Proposal of a Method to Identify Vascular Endothelial Cells from Images of Mouse Myocardial Tissue

Abstract

Three-dimensional (3D) analysis of cardiomyocytes using computers is urgently required to elucidate the pathogenesis of cardiac diseases. As there exists several different types of cells in microscopic images, it is necessary to classify the cells prior to analysis. To classify cells, the fact that the cell nuclei of vascular endothelial cells are covered by vascular endothelial cell membrane is considered. In this study, the cell nuclei and areas of the vascular endothelial cell membrane are extracted from microscopic images of myocardial tissue in the left atrium of newborn mice and classified into vascular endothelial cell nuclei and non-endothelial cell nuclei based on the quantity of the cell nuclei that is covered by the vascular endothelium. The accuracy was calculated from the experimental results, and it was more than 85%. Future studies will include improving the accuracy of the classification and further classifying the cell nuclei that were determined to be other than endothelial cell nuclei to analyze the cells. These are expected to contribute to the research to clarify the mechanism of blood vessels and cells in myocardium.