^2$$ , all ROC values >0.99). We conclude that our technique has sufficient accuracy for quantifying coronary lumen area. The accuracy and efficiency demonstrated that our approach can facilitate quantitative evaluation of coronary stenosis and potentially help in real-time assessment of CAD.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call