Tuning of boundary conditions parameters for hemodynamics simulation using patient data

This thesis describes an engineering workflow, which allows specification of boundary conditions and 3D simulation based on clinically available patient-specific data. A review of numerical models used to describe the cardiovascular system is provided, with a particular focus on the clinical target disease chosen for the toolkit, aortic coarctation. Aorta coarctation is the fifth most common congenital heart disease, characterized by a localized stenosis of the descending thoracic aorta. Current diagnosis uses invasive pressure measurement with rare but potential complications. The principal objective of this work was to develop a tool that can be translated into the clinic, requiring minimum operator input and time, capable of returning meaningful results from data typically acquired in clinical practice. Linear and nonlinear 1D modelling approaches are described, tested against full 3D solutions derived for idealized geometries of increasing complexityand for a patient-specific aortic coarctation. The 1D linear implementation is able to represent the fluid dynamic in simple idealized benchmarks with a limited effort in terms of computational time, but in a more complex case, such as a mild aortic coarctation, it is unable to predict well 3D fluid dynamic features. On the other side, the 1D nonlinear implementation showed a good agreement when compared to 3D pressure and flow waveforms, making it suitable to estimate outflow boundary conditions for subject-specific models. A cohort of 11 coarctation patients was initially used for a preliminary analysis using 0D models of increasing complexity to examine parameters derived when tuning models of the peripheral circulation. The first circuit represents the aortic coarctation as a nonlinear resistance, using the Bernoulli pressure drop equation, without considering the effect of downstream circulation. The second circuit include a peripheral resistance and compliance, and separate ascending and descending aortic pressure responses. In the third circuit a supra-aortic Windkessel model was added in order to include the supra-aortic circulation. The analysis detailed represents a first attempt to assess the interaction between local aortic haemodynamics and subject-specific parameterization of windkessel representations of the peripheral and supra-aortic circulation using clinically measured data. From the analysis of these 0D models, it is clear that the significance of the coarctation becomes less from the simple two resistance model to the inclusion of both the peripheral and supra-aortic circulation. These results provide a context within which to interpret outcomes of the tuning process reported for a more complex model of aortic haemodynamics using 1D and 3D model approaches. Earlier developments are combined to enable a multi-scale modelling approach to simulate fluid-dynamics. This includes non-linear 1D models to derive patient-specific parameters for the peripheral and supra-aortic circulation followed by transient analysis of a coupled 3D/0D system to estimate the coarctation pressure augmentation. These predictions are compared with invasively measured catheter data and the influence of uncertainty in measured data on the tuning process is discussed. This study has demonstrated the feasibility of constructing a workflow using non-invasive routinely collected clinical data to predict the pressure gradient in coarctation patients using patient specific CFD simulation, with relatively low levels of user interaction required. The results showed that the model is not suitable for the clinical use at this stage, thus further work is required to enhance the tuning process to improve agreement with measured catheter data. Finally, a preliminary approach for the assessment of change in haemodynamics following coarctation repair, where the coarctation region is enlarged through a virtual intervention process. The CFD approach reported can be expanded to explore the sensitivity of the peak ascending aortic pressure and descending aortic flow to the aortic diameter achieved following intervention, such an analysis would provide guidance for surgical intervention to target the optimal diameter to restore peripheral perfusion and reduce cerebral hypertension