Dimensionality reduction of flow cytometric data through information preservation

Like many biomedical applications, flow cytometry is a field in which dimensionality reduction is important for analysis and diagnosis. Through expression patterns of various fluorescent biomarkers, flow cytometry is often used to characterize the malignant cells in cancer patients, traced to the level of the individual cell. Typically, diagnosticians analyze cytometric data through a series of 2-dimensional histograms of the expression of various marker combinations, which does not exploit the high-dimensional nature of the data. In this paper we utilize a form of dimensionality reduction - which we refer to as Information Preserving Component Analysis (IPCA) - that preserves the information distance between multi-dimensional data sets. As such, we offer a method for clinicians to visualize patient data in a low-dimensional projection space defined by a linear combination of all available markers. We illustrate these results on actual patient data.