Multiparameter image cytometry: From confocal micrographs to subcellular fluorograms

Abstract

Multifluorescence labeling is routinely performed to detect the spatial coincidence between several markers within biological specimens. We have recently developed image correlation methods to identify double fluorescent structures by virtue of local similarities between fluorescence distributions. We extend this approach here to analyze statistically the fluorescence distribution of structures of interest on micrographs. This digital cytometry relies mainly on the segmentation of multifluorescence images. Once identified, all objects are analyzed through a range of attributes estimating size, morphology, fluorescence content and mean colocalization level between fluorescence channels. The data sets which are saved in flow cytometry standard (FCS) files, allow multiparameter classification of objects and subpopulation counting. The combination of fluorescence, morphometric and local image correlation attributes has been applied here to compare the frequency of single- and multiple-labeled structures at the subcellular level.