Segmentation of Common Carotid Artery Using Localized Region Based Active Contour

摘要

Carotid Intima-Media Thickness (IMT) is an important index in the evaluation of atherosclerotic disease. In this paper, the authors have attempted to segment the carotid IMT in longitudinal B mode ultrasound image of the common carotid artery (CCA) using Fast Fuzzy C-Mean Clustering (FFCM) and Localized region based Active Contour (LAC) scheme. The proposed method was evaluated on 66 samples. Authors have reported a comparison between the proposed method, manual segmentation, level set evolution without re-initialization (LER) and LAC method. It is observed that the proposed method has results better than LAC method and inferior to LER. Furthermore have investigated the effect of Median filter's window size and number of times it is applied on an image before segmentation. Inter-method (auto versus manual) variability has been studied by computing: average & standard deviation, coefficient of variation (CV %) and correlation.