Identification of blood cells subpopulations from angle resolved light scattering

Flow cytometry has experienced a considerable expansion of available parameters during the last years. Various techniques that allow for the direct analysis of leukocytes by direct labeling with antigen-specific fluorescent labeled antibodies or according to functional parameters will have enormous impact on immunological research and hematopathological diagnosis. Scanning flow cytometry (SFC) [1] with its additional ability to measure morphological and/or biophysical differences between different blood cell subpopulations can make analysis of leukocytes more accurate and more comfortable. The angle resolved light scattering pattern (LSP) measured with SFC contains encoded information about morphological and biophysical properties of cells. Cells having subtle differences in their morphological and/or biophysical properties can potentially be better discriminated by LSP compared to “classical” integrated side scatter measurements. Eventually this could lead to a reduction in the number of antibodies needed for proper subset identification, thus allowing for either less expensive testing or adding additional antigenic evaluations to the test. The aim of this study is to develop multiparametric methods of classification for peripheral blood (PB) cells based on the LSP only and using them for cell subset identification. We adapted Bayes classification algorithm to classify different types of blood cells. The antigen-specific (antibody) labeling was used for building of a learning sample but not for the classification itself. The integrals of LSP over different angular ranges and spectrum parameters of LSP were used in the classification. The optimization of algorithmwas done by correctly definiting parameters.We verified the obtained algorithm on different healthy donors. Though some overlap in cell subsets is present, good evaluations with subset classification errors less then 15% can be made without the need for antigen-specific labeling.