Image analysis for the study of radionuclide distribution in tissue sections.
Academic Article
Overview
abstract
UNLABELLED: Tissue section autoradiographs are often prepared to review the precise spatial locations of a radiolabeled molecule relative to cells, such as in the study of radiolabeled antibody distribution. The objective of this work was to develop and evaluate a method to automatically detect both grains and cell nuclei from stained tissue autoradiographs using a microscope and an image analyzer. METHOD: Using a sequence of morphological image operations, the densely stained regions of the section, representing the cell nuclei are identified first, and then subtracted from the original image. This enables the identification of autoradiographic grains under conditions of variable contrast, by separation of the grains overlapping the cell nuclei from the extracellular spaces, permitting a more accurate and robust automatic segmentation routine. RESULTS: The accuracy of the method to detect grains has been evaluated at different threshold levels. The highest accuracy obtained was approximately 90%. The accuracy in the detection of cell nuclei was histology-dependent. As examples, we have estimated accuracies of approximately: 86%, 81% and 77% for kidney, EL-4 lymphoma and pneumonocyte sections, respectively. CONCLUSION: This method was tested using specimens designed to study radiolabeled antibody distribution, but it should be applicable with comparable accuracy to other radiolabeled compounds for which quantitative information on the heterogeneity of distribution is required.
