Development of finite element analysis of magnetic resonance elastography to investigate its potential use in abdominal aortic aneurysms
Hollis2016 Matlab codes.zip (8.344Kb)
Hollis2016 Python codes.zip (74.53Kb)
Hollis, Lyam Mark
MetadataShow full item record
Abdominal aortic aneurysm (AAA) is a form of cardiovascular disease whereby a change in the material properties of the vessel wall results in a localised dilation of the abdominal aorta. The primary risk of AAAs is rupture with mortality rates close to 90%. Whilst surgical intervention can be performed to repair AAAs, such procedures are considered high risk. As a result, surgery is only performed upon AAAs that are considered likely to rupture. The current method of prediction is the diameter criterion, with surgical intervention performed if the diameter of the AAA exceeds 5.5cm. Research has demonstrated that this is a weak method of predicting rupture and as such other methodologies are sought. One promising method is patient specific modelling (PSM) which involves the reconstruction of individual patient AAA geometries from imaging datasets, and finite element analysis (FEA) to calculate the stresses acting on the AAA wall, with the peak stress typically used as the predictor. A weakness of this methodology is the lack of patient specific material property values defined in the simulation. A potential technique to address this limitation is magnetic resonance elastography (MRE), an MR-based technique which utilises a phase-contrast sequence to characterise displacements caused by shear waves induced into the tissue by an external mechanical driver. An inversion algorithm is used to calculate local material property values of the tissue from these displacements. The aim of this thesis was to investigate the capability of utilising MRE to obtain material property measurements from AAAs that could be incorporated into PSM. To achieve this an FE method of modelling MRE was developed. The influence of modelling parameters upon the material property measurements made using the direct inversion (DI) algorithm was investigated, with element type and boundary conditions shown to have an effect. The modelling technique was then utilised to demonstrate the influence that the size of an insert had upon shear modulus measurements of that insert using DI in both 2- and 3-dimensions, and the multi-frequency dual elasto-visco algorithm (MDEV), an extension of DI combining information from multiple frequencies. Meanwhile a comparison of the modelling technique against an MRE scan of a phantom showed that whilst measurements made from the two techniques were different at low frequencies, they became similar as the frequency increased. This suggested that such differences were attributable to increased noise in the scanned data. FEA of MRE performed on idealised AAA geometries demonstrated that AAA size, shear viscosity of the thrombus and shear modulus of the AAA wall all influenced the accuracy of MRE measurements in the thrombus. Meanwhile MRE scanning of a small cohort of AAA patients had been undertaken and phase images investigated for signs of wave propagation to investigate the capabilities of the current MRE setup. Phase images were dominated by noise and there was no wave propagation visualised in any of the AAAs. This thesis demonstrates that the current MRE setup is not capable of achieving accurate measurements of material properties of AAA for PSM. Visualisation of wave propagation in AAAs is technically demanding and requires further development. A more fundamental concern however is the size dependence of the inversion algorithm used and the inability to consistently make accurate measurements from AAA geometries.