基于分层方法的白细胞五分类算法  

A Classification Algorithm for Five Types of White Blood Cells Based on Hierarchical Method

作　　者：赵建伟[1] 张敏淑 周正华[1] 楚建军 曹飞龙[1] 

机构地区：[1]中国计量学院理学院,浙江杭州310018 [2]嘉善加斯戴克医疗器械有限公司,浙江嘉兴314100

出　　处：《山西大学学报（自然科学版）》2015年第3期420-425,共6页Journal of Shanxi University(Natural Science Edition)

基　　金：浙江省自然科学基金(No.LY14A010027);国家自然科学基金(No.61272023;91330118)

摘　　要：针对白细胞自动识别问题,在已分割好的细胞基础上结合五类白细胞(中性粒细胞、嗜酸性粒细胞、嗜碱性粒细胞、单核细胞和淋巴细胞)的具体特征,利用分层的思想对其分类。首先提取白细胞中细胞核的分叶特征和圆形度特征,对该类特征明显的细胞进行筛选;而对该类特征不明显的细胞,提取对偶旋转不变共生局部二进制(PRICoLBP)纹理特征作为判定标准,将它们分为颗粒细胞与无颗粒细胞;然后对颗粒细胞,利用PRICoLBP纹理特征区分出嗜碱性粒细胞、嗜酸性粒细胞和中性粒细胞;而对无颗粒细胞,则用圆度核质比区分出淋巴细胞和单核细胞。实验表明,本文所提的方法比已有的分层方法在总体识别率上提高了十几个百分点,并且各类细胞的分类精度都有所提高。For the problem of automatic identification of white blood cells,this paper proposes a classification algorithm for its five types（neutrophils,eosinophils,basophils,monocytes and lymphocytes）using their specific features based on the idea of hierarchical method.Firstly,we extract the leaflet feature and the circularity feature of the nuclei of white blood cells to screen the cells with these obvious characteristics.While for the cells without those obvious characteristics,we use the pairwise rotation invariant cooccurrence local binary pattern（PRICoLBP）feature to divide them into granular cells and nongranular cells.Finally,we divide the granular cells into basophils,eosinophils and neutrophils using PRICoLBP feature,and divide the nongranular cells into lymphocytes and monocytes using the ratio of the texture features and circularity features.Some experiments illustrate that our proposed method gets better performance than the existing corresponding hierarchical method.

关 键 词：白细胞分类 分层方法 纹理特征 对偶旋转不变共生局部二值模式(PRICoLBP) 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]