Figure 1.
Left arm vasculature divided into arterial, venous and anastomosis segments (middle).
These segments locally describe the relation between pressure p and flow q via a lumped parameter approach (right), and consists of a resistor R (viscous resistance to blood flow), a resistor RL (viscous resistance of blood flow to small side-branches), an inductor L (blood inertia) and a capacitor C (vascular compliance). The anastomosis is modeled with two nonlinear resistors Rv and Rd. The windkessels consist of two resistors, Zwk and Rwk (together the peripheral resistance) and a capacitor Cwk (peripheral compliance). This figure is adapted from Huberts et al.
Table 1.
The names of all vessels included in the computational domains.
Figure 2.
Schematic picture of the locations of the diameter measurements in the preoperative DUS examination (left).
At the right a schematic picture of the postoperative flow determination is presented.
Table 2.
Patient-specific measurements performed in the context of the study.
Figure 3.
Schematic visualization of predicted and measured postoperative flows for the AVF configuration created by the vascular surgeon.
The error bars in predicted flow are the result of inaccuracies in input parameters, while the error bars in postoperative flow are the result of measurement inaccuracies21. For patient #15 prosthetic graft material was used for VA creation. For patient #21 no postoperative flow measurements could be obtained due to immediate thrombosis. A green circle around the patient identification represents overlap between predicted and measured postoperative flow (16 patients). A red square around the patient identification represents a discrepancy between predicted and measured postoperative flow (7 patients). Note that patient #24 received an alternative AVF configuration (cephalic vein was anastomosed with the ulnar artery on the upper arm due to a high brachial artery bifurcation).
Figure 4.
The predicted postoperative flows for a RCAVF, BCAVF and BBAVF configuration.
The flows are presented as the median of all Monte Carlo simulations with their 25th and 75th percentile interval. In 4 patients postoperative brachial artery flow could not be simulated for all three AVF configurations because essential patient-specific data were missing due to thrombosis of the cephalic vein (patient #19, #23, and #25) or because the computations did not converge for all Monte Carlo simulations (patient #24). An asterix represents the fistula configuration chosen by the surgeon.