The probabilities mixture model for clustering flow-cytometric data: an application to gating lymphocytes in peripheral blood

Data clustering is a major data mining technique and has been shown to be useful in a wide variety of domains, including medical and biological statistical data analysis. A non trivial application of cluster analysis occurs in the identification of different subpopulations of particles in large-sized heterogeneous flow-cytometric data. Mixture-model based clustering has been several times applied in the past to medical and biological data analysis; to our knowledge, however, non of these applications was involved with flow-cytometric data. We claim, that utilizing the probabilities mixture model offers several advantages compared to other proposed flow-cytometric data clustering approaches. We apply this model in order to gate lymphocytes in peripheral blood, which is a necessary first-step procedure when dealing with various hematological diseases diagnoses, such as lymphocytic leukemias and lymphoma.