基于免疫算法和水平集的乳腺肿块分割方法  

Method of Breast Lumps Segmentation Based on Immune Algorithm and Level Set

作　　者：杨铁军[1] 吴效明[2] 黄琳[1] 

机构地区：[1]桂林工学院电计系,广西桂林541004 [2]华南理工大学生物力学研究所,广东广州510641

出　　处：《计算机仿真》2010年第3期243-246,共4页Computer Simulation

摘　　要：给出了一种乳腺肿块的自动分割方法。首先,使用肿块局部高亮、灰度均匀等区域信息的免疫算法检测器快速检测出包含肿块的感兴趣区域(ROI),显著减少了ROI的非目标信息;检测器可自动设置C-V水平集的初始位置,对ROI进行精确分割,不仅减少了计算量和提高了肿块分割的自动化程度,还增强了"目标和背景是同质的"命题的真实性,提高了水平集分割算法的性能。实验结果表明,方法能自动、快速、准确的分割肿块(包括毛刺征等),使测得结果与诊断结果相符。This paper proposes an automated breast lumps segmentation method. First, according to the features of lumps, such as local high intensity and local uniformity, an immune algorithm based detector is used to find the region of interest （ROI） quickly, and the non - object area of ROI is reduced remarkably. Second, C - V level set model is employed to segment the ROI more precisely by setting the initial level set automatically. The method not only improves the performance and the extent of segment automation, but also enhances the truth that the object and the background are homogeneous. Experiments show that the method can segment lumps automatically, rapidly and correctly, even if a lump is spiculate. And the segment results are in agreement with the radiologists＇ opinions.

关 键 词：水平集 免疫算法 图像分割 肿块 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]