基于竞争最优原型的淋巴细胞协同分类

Lymphocyte synergetic classification based on competition optimum prototype

目的提出一种基于竞争最优原型的淋巴细胞协同分类方法。方法首先,采用改进的ISODATA算法聚类同型多态的淋巴细胞原型集;然后,依据待分类细胞样本,建立协同竞争机制求出各原型集中的最优原型;最后,将待分类样本分别与各型最优原型进行相似匹配,以最大相似结果作为分类结果。结果实现了淋巴滤泡套细胞、淋巴滤泡中心细胞、淋巴滤泡中心母细胞、中心母细胞性淋巴瘤细胞等4组类型的细胞分类。结论实验结果表明该方法分类效果理想,为下一步淋巴细胞定量分析打下基础。

Objective To mum prototype. Method put forward a lymphocyte synergetic classification based on competition optiFirstly, according to ISODATA arithmetic, set up the lymphocyte prototype set. Secondly, in accordance with the cell swatch, pick out the optimum prototype based on synergetic competition model. Lastly, match the cell swatch with the optimum prototype and obtain the most similar result. Results The classification of mantle cells, follicular centrocytes cells, follicular centroblastic cells and centroblastic lymphoma cells is achieved. Conclusion The experimental results show that the method can be further extended to lymphocyte quantitative analysis.