NOVEL STRATEGIES FOR THE MORPHOLOGICAL AND BIOMECHANICAL ANALYSIS OF THE CARDIAC VALVES BASED ON VOLUMETRIC CLINICAL IMAGES

This work was focused on the morphological and biomechanical analysis of the heart valves exploiting the volumetric data. Novel methods were implemented to perform cardiac valve structure and sub-structure segmentation by defining long axis planes evenly rotated around the long axis of the valve. These methods were exploited to successfully reconstruct the 3D geometry of the mitral, tricuspid and aortic valve structures. Firstly, the reconstructed models were used for the morphological analysis providing a detailed description of the geometry of the valve structures, also computing novel indexes that could improve the description of the valvular apparatus and help their clinical assessment. Additionally, the models obtained for the mitral valve complex were adopted for the development of a novel biomechanical approach to simulate the systolic closure of the valve, relying on highly-efficient mass-spring models thus obtaining a good trade-off between the accuracy and the computational cost of the numerical simulations. In specific: • First, an innovative and semi-automated method was implemented to generate the 3D model of the aortic valve and of its calcifications, to quantitively describe its 3D morphology and to compute the anatomical aortic valve area (AVA) based on multi-detector computed tomography images. The comparison of the obtained results vs. effective AVA measurements showed a good correlation. Additionally, these methods accounted for asymmetries or anatomical derangements, which would be difficult to correctly capture through either effective AVA or planimetric AVA. • Second, a tool to quantitively assess the geometry of the tricuspid valve during the cardiac cycle using multidetector CT was developed, in particular focusing on the 3D spatial relationship between the tricuspid annulus and the right coronary artery. The morphological analysis of the annulus and leaflets confirmed data reported in literature. The qualitative and quantitative analysis of the spatial relationship could standardize the analysis protocol and be pivotal in the procedure planning of the percutaneous device implantation that interact with the tricuspid annulus. • Third, we simulated the systolic closure of three patient specific mitral valve models, derived from CMR datasets, by means of the mass spring model approach. The comparison of the obtained results vs. finite element analyses (considered as the gold-standard) was performed tuning the parameters of the mass spring model, so to obtain the best trade-off between computational expense and accuracy of the results. A configuration mismatch between the two models lower than two times the in-plane resolution of starting imaging data was yielded using a mass spring model set-up that requires, on average, only ten minutes to simulate the valve closure. • Finally, in the last chapter, we performed a comprehensive analysis which aimed at exploring the morphological and mechanical changes induced by the myxomatous pathologies in the mitral valve tissue. The analysis of mitral valve thickness confirmed the data and patterns reported in literature, while the mechanical test accurately described the behavior of the pathological tissue. A preliminary implementation of this data into finite element simulations suggested that the use of more reliable patient-specific and pathology-specific characterization of the model could improve the realism and the accuracy of the biomechanical simulations.