The Geometric Dependence of Wave Velocity in Carotid Arteries: Phantom and Finite Element Study and Implications for Vascular "Shear Wave" Elastography.

Objective: The intrinsic elasticity and the structural stiffness of blood vessels are widely regarded as important biomarkers for prediction of cardiovascular disease risk, the leading cause of death worldwide. Ultrasound-based shear wave elastography (SWE) has been used to measure these properties in several clinical studies. However, the geometric properties of blood vessels complicate the relationship between wave speed and elasticity in blood vessels compared to bulk tissue. Here we quantify these effects with a semi-analytical finite element (SAFE) model and ultrasound experiments and discuss their implications for the vascular "shear wave" (better described as "guided wave") elastography literature.

Methods: A previously developed SAFE model was employed to simulate wave propagation after insonification with an acoustic radiation force (ARF) in 4,437 combinations of vascular geometry and elasticity. Group velocities were extracted and underwent processing analogous to what a SWE-equipped ultrasound scanner would perform to estimate Young's modulus (E=3ρC_s²), and were compared to the true Young's modulus used in the simulation. Additionally, 23 polyvinyl alcohol cryogel tubes of different geometries and elasticities were constructed and underwent ARF-based SWE and reference inflation-based mechanical testing. Wave speeds were converted to Young's modulus using the same method as in the finite element study and were compared with the Young's modulus obtained from mechanical testing.

Results: Both the SAFE simulations and PVA tube ultrasound experiments confirm the dependence of wave speed on vascular geometry which leads to a severe, geometry-dependent underestimation of Young's modulus if geometry is not considered. We identify and discuss 11 recent papers that have used clinical ultrasound systems to measure elasticity in common carotid arteries and argue that geometry (distinct from elasticity) may have contributed to differences in scanner-reported elasticities between groups in some cases.