基于改进变分水平集法提取超声图像乳腺肿瘤边缘  

Extraction of Breast Tumor Boundary in Ultrasonic Image Based on Improved Variational Level Set Algorithm

作　　者：杨晓霜[1] 汪源源[1] 王涌[2] 王怡[2] 

机构地区：[1]复旦大学电子工程系,上海200433 [2]复旦大学附属华山医院,上海200040

出　　处：《航天医学与医学工程》2008年第6期500-504,共5页Space Medicine & Medical Engineering

基　　金：国家基础研究项目(2006CB705707);国家自然科学基金资助项目(30570488);上海市重点学科建设项目(B112)

摘　　要：目的研究准确提取超声图像乳腺肿瘤边缘的算法。方法在变分水平集方法中提出一种新型能量函数—全局—局部能量函数,通过加强图像的局部信息,实现超声图像乳腺肿瘤边缘的准确提取。结果对仿真超声图像的实验表明,本方法比全局最优化的Chan-Vese方法能更准确区分目标和背景。对103例超声乳腺肿瘤图像的实验表明,本方法无需任何预处理步骤即可准确提取肿瘤边缘。结论本方法有望实现超声图像乳腺肿瘤边缘的快速、准确提取。Objective To propose an extraction algorithm for extraction of breast tumor boundary in its ultrasonic image. Methods A novel Global-Local Energy function was proposed in the variational Level Set formulation for the curve evolution. By enhancing the intensity of local information, the breast tumor boundary was obtained from ultrasonic images accurately. Results Experiments on a synthetic ultrasound image showed that this method was more accurate at segmenting the object from the background than the Chan-Vese algorithm based on the global optimization. Experiments on 103 ultrasonic breast tumor images also showed that the boundaries were extracted accurately with this method without any pre-processing step. Conclusion This proposed method is expected for the fast and accurate boundary extraction of ultrasonic breast tumor boundary from ultrasonic image.

关 键 词：超声图像 乳腺肿瘤 边缘提取 变分水平集 Chan-Vese方法 全局—局部能量 

分 类 号：R445.1[医药卫生—影像医学与核医学] TP391[医药卫生—诊断学]