Abstract
We developed a system for quantifying the numbers of bromodeoxyuridine (BrdU)-labeled hepatocyte nuclei in rat and mouse liver with an automated image analysis system. We began by developing a protocol for BrdU staining that would provide consistently intense staining to facilitate identification of both labeled and unlabeled nuclei by image analysis. Preliminary studies detected and characterized hepatocyte nuclei and differentiated them from non-hepatocyte nuclei using area and form factors. The parameters were selected to optimize discrimination between the two populations, selecting 90% of hepatocyte and 5% non-hepatocyte nuclei. Finally, we developed a program for automatic counting of BrdU-labeled hepatocyte and total hepatocyte nuclei. Results obtained from this method correlated well with data collected by a microscopist over a wide range of labeling indices. The automated system reduces interobserver variation and should minimize intraobserver error, as well as reducing the tedium of measuring labeling indices in the liver. Moreover, the techniques described should be applicable to other tissues.