|Abstract:|| The ability of automatically reconstructing patient-specific physiological shapes from medical images and of generating computational meshes is enabling the use of computational fluid dynamics as an additional tool to investigate the functioning of many human body systems, such as cardiovascular and respiratory systems, to comprehend the occurrence of organs pathologies and to provide aid to clinical practice.
Starting from geometries reconstructed from medical images, the use of efficient instruments able to create high quality meshes even in very complicated geometries
is required to build accurate surface meshes suitable for fluid dynamic simulations.
This thesis presents a set of procedures which aims at improving the mesh used for visualizations of the investigated geometry, whose quality is often neither suitable for mesh generation nor for flow solution. We present the use of mesh enhancement techniques to improve and maximize the quality of the initial triangulation, together with a curvature based method to optimize mesh resolution according to a geometric
error estimator. The proposed optimization strategy is applied to reconstructed geometries of two cases of physiological interests, a nasal cavity and a by-pass
Finally, on the graft geometry, we executed a fluid dynamic simulation in order to show the benefits provided by an algorithm capable of generating meshes that capture all the details of the shape of a complicated geometry, by means of a comparison between the results obtained in a non-optimised mesh. The results are
illustrative of the sensitivity of the flow dynamics to the geometry, and therefore to the mesh generation process.|