Estimation of model parameters characterizing dispersion in ARFI induced shear waves in in vivo human liver

Abstract

We consider the analysis of shear wave dispersion following ARFI excitation in a cohort of 135 Non-Alcoholic Fatty Liver Disease patients traditionally characterized as “difficult-to-image.” Three analysis methods are considered: (1) measuring k(ω) by locating the maximum signal from the two-dimensional Fourier transform (2D-FT) of propagating shear wave data, (2) an extension of this method using model-based sums through 2D-FT data, and (3) shear wave spectroscopy. A linear dispersion model is used to characterize the frequency-dependent phase velocity. The analysis methods are evaluated in terms of robustness as determined by the rate of successful measurements in the patient data, and by the degrees of correlation and bias inherent with each method as determined by viscoelastic FEM validation studies. For these patient data and analysis procedures, although the 2D-FT methods are more robust than the shear wave spectroscopy method, they are also systematically biased. Even so, they can be used can be used to characterize the relative viscoelastic properties of liver.