RESUMEN
BACKGROUND: Venous sinus stenosis can be associated with cerebrovascular disorders. Understanding the role of blood flow disturbances in these disorders is often hampered by the lack of patient-specific flow rates. Our goal was to demonstrate the impact of this by predicting individual flow rates retrospectively from routine manometry and angiography. METHODS: Ten cases, spanning a range of stenosis severities and pressure gradients, were selected from a cohort of patients who had undergone venous stenting for pulsatile tinnitus. Lumen geometries were digitally segmented from CT venograms. A simplified Bernoulli formula was derived to estimate individual cycle-average flow rates from clinical pressure gradients and minimum lumen cross-section areas. High-fidelity pulsatile computational fluid dynamics (CFD) simulations were performed to compare predictions of flow disturbances using generic versus individual flow rates, and to validate the Bernoulli formula. RESULTS: Individual flow rates derived from the Bernoulli formula deviated by up to 47% from the assumed generic flow rate, resulting in substantial differences in CFD predictions of post-stenotic flow instabilities. Pressure gradients estimated by the simplified Bernoulli formula were, however, highly predictive of pressure gradients from the full CFD simulations (R2=0.95; slope=0.98, 95% CI 0.88 to 1.09). CONCLUSIONS: A simple Bernoulli formula can predict CFD-estimated trans-stenotic pressure gradients in realistic venous geometries. As demonstrated here, this may be used to recover individual flow rates from routine-but-invasive clinical measurements; however, it also suggests a simpler path towards non-invasive estimation of trans-stenotic pressure gradients that may avoid some of the challenges associated with 4D flow MRI approaches.
RESUMEN
Computational fluid dynamics (CFD) of cerebral venous flows has become popular owing to the possibility of using local hemodynamics and hemoacoustics to help diagnose and plan treatments for venous diseases of the brain. Lumen geometries in low-pressure cerebral veins are different from those in cerebral arteries, often exhibiting fenestrations and flattened or triangular cross section, in addition to constrictions and expansions. These can challenge conventional size-based volume meshing strategies, and the ability to resolve nonlaminar flows. Here we present a novel strategy leveraging estimation of length scales that could be present if flow were to become transitional or turbulent. Starting from the lumen geometry and flow rate boundary conditions, centerlines are used to determine local hydraulic diameters and cross-sectional mean velocities, from which flow length scales are approximated using conventional definitions of local Kolmogorov and Taylor microscales. By inspection of these scales, a user specifies minimum and maximum mesh edge lengths, which are then distributed along the model in proportion to the approximated local Taylor length scales. We demonstrate in three representative cases that this strategy avoids some of the pitfalls of conventional size-based strategies. An exemplary CFD mesh-refinement study shows convergence of high-frequency flow instabilities even starting from relatively coarse edge lengths near the lower bounds of the approximated Taylor length scales. Rational consideration of the length scales in a possibly nonlaminar flow may thus provide a useful and replicable baseline for denovo meshing of complicated or unfamiliar venous lumen geometries.