论文部分内容阅读
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 propose a lymphocyte synergetic classification method based on competitive optimal prototype. Methods Firstly, the ISODATA algorithm was used to cluster the homologous polymorphic lymphocyte prototype set. Then, according to the cell samples to be sorted, a collaborative competition mechanism was established to obtain the optimal prototype of each prototype set. Finally, Similar prototypes for the best match, with the largest similarity as the classification results. Results Four groups of cell types, lymphocytic follicular cells, lymphoid follicle central cells, lymphoid follicular centroblasts and centroblastic lymphoma cells, were achieved. Conclusion The experimental results show that the method is effective in the classification and lay the foundation for the next quantitative analysis of lymphocytes.