Efficient cardio-vascular 4D-Flow MRI enabled CFD to improve in-silico predictions of post-surgical haemodynamics in individual patients

This thesis focuses on creating a workflow that combines four dimensional flow magnetic resonance imaging with computational fluid dynamics techniques, and identifying the main difficulties that are associated with patient-specific modelling. With further development, the proposed work- flow will allow post-surgical haemodynamics to be predicted prior to surgical intervention taking place, ensuring the best possible outcome is achieved for the individual patient.
The use of patient-specific computational fluid dynamic modelling in diagnostics and risk stratification, treatment planning, and surgical intervention is quickly becoming an invaluable tool and has proven key in multiple medical advances and breakthroughs. However, existing methods to combine medical imaging and computational fluid dynamics techniques often require invasive procedures to collect appropriate patient-specific data, require expensive software licenses, or have significant limitations within the methodologies, such as inlet conditions or spatial resolutions.
The research within this thesis provides a workflow to combine four dimensional flow magnetic resonance imaging and computational fluid dynamics, using open source software when possible, and a non-invasive and non-ionising imaging technique. The major challenges of patient-specific modelling are investigated. By increasing the complexity of the workflow incrementally, the impacts of physiologically accurate inlet boundary conditions are assessed, as is the human error that is introduced into patient-specific modelling through the geometry reconstruction process. The workflow created is tested on a wide age range of patients and bicuspid aortic valve phenotypes.
To validate the workflow created, the methods used were applied to an anatomical flow phantom, therefore the in-vivo challenges of the thoracic aorta moving radially and vertically, and the systemic circulatory system distal to the outlets were removed. This research has shown that the workflow proposed produces good agreement with four dimensional flow magnetic resonance imaging data, notably in the ascending aorta during the systolic phase of the cardiac cycle.
A significant challenge of patient-specific modelling that is often acknowledged yet not fully quantified is the spatial resolution of the four dimensional flow magnetic resonance imaging. Research therefore focused on determining how the spatial resolution at which the four dimensional flow magnetic resonance imaging data is acquired at impacts the subsequent patient-specific computational fluid dynamics simulations. The results presented show that coarse spatial resolutions have a significant impact on the results of numerical simulations. From the results presented, a recommendation of a minimum spatial resolution that should be used when conducting patient-specific simulations was made to avoid errors being introduced into the numerical simulations.