中文摘要：

为了有效地提取动脉超声造影图像中的血管边界，提高动脉内膜附近血流流场超声测量的准确性，提出一种结合时域信息的区域增长算法进行动脉超声造影图像的血管边界分割．该算法根据超声造影微泡在血液内的流动特性，利用连续两帧图像之间的灰度差异构造窗帧差，令其作为生长条件，结合邻域灰度进行区域生长；采用数学形态学的闭合运算与一种自定义的边界平滑方法对区域生长结果进行处理，填补了区域生长中产生的空洞并平滑了不规则边界．最后通过大鼠颈动脉的超声造影图像实验，验证了文中算法的有效性和可行性，该算法能够准确地提取出模糊的动脉边界，时间复杂度低．

英文摘要：

An image segmentation algorithm based on region growing combined with time domain information is proposed for delineating the arterial boundaries in ultrasound contrast images and improving the measurement accuracy of the velocimetry. Firstly, window frame difference is constructed by the intensity difference of two consecutive frames according to the flow characteristics of ultrasound contrast microhubbles in blood. Region growing is proceeded by the growing condition combined with window frame difference and neighborhood intensity. Then the close operator of morphology is used to fill the holes generated in region growing and a customized method is developed to smooth the irregular boundaries. Finally, the experimental results verify the efficiency and feasibility of the proposed algorithm in segmenting ultrasound contrast images of rat carotid arteries. The proposed algorithm is able to extract the obscure arterial boundaries and has low time complexity.