Maximum likelihood estimation of Young's modulus in transient elastography with unknown line-of-sight orientation

Transient elastography can be used to measure tissue elasticity by applying a mechanical stress constraint and measuring the velocity of propagation of the induced shear wave, assumed to be proportional to tissue elasticity. In this paper we study two original maximum-likelihood (ML) approaches for shear wave velocity estimation on RF ultrasound signals acquired with a transient elastography setup. For acquisitions made with a line of sight (LOS) aligned with the directions of propagation (DOP) of the shear wave, a simple parametric model was derived from the theoretical Green's function, enabling ML estimation of the elasticity. For non-aligned LOS and DOP, an empirical approach was considered to learn a simple time-delay model of the displacement field, using an annotated database of simulated data. A ML estimator was then defined to jointly estimate the angle of the LOS and the elasticity of the tissue. The proposed methods were evaluated on simulations, and RF signals acquired on phantom objects and on volunteers, for liver screening. Results reported very high accuracy, with elasticity errors of measures below 10%.