A patient-specific FSI model for vascular access in haemodialysis

This research forms part of an interdisciplinary project that aims to improve the understanding of haemodynamics and vascular mechanics in arteriovenous shunting. To achieve the high flow rates that enable patients with renal disease to receive haemodialysis, a fistula is created between an artery and a vein. The patency rate of fistulas, especially those located in the upper arm, is low. The approach adopted here makes use of new magnetic resonance image (MRI) technology and computational modelling of blood flow, with a view to improving therapeutic strategies of disease requiring vascular interventions. This thesis presents the construction and development of a 3D finite element model of the fluid-structure interaction in a brachial–cephalic patient–specific fistula. An overview of the mathematical models that describe the vessel wall and fluid behaviour as well their interaction with each other is given. An Arbitrary Lagrangian- Eulerian (ALE) framework is used together with a transversely isotropic hyperelastic constitutive model for the vessel walls, while blood flow is modelled as a Newtonian fluid. A three-element Windkessel model is used to allow the fluid to move through the outlets of the computational domain without causing non–physical reflections. Flow data acquired from MRI is used to prescribe the flow at the inlet. The parameters of the Windkessel-model at the two outlets are calibrated to resemble the flow acquired from the 2D MRI. The model is validated against the flow patterns acquired from the 4D MRI. The flow patterns of the blood, and stress present in the vessel are investigated. Of special significance are the flow and wall shear stress at the anastomosis. An area of very high velocity in the anastomosis is followed by an area of recirculation and low velocity. The propagation of pressure waves and their reflection at the anastomosis are studied. Areas that are subjected to low wall shear stress, high oscillatory wall shear stress or flow circulation are identified as areas where intimal hyperplasia may develop. The flow results from the simulation show good qualitative agreement with the MRI data