Automated Segmentation of Common Carotid Artery in Ultrasound Images

We propose a basis splines-based based active contour method for the segmentation of lumen boundary and media adventitia boundary from transverse and longitudinal B-mode ultrasound images. Basis-spline has
$M$
knots in the shape template and five free parameters i.e., a pair of center coordinates, scaling in the horizontal and vertical directions, and the rotation angle. The segmentation of the region of interest in ultrasound images is done by minimizing the local energy function using gradient descent technique. Further optimization is carried out using Green’s theorem. Automatic localization poses an equal importance as segmentation and is achieved using sum of absolute difference method. The result of experimental validation has been done on the Signal Processing Lab, Brno University’s database of transverse and longitudinal B-mode ultrasound images consisting of 971 and 84 images, respectively. We attained accurate segmentation of the lumen boundary from transverse and longitudinal B-mode ultrasound images with an accuracy of 99.68% and 96.98% and Dice index of 93.33% and 91.70%, respectively. In addition, we attained an accuracy of 99.29% with Dice index of 91.78% for the segmentation of media adventitia boundary from transverse B-mode ultrasound images, which is the highest among the previously proposed methods.