Numerical and experimental study of blood flow through a patient-specific arteriovenous fistula used for hemodialysis

Abstract 

Arteriovenous fistula (AVF) pathologies related to blood flow necessitate valid calculation tools for local velocity and wall shear stress determination to overcome the clinical diagnostic limits. To illustrate this issue, a reconstructed patient-specific AVF was investigated, using computational fluid dynamics (CFDs) and particle image velocimetry (PIV). The aim of this study was to validate the methodology from medical images to numerical simulations of an AVF by comparing numerical and experimental data. Two numerical grids were presented with a refinement difference of a factor of four. A mold of the same volume was created and mounted on an experimental bench with similar boundary conditions. The patient's acquired echo D006Fppler flow waveform was injected at the arterial inlet. Experimental and numerical velocity vector cartography qualitatively produced similar flow fields. Quantification with a point-to-point approach thoroughly investigated the velocity profiles using the mean difference between both results. The finest mesh generated CFD results with a mean percentage of the difference in velocity magnitude, taking the PIV as reference, did not exceed 10%. At specific zones, the coarse mesh required adaptive meshing to improve fitting with experimental data. Meshing refinement was necessary to improve velocity accuracy at wide diameters and wall shear stress at narrow diameters. Provided that these criteria were properly respected, we show through this difficult example the validity of using CFD to properly describe flow patterns in image-based reconstructed blood vessels.

Keywords: Particle image velocimetry PIV, Computational fluid dynamics CFD, Reconstruction, Angioscanner, Vascular access, Hemodynamics