Automatic Segmentation of Common Carotid Artery Using Affine Snake in Gradient Vector Field Technique

Manual examination of the common carotid artery in ultrasound images requires time and effort. We propose affine snakes in gradient vector field algorithm that generates automatic segmentation of common carotid artery in transverse mode ultrasound images and assist in the automated diagnosis of atherosclerosis. The segmentation is achieved following the transformation of affine snake from initialization to convergence on the region of interest under the presence of gradient vector field. To determine affine transformation parameters, a force field on the image is computed and then the snake is deformed until the net force on it drops down to zero. Deformation of the snake is done by the gradient descent algorithm which helps it converge into the common carotid artery region. Automated initialization of the active contour is done using normalized cross-correlation technique. We have validated the performance of the proposed algorithm on a publicly available ultrasound image database provided by SPLab, Brno University, which consists of 971 transverse mode ultrasound images of the common carotid artery. We have achieved carotid artery segmentation accuracy and Dice similarity index of 99.53% and 87.30%, respectively.