Full-hexahedral structured meshing for image-based computational vascular modeling

Abstract

Image-based computational modeling offers a virtual access to spatially and temporally high resolution flow and structural mechanical data in vivo. Due to inter-subject morphological variability, mesh generation represents a critical step in modeling the patient-specific geometry and is usually performed using unstructured tetrahedral meshing algorithms. Although hexahedral structured meshes are known to provide higher accuracy and reduce the computational costs both for Finite Element Analysis and Computational Fluid Dynamics, their application in computational cardiovascular studies is challenging due to the complex 3D and branching topology of vascular territories. In this study, we propose a robust procedure for structured mesh generation, tailoring the mesh structure to the subject-specific vessel topology. The proposed methodology is based on centerline-based synthetic descriptors (i.e. centerlines, radii and centerlines' normals) which are used to solve the meshing problem following a bottom-up approach. First, topologically equivalent block-structures are placed inside and outside the lumen domain. Then, a projection operation is performed, returning a parametric volume mesh which fits the original triangulated model with sub-micrometric accuracy. Additionally, a three-layered arterial wall (resembling the intima, media and adventitia) is artificially generated, with the possibility of setting variable thickness (e.g. proximal-to-distal tapering) and material anisotropy (e.g. position-dependent collagen-fibers' orientation). This new meshing procedure, implemented using open-source software packages only, is demonstrated on two challenging human cases, being an aortic arch and an abdominal aortic aneurysm. High-quality meshes are generated in both cases, according to shape-quality metrics. By increasing the computation accuracy, the developed meshing tool has the potential to further add "confidence" to the use of computational methods in vascular applications.