摘要： Kidney stones are a common clinical problem that affects up to 20% of the population. For larger kidney stones a percutaneous nephrolithotomy (PCNL) is the first line treatment. The procedure begins with a percutaneus puncture of the kidney to gain an access to intrarenal part of urinary tract, where the stone is located. Needle must reach precisely the targeted spot, passing through retroperitoneal cavity and kidney parenchyma without collision with adjacent organs or larger intrarenal blood vessels. In this paper, an algorithm dedicated for kidney arterial system segmentation is proposed. For validation, its’ results were compared with manual segmentations of ten exemplary 3D images providing satisfactory outcomes. The effectiveness and reliability of this algorithm was also demonstrated by incorporating its results into 3D anatomical model prepared as an image-guidance for real PCNL procedure. The description of the proposed algorithm and methodology of the construction of aforementioned model is presented in this paper. It consists of a registration of different phases of CT images together with segmentation of particular structures of interest (kidney, stones, urinary tract, kidney vasculature). The proposed solution allows to obtain 3D patient-specific models in a far more efficient manner than manually or by the use of semiautomated methods giving excellent and reliable visualization of intrarenal structures and by thus can stand clinically useful add in preoperative planning and intraoperative support in minimally invasive urological interventions.