Quantitative precision of an automated, fluorescence-based image cytometer

Abstract

Digital image-based cytometry of clinical specimens labeled with fluorescent, disease-specific markers holds promise for becoming an important diagnostic and prognostic technique because the technique can make a diverse range of quantitative biochemical, morphologic, densitometric and contextual measurements on intact specimens. It has been previously shown by us, using an image cytometer (IC) consisting entirely of commercially available components, that the nuclei of individual cells in slide-supported specimens can be detected automatically using a fluorescent DNA stain and image analysis software. The purpose of this study was to determine the precision of the IC for quantifying the integrated fluorescence intensity and area of fluorescent standard beads and nuclei. Integrated intensities could be quantified to between 2.3% and 3.5% precision using a 40x objective lens and between 1.6% and 2.3% using a 20x objective. The main contribution to this uncertainty was 2% inaccuracy in determining the variations in sensitivity over the imaging area. Areas could be quantified to between 0.91% and 2.1% using a 40x objective and between 2.8% and 3.2% using a 20x objective. Significant quantification errors were introduced if the objects were not in focus or were touching each other. Overall, however, these results demonstrated that image cytometry of fluorescence-stained specimens can yield quantitative results with sufficient precision for determining DNA ploidy distributions and for making other measurements on clinical specimens.