Using two methods for recognition common carotid artery of B-mode longitudinal ultrasound image

In this paper two novel methods are proposed for recognizing B-mode ultrasound (US) imaging of common carotid artery (CCA) in longitude. In the single frame method, the US is segmented by k-means fuzzy clustering algorithm. Then a series points are extracted based on geometric features. Thirdly, the feature points are selected according to our new cluster method. Then curve fitting is applied by feature points. Finally CCA is recognized in US imaging. According to the experiment on 7402 frames of 104 US videos from different persons, 95.66% accuracy rate is achieved. Based on multi-frame algorithm, it integrates the first algorithm for present frame and a buffer strategy for previous frame information. 98.97% accuracy rate is higher than the single frame method. Our novel technique outperforms the current technique proposed in other papers. The recognition CCA is totally automatic in our database by our proposed methods. Those proposed automatic methods are suitable for real-time recognition of the CCA.