Reconstruction of the Intrarenal Vascular Tree From μ-CT Scans of Corrosion Cast Specimens

Reconstructing renal arterial trees from μ-CT scans of corrosion cast specimens is challenging due to imaging artifacts, noise, and inconsistent material properties that hinder automated analysis. This study aims to develop a robust methodology for transforming μ-CT data into semantically and topologically meaningful models, enabling accurate morphometric and topological analysis. We propose a three-stage algorithmic pipeline consisting of adaptive binarization tailored to each sample’s characteristics, artifact removal using region labeling and manual trimming, and multi-scale convolution with spherical kernels for smooth reconstruction. The methodology was validated using artificial renal models generated with FractaL-Tree software, 3D printed, μ-CT scanned, and processed using the proposed pipeline to replicate real-world conditions. Visual inspection confirmed that the algorithm effectively addressed discontinuities and preserved anatomical integrity. Numerical validation, based on ground-truth models, demonstrated high agreement in key geometric parameters such as branching angles and edge counts, with minor discrepancies limited to the smallest vessels. Of the 33 human intrarenal vascular trees, 30 were successfully reconstructed into clean, artifact-free, anatomically realistic models. These reconstructions support advanced structural analysis and have potential applications in anatomical research, disease modeling, and preoperative planning for minimally invasive kidney procedures.