Automatic Detection of Common Carotid Artery in Transverse Mode Ultrasound Images

Automatic detection of common carotid artery is a precursor to the computer-aided analysis of atherosclerosis. In this paper, we present an approach for common carotid artery detection based on normalized matched filtering technique. The algorithm performance is optimized by introducing vector inner products and norms instead of conventional mean and variance computations. The difficulties of matched filtering are overcome by normalizing the image and template vectors to unit length, yielding a cosine similarity metric. We report common carotid artery detection results on Brno University of Technology, Signal Processing laboratory (SP lab) carotid artery database amounting to a total of 971 transverse mode ultrasound images out of which 538 and 433 images are taken from Ultrasonics and Toshiba ultrasound imaging devices, respectively. The proposed method results in a common carotid artery detection accuracy of 97.63% with a detection time per image of 2.21 seconds. We show that the normalized matched filtering is a reasonable choice for detection of common carotid artery in ultrasound images.