Mice are widely used in cardiovascular research. Quantitative characterization of the vascular structure is crucial for the analysis of the coronary network, which can be imaged by microtomodensitometry (microCT) after perfusion with a contrast medium. The aim of this study was to propose a computational processing of microCT imaging of mouse heart that ensures objective segmentation of the coronary vasculature and quantification of its 3D architecture. All procedures, performed on 6 C57BL/6J mice, were done accorded with current national and European legislations, and agreed by local ethical committee. 2 hours after injection of anticoagulant and vasodilator treatment, mice were euthanatized by pentobarbital intraperitoneal injection (600 mg.kg-1), exsanguinated, and perfused with a contrast medium (Neoprene latex and barium sulphate) via the brachiocephalic artery trunk. After dissection, hearts were imaged with a microCT system (Skyscan, Brucker) with 12x12x12 μm resolution, followed by 3D reconstruction (NRecon, Brucker). Image processing was done using CTAn (Brucker) and Image J with BoneJ plugin. The cardiac muscle was segmented from background and ventricular cavities, and then the coronary artery from the myocardium, by 2 successive thresholds based on the statistical distribution of pixels according to their grey levels and independently from any visual estimate. In 3 specimens, the aortic valve did not resist the perfusion pressure, and the left ventricle resulted perfused. A shape-based segmentation was applied to delete the ventricle from the coronary network. After segmentation, the following parameters were calculated: the volumes of the organ (Vh) and of the coronary arteries network (Vc), and their ratio as an estimate of the density of the coronary artery network (Dc), and its fractal dimension (Df), which measures the capacity of a self-similar structure to fill the three-dimensional space it occupies. The thickness (Tc) and the length (Lc) of the vascular segments, and their frequency distribution, and the total length of the coronary arteries (Lcn), were calculated. Statistical analyses were done with Gradpad Prism software. Data are given as mean±SEM. Vh was 134.7±9.0 mm3, Vc was 1.4±0.1 mm3, Dc was 1.05±0.06%, and Df was 1.65±0.02. Tc was 9.8±0.2 µm. Thickness frequency distribution followed a Gaussian law, and showed that vessels of 12 µm diameter or more can be detected, but with underestimation of the smaller ones. Total Lcn was was 158.3±13.9 mm, and average Lc was 249±18 µm. Segment length frequency distribution followed a one-order exponential decay from shortest (0-100 µm) to longest segments. We conclude that this computational processing applied to 3D microCT imaging ensures objective calculation of quantitative parameters that characterize the 3D structure of the coronary vasculature, including the microcirculation, down to 36 µm (3 pixel size).