Automatic Classification and Labeling of Cell Subsets in FCM Data
Objective To study the value of automatic analysis method for flow cytometry data analysis and solving the problem of automatic classification and labeling of cell subsets,so as to provide reference for disease diagnosis.Methods The data of bone marrow flow cytometry from 528 cases of acute leukemia in 2021 were collected,and the original flow cytometry data were preprocessed by compensation,conversion and deadhesion.The preprocessed data were analyzed by unsupervised clustering method,and the supervised classification model was trained by using the central location of the generated cell subsets,namely the distribution rule of macro cells,to further classify the subsets.Finally,the cell subsets were labeled as known cell types by manual recognition and labeling.Results Unsupervised clustering method and supervised classification method could be used in flow cytometry data analysis,which can realize automatic classification and labeling of cell subsets,and the accuracy can reach or almost reach the level of manual analysis.Conclusion The method of automatic classification and labeling of flow cytometry data proposed in this study bridge the gap between cell clustering and patient classification existing in current flow cytometry automation,and provid a solution for the whole process automation.The intermediate results required for clinical diagnosis can be used for quality control of disease diagnosis.
万方数据
维普
