基于CT图像的周围型肺部疑似肿瘤分割  

Segmentation of Peripheral Pulmonary Suspicious Nodules in CT Images

作　　者：周林[1] 张建州[1] 朱建峰[1] 蒋平[1] 

机构地区：[1]四川大学计算机学院,四川成都610065

出　　处：《计算机应用研究》2007年第4期154-155,共2页Application Research of Computers

基　　金：国家自然科学基金资助项目(60371024)

摘　　要：以CT图像为基础,针对周围型肺部肿瘤分割问题,首先将经过预处理的CT图像利用阈值技术二值化,并采用形态滤波方法去掉散点和小桥,然后将每个连通区域标记出来,根据判断准则确定可疑的肿瘤区域。该方法使对阈值选取的敏感性和所需的人机交互减小到最低限度。Segmentation of peripheral pulmonary nodules based on CT images was proposed. This method first made the preprocessed image a binary one, and employed morphology techniques to get rid of solitary points and small bridges, then labeled the white connectivity components in that binary image respectively and drew the contour of components which was suspected as peripheral pulmonary nodules according to the rules. This method can make the both the sensitivity to the selection of threshold value and necessitated interaction reduced to the least extent.

关 键 词：高分辨率计算机断层扫描 分割 阈值 区域生长 连通区域标记 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术] R318[自动化与计算机技术—计算机科学与技术]