Particle image velocimetry for personalised patient specific arteriovenous access evaluation

Chronic kidney disease (CKD) is the condition where renal function declines or renal structure is damaged. The disease has five stages of severity, where the final stage is known as end stage renal disease (ESRD) or end stage kidney disease (ESKD). This occurs when patients have lost complete function of their kidneys. To deal with this loss, patients require renal replacement therapy. Kidney transplantation is the ideal form of therapy, however due to a shortage of donors, hemodiaylsis is the most popular form of treatment used. To perform hemodialysis, vascular access with high flow rates is required. Three types of vascular access are used for hemodialysis, arterio-venous fistula (AVF), arterio-venous graft (AVG) and central venous catheter (CVC). AVFs are the preferred choice for hemodialysis as they have better patency rates and present less complications than other vascular access.

However AVFs continue to present issues with primary and secondary patency rates. Common complications of AVFs are: failure to mature, stenosis development or thrombosis formation. The hemodynamics in AVFs stimulate remodelling however they can also damage the endothelial cells lining the fistula walls triggering intimal hyperplasia formation eventually leading to stenosis. To improve AVF patency and gain an insight into AVF failure, an understanding of the hemo?dynamics involved is necessary.

The focus of this thesis is to examine the hemodynamics within a patient-specific brachio-cephalic AVF under different flow and wall conditions using particle image velocimetry. Steady (Re = 1817) and patient-specific pulsatile flow (Reav = 1817, Remax = 2233) were modelled through a rigid AVF to investigate the velocity flow conditions. Velocity magnitude and streamline plots were compared for steady, phase average pulsatile and average of the waveform. The bulk flow features were similar in all three, with a large region of recirculation in the distal artery with a smaller area in the proximal vein and flow impingement at the toe. These areas of recirculating flow can affect the efficiency of dialysis and possibly lead to failure. Disturbed flow in AVFs can also initiate intimal hyperplasia. The flow pattern did not vary significantly along the pulsatile waveform.

Two velocity based wall shear stress parameters, coefficient of variation and directional variability, were also studied. Directional variability was greatest in the regions of flow recirculation and impingement while the coefficient of variation was highest at the walls and where the flow begins to separate. Areas with high values of directional variability and coefficient of variation suggest regions of high transverse wall shear stress and temporal wall shear stress gradient at the walls respectively.

To determine the affect of wall compliance on AVF hemodynamics, a compliant phantom of the same AVF was fabricated and PIV results were compared. Streamline and non-dimensional velocity plots were compared for both steady (Re = 1817) and pulsatile flow (Reav =1817). The main flow features illustrated a similar pattern, a high speed jet of flow entering through the proximal artery, splitting in the anastomosis where most of the flow leaves the proximal vein. Recirculation zones were present in the distal artery and in the heel of the anastomosis while flow impingement was observed at the anastomosis toe. The only difference noted was that the non-dimensional velocity was lower in the rigid model. Different time points on the waveform were also set side by side and depicted similar flow features however a difference was also detected in the non-dimensional velocity.

Lastly the differences in coefficient of variation and directional variability were inspected. In both models, the coefficient of variation was largest at the walls and where the flow begins to separate while the directional variability was greatest in the areas of recirculation and flow impingement, however the extent of these regions is greater in the compliant model. This suggests that while a rigid model is capable of capturing the flow pattern in an AVF, a compliant model is necessary to obtain the real flow and finer hemodynamics details.