摘要
骨髓细胞分类识别过程中,对细胞形态属性识别的模糊性以及各属性与细胞类判定之间关系的模糊性,极大影响了细胞类的正确识别,致使目前对细胞图像分析处理仅能进行到六类细胞类粗略划分。结合模糊理论的隶属度定义及综合有权评判思想,给出了隶属函数及权重的确立方法,进而提出了基于专家知识的隶属度模糊识别机制,大大增加了骨髓细胞可识别类的细分程度。并在不影响识别效果的条件下,结合决策树判别,提高了识别的效率。
The accurate recognition of marrow cells is very difficult due to the vagueness of recognition to the morphological characters of cells and the vague relation between characters and certain cell. The current methods can only roughly deal with six categories of marrow cells through analysis to bone marrow microscope images. With the idea of membership and synthesis evaluation with weight in fuzzy theory, proposing a way to provide membership function and weight,a fuzzy recognition mechanism based on the expert knowledge is presented,which greatly increases the number of recognizable marrow cells. Besides, combining with decision tree improves the efficiency of recognition without influence to the effect of recognized result.
出处
《计算机应用与软件》
CSCD
北大核心
2006年第9期11-13,33,共4页
Computer Applications and Software
基金
国家自然科学基金资助项目(60303025)
关键词
骨髓细胞分类
白血病
模糊性
隶属度
决策树
Marrow cell classification Leukemia Fuzzy Membership Decision tree
