COMPLETE MATHEMATICAL-MODELING METHOD FOR THE ANALYSIS OF IMMUNOFLUORESCENCE DISTRIBUTIONS COMPOSED OF NEGATIVE AND WEAKLY POSITIVE CELLS

In a recent paper (Lampariello: Cytometry 15:294-301, 1994), we propos ed a method for the automated evaluation of the percentage of positive cells from how cytometric immunofluorescence histograms, The method i s based on a suitable mathematical representation of the control histo gram, which is used to identify the negative cell distribution in the test histogram. In this paper we present an improvement of the previou s method, where we assume that the positive cell distribution in the t est can also be modeled making use of an empirical distribution of the same kind as employed for modeling the control. The parameters of thi s distribution are estimated directly from the test. In this way, a ma thematical representation of the whole test distribution is calculated without having to set up a purely positive control. In order to evalu ate the accuracy of the method in the determination of the positive pe rcentage, we carried out a set of measurements of double-labeled and s uitably treated cells, mixed in different ratios with control cells, a nd from each sample we obtained histograms with overlapped and well-se parated positive and negative distributions. These last histograms all ow us to determine the actual positive percentages and thus to evaluat e the performance of the analysis method applied to the histograms wit h overlapped distributions. (C) 1998Wiley-Liss, Inc.