Abstract:
This thesis investigated methods to evaluate myocardial mechanics using non-invasive Magnetic Resonance Imaging (MRI), in order to aid in the diagnosis and assessment of heart disease. Firstly, a statistical analysis of cardiac shape and function was developed to reduce the number of parameters required to describe heart geometry and motion with finite element models. The Principal Component Analysis (PCA) resulted in a compact description of left ventricular deformation and visualisation of the main modes of motion. The PCA was then used to aid the segmentation of cardiac magnetic resonance images. Secondly, an apparatus and method was developed for estimation of soft tissue material properties using MRI. The apparatus allowed controlled mechanical experiments to be performed in which boundary conditions are recorded. The method was validated using a silicon gel phantom with known material parameters. Monte Carlo simulations were performed to estimate the robustness of the parameters. Thirdly, a method was developed for fitting finite element models of the myocardial fibre field architecture to Diffusion Tensor Magnetic Resonance Imaging (DTMRI) data. DTMRI provided a high resolution, fast through put method of determining the fibre architecture in the heart, suitable for quantitative studies of disease states in the entire heart in statistically significant numbers. The fibre models can then be used for stress analysis and activation studies. Fourthly, four isolated arrested porcine hearts were imaged with MRI whilst undergoing passive inflation. Using geometry from MRI, recorded boundary conditions, muscle fibre architecture from diffusion tensor imaging, and deformation from magnetic resonance tagging, finite element models were constructed to solve the finite elasticity stress estimation problem. Optimisation was performed to estimate the constitutive parameters of a Fung type exponential material law by minimising the difference between the modelled deformation and the imaged deformation. The optimised parameters were in a similar range as those reported by other studies. Finite element models of myocardial mechanics can thus be used to extract meaningful biophysical parameters from non invasive MRI data.