肺癌细胞识别彩色图像处理系统  

A CYTOLOGICAL COLOR IMAGE PROCESSING SY STEM FOR RECOGNITION OF LUNG CANCER CELLS

作　　者：刘雷健[1] 杨静宇[1] 陆建峰[1] 

机构地区：[1]南京理工大学计算机科学系

出　　处：《计算机应用与软件》1996年第4期34-44,共11页Computer Applications and Software

基　　金：国家"八五"重点科技攻关项目

摘　　要：本文介绍了一个用于识别痰液涂片彩色细胞图像中肺癌细胞的彩色图像处理系统。为了提高系统识别癌细胞的稳定性与有效性,我们采用了一种细胞分割与分类的分层处理结构。首先,在某一特定的归一化彩色空间中,利用自适应阈值的方法对细胞核进行分割;然后,利用细胞核的形态学特征检出可疑癌细胞;最后用细胞核区的色度学特征对可疑癌细胞进行分类与识别,从而确定是否存在癌细胞。实验结果表明,本文提出的方法是可行、有效的。This presentation describes a color image processing system for the detection of lung cancer cells in the cytological color images of the examinees' sputum slides. To increase the stability and efficiency of the detection of the cancer cells, we adopt a hierarchical processing architecture for the segmentation and recognition of cancer cells. First, the nuclei are segmented by thresholding in a normalized chromatic space. Then, the candidate cancer cells are selected by using their nuclear morphological features. Finally, by utilizing the chromatic features of the candidate nuclei, all the cancer cells are recognized. Experiment results are given to show the feasibility of the approach proposed hare.

关 键 词：肺癌 癌细胞 识别 图像处理 彩色细胞图像 

分 类 号：R734.2[医药卫生—肿瘤]