An Integrated Computational Framework for Cardiovascular Shape and Flow Analysis

Cardiovascular diseases remain the leading cause of death worldwide, with conditions such as thoracic aortic aneurysms and congenital heart diseases representing a significant clinical burden. While imaging modalities such as CT and MRI provide detailed anatomical information, they are not sufficient to capture hemodynamic conditions, which are crucial for diagnosis, prognosis, and treatment planning. In this context, computational modeling, through Statistical Shape Modeling (SSM) and Computational Fluid Dynamics (CFD), enables a quantitative link between anatomy and biomechanics. This thesis introduces a novel non-rigid registration algorithm that allows the inclusion of supra-aortic vessels in statistical models of the thoracic aorta, improving anatomical realism and the definition of boundary conditions in CFD simulations. The proposed framework is applied to investigate the relationship between aortic morphology and hemodynamics, to study the impact of vascular geometry in VA-ECMO patients, and to demonstrate its generalizability to pulmonary arteries in Tetralogy of Fallot. Furthermore, integration with artificial intelligence techniques enables the generation of realistic synthetic datasets for training deep learning models. Overall, this work shows how combining shape modeling, CFD, and AI can support the advancement of precision cardiovascular medicine.