On the development of assessment techniques for improved prediction of arteriovenous graft performance in facilitating haemodialysis

Renal failure is a condition where the kidneys cannot provide sufficient filtration of waste product from the blood. A kidney transplant is the ideal treatment, but donor numbers fall short of demand. The alternative treatment option is filtration of the blood through a dialyzer. To facilitate the extraction and returning of the blood, an artificial conduit is constructed between a high pressure artery to a low pressure vein. This ArterioVenous (AV) junction facilitates the required flow rates for the haemodialysis treatment. This is typically made possible directly connecting the artery to the vein (fistula) or by using an artificial material instead (graft), which is the primary focus of this work. This thesis explores the implications of a simplified flow regime being used in Computational Fluid Dynamics (CFD) to study a frequently noted complex fluid flow scenario. A factor that is acknowledged within vascular access (VA) studies is the frequent presence of disturbed flow. Turbulent flow can have dramatic effects at cellular level, affecting both endothelial and blood cells. The presence of turbulence alters the wall shear stress (WSS) distribution from laminar flow even at similar Reynolds numbers. Despite this, it is frequently ignored and numerical models employ laminar assumptions. To examine the differences induced with this assumption, various pseudo-realistic models are computationally modelled in CFD using Large Eddy Simulation (LES). These models vary in feeding vein diameter, graft angle, flow distribution and Reynolds number. These are then solved using a steady Navier{Stokes approximation to examine differences from the LES solutions. The laminar assumption performed poorly in several cases, often due to the presence of turbulence below the critical Reynolds number. High fluctuations were also noted that may affect correlations between WSS related parameters and reported disease correlation. However, no commonly used disease metric correlated to these high fluctuations. These same models are then tested against various turbulence mod- els, such as the k - RNG model, k- SST model and Transitional k -SST model, that are less computationally expensive than LES to provide a solver that describes the AV flow conditions but that also is computationally less expensive than LES. No optimal option was found with the models being representative at best. Currently a wealth of geometric data in VA is not captured since there is infrequent need for MRI or CT data, as angiograms regularly suffice. Since the VA surgery is open and reasonably visible, a novel single camera approach is proposed to capture the geometric data. The proposed concept is ideal for the surgical environment, since it is requires little in the way of setup, is non contact and rapid. It there- fore minimally interferes with a surgery procedure. To examine the efficcacy of the method, a contorted silicone model of known geometry is employed. The geometric error and repeatability was examined. Computer simulations (structural and fluid flow) were conducted on the reproduced geometries to examine if computational values and trends were comparable to the control model values and an assessment of the influence geometry had on these values and trend correlations was established. Finally, a novel AV graft design is designed by using a CT-scan data of a subjects arm. The subjects arm was reconstructed virtually allowing for placement of the graft in an augmented design space. The graft is then connected to the artery and vein with the connections positioned around bone and under surface tissue. This allows adjustment of geometric parameters of the graft on a more realistic scale. This graft was manufactured and then used in porcine preclinical studies. In this preclinical study, the single camera approach is utilised to examine its functionality in a surgical environment. Models were reconstructed and then simulated using the chosen CFD solver, to examine the flow in the novel graft in the preclinical model. Results highlighted the positive features of the novel graft when compared to a standard graft.