A Semi-Automatic Segmentation Method for the Structural Analysis of Carotid Atherosclerotic Plaques by Computed Tomography Angiography

Aim: Computed tomography angiography (CTA) is currently the most reliable imaging technique for evaluating and planning the treatment of atherosclerosis. The drawbacks of the technique are its low spatial resolution and challenging manual measurements. The purpose of this study was to develop a semi-automatic method to segment vessel walls, surrounding tissue, and the carotid artery lumen to measure the severity of stenosis. Methods: In vivo contrast CTA images from eight patients undergoing endarterectomy were analyzed using a tailored five-step process involving an adaptive segmentation algorithm and region growing to measure the maximum percent stenosis in the cross-sectional area of the carotid artery. The accuracy of this method was compared with that of manual measurements made by physicians. Results: There were no significant differences between the maximum percent stenosis value obtained using the semi-automatic tool and that obtained using manual measurements (6%; p = 0.31). The data acquisition and analysis required an average of 145 seconds. Conclusion: This new semi-automatic segmentation method for CTA provides a fast and reliable tool to quantify the severity of carotid artery stenosis.