Full-automated system for the segmentation of the common carotid artery in ultrasound images

Abstract

The full segmentation of the common carotid artery (CCA) in ultrasound images is important for the evaluation of the intima media thickness (IMT) and for the measurement of the artery stenosis which are considered to be the significant markers for the clinical evaluation of the risk of stroke. The current research proposes full-automated segmentation system for the segmentation of the CCA, which is based on an adaptive snake-contour segmentation algorithm. The CCA is segmented by the proposed algorithm into different distinct regions, namely the IMT, intima-media (IL), media-layer (ML), carotid plaque and lumen. The proposed method is automatically processing image normalization, binarization, adaptive hybrid median filter, and morphology prior the application of the snake segmentation algorithm. The mean and standard deviation of the IMT diameter in y-axis of the full-automatically segmented regions for the snakes-based and level-set method are 0.12 mm +/− 0.01 mm and 0.09 mm +/− 0.01 mm respectively in comparison with the ground truth IMT extracted from the manual clinical segmentation. The Wilcoxon rank sum test shows the significant improvements of the proposed method.