超声图像处理中Snake模型研究  

Study on Snake Model in the Ultrasound Image Processing

作　　者：赵暖[1] 陈亚青[2] 余建国[1] 王威琪[1] 

机构地区：[1]复旦大学电子工程系,上海200433 [2]上海市第六人民医院,上海200233

出　　处：《上海生物医学工程》2004年第4期3-9,共7页Shanghai Journal of Biomedical Engineering

摘　　要：Snake模型是一种基于高层信息的有效目标轮廓提取算法,其优点是作用过程及最后结果的目标轮廓是一条完整的曲线,从而引起广泛的关注。鉴于医学超声图像的信噪比较低,用经典的边缘提取算法无法得到较好的结果,因此人们将Snake模型进行了各种各样的改进,并且越来越多地将它运用到医学超声图像处理中来。本文对乳腺超声图像进行阈值分割、形态滤波等一系列预处理后,将改进的Snake模型对乳腺超声图像进行肿瘤的边缘提取,得到了比较好的结果。Snake model is an effective edge extraction algorithm based on advanced information. Its advantage is that during and after the calculation the edge of the object is a continuous curve, so it attracts more and more attention. Whereas the signal to noise ratio of medical ultrasound images is quite low, it can not get optimal result using classical edge extraction algorithm. So Snake model after kinds of improvement is applied in medical ultrasound images increasingly. In this paper, improved Snake model is applied to the ultrasound images of breast lesion after a lot of preprocessing like threshold segmentation and shape filtering. And this new method can extract the edge of the lesion and get preferable result.

关 键 词：超声图像处理 SNAKE模型 乳腺超声图像 阈值分割 

分 类 号：R318[医药卫生—生物医学工程]