基于改进分水岭算法和凹点搜索的乳腺癌粘连细胞分割  

Automatic Segmentation of Clustered Breast Cancer Cells Based on Modified Watershed Algorithm and Concavity Points Searching

作　　者：童振[1] 蒲立新[1,2] 董方杰[1] 

机构地区：[1]电子科技大学自动化学院,成都610054 [2]成都金盘电子科大多媒体技术有限公司,成都611731

出　　处：《生物医学工程学杂志》2013年第4期692-696,共5页Journal of Biomedical Engineering

基　　金：科技型中小企业技术创新基金资助项目(08C26215102272)

摘　　要：乳腺癌作为一种常见的恶性肿瘤,已经严重影响妇女身心健康甚至危及生命,在某些地区还呈现高发趋势。免疫组织化学技术作为一种常用的辅助病理诊断技术,在乳腺癌的诊断上发挥了重要作用,从免疫组化技术处理的标本中,识别阳性细胞,并统计阳性细胞百分率这一个重要的诊断指标。本文提出一种基于改进分水岭算法和凹点搜索的方法,识别阳性细胞并自动分割粘连细胞,最后实现自动计数,试验结果对比显示,本方法能够在不损失细胞几何特性的基础上准确地分离粘连细胞,实现自动计数并辅助完成统计。As a common malignant tumor, breast cancer has seriously affected women＇s physical and psychological health even threatened their lives. Breast cancer has even begun to show a gradual trend of high incidence in some I places in the world. As a kind of common pathological assist diagnosis technique, immunohistochemical technique plays an important role in the diagnosis of breast cancer. Usually, Pathologists isolate positive ceils from the stained specimen which were processed by immunohistochemical technique and calculate the ratio of positive cells which is a core indicator of breast cancer in diagnosis. In this paper, we present a new algorithm which was based on modified watershed algorithm and concavity points searching to identify the positive ceils and segment the clustered cells automatically, and then realize automatic counting. By comparison of the results of our experiments with those of other methods, our method can exactly segment the clustered ceils without losing any geometrical cell features and give the exact number of separating ceils.

关 键 词：免疫组化 粘连细胞 分水岭算法 凹点搜索 阳性细胞百分率 

分 类 号：R737.9[医药卫生—肿瘤]