We are presenting an algorithm for creation the three-dimensional model of cerebral vascular network based on two types of non-contrast enhanced magnetic resonance angiograms. Those types of images allow for arterial tree visualization (Time of Flight) or both arterial and venous trees visualization (Phase Contrast). First step of the algorithm is pre-processing of the images. It includes resizing of the images, segmentation of the skull interior area and application of the morphological white top-hat (WTH) transformation in order to highlight the sections of blood vessels since their intensities have higher values than the image background. All images were also subjected to the nonlinear gamma correction in order to improve the contrast of the images. Developed algorithm for creation the structure of blood vessels uses either binarization with hysteresis or binarization with threshold based on the image histogram for the segmentation of the vessel cross-sections. The further process rely on selection of the vessels centroids, and construction of resulting graph structure. The vascular trees obtained from two types of images were compared. Obtained three-dimensional model of cerebral vascular network is useful for the diagnosis of many brain diseases (e.g., intracranial saccular aneurysms ). It can serve for the further research including numerical simulations of blood flow/perfusion or prediction of the drug distribution in the brain.