肿瘤癌变细胞FISH图像分析系统的研究  

Research on the FISH image analysis system of tumor cell

作　　者：陈舒[1] 陈若寒 

机构地区：[1]福建警察学院计算机与信息安全管理系,福建福州350007

出　　处：《微型机与应用》2017年第1期48-51,55,共5页Microcomputer & Its Applications

摘　　要：肿瘤癌变细胞FISH图像分析系统中,需要解决粘连细胞核的分割问题。FISH属于新兴技术,产生的是特殊的荧光彩色细胞图,现有细胞图像分析方法并不适用。文章创新设计了基于深度凹陷检测和构造自然凹陷力的方法,分离粘连细胞核。首先,针对参差不齐的实验成像,在RGB模型下结合统计思想,将图像分为三类,分别进行预处理。继而,利用融合了K-means聚类算法的改进马尔科夫随机场(MRF)方法,将细胞核与癌变信号点进行有效提取。在此基础上,利用几何原理,创新设计了一套粘连细胞核分离算法。最后,给出细胞核快速计数和信号点提取方法。该系统设计基本完整,并达到预期效果。In FISH image analysis system, the segmentation of the cell nucleus is needed. FISH is a new technology, it is a special fluoreseent color cell image, the existing cell image analysis method is not applicable. In this paper, we design a method based on deep depression detec- tion and constructing the natural sag force to separate the adhesion nucleus. Firstly, according to the different experimental imaging, the images are divided into three kinds according to the RGB model, and the images are preprocessed. Then, using the improved Markov Random Field （MRF） method with the integration of K-means clustering algorithm, to extract nuclei and carcinogenesis signal points. On this basis, using geometric principles, a set of adhesion cell nuclear separation algorithm is designed. In the end, the method of nuclear fast counting and signal point extraction is presented. The design of the system is basically complete, and achieve the desired results.

关 键 词：FISH图像 细胞核提取 MRF模型 细胞核粘连 自然凹陷力 

分 类 号：R857.3[医药卫生—航空、航天与航海医学] TP391[医药卫生—临床医学]