2-D Bayesian Displacement Estimation Improves Contrast-to-Noise/Resolution Trade-off in Shear Wave Elasticity Imaging

For time-of-flight reconstructions of shear wave speed (SWS) data, tissue displacements are typically tracked using windowed 1-D normalized cross correlation (NCC). The shear wave arrival times at laterally offset tracking positions are used for the reconstruction of the shear wave speed. Recent work has pointed out that one main source of error in SWS imaging is the uncertainty in the lateral tracking positions. Here, the tracking positions can be laterally shifted towards constructive interference of scatterers within the PSFs, used for the axial window during NCC estimation. Since the variance of the tracking position has a strong effect on the SWS estimates, large regression kernels and regularization are used, and a trade-off between an increased contrast-to-noise ratio (CNR) and a high lateral resolution is required. In this work, we present a novel Bayesian displacement estimation scheme incorporating RF data from a local 2-D neighborhood of uncorrelated speckle ensembles. The incorporation of a-priori knowledge from a 2-D neighborhood leads to a reduced variance of the tracking position and thus to an overall reduced variance of the shear wave arrival times. In simulation experiments, SWS images based on the proposed Bayesian estimator and the NCC estimator are compared with regard to the CNR and the resolution. The SWS images based on the proposed method yield an increase in CNR of up to twofold compared to the SWS images of the NCC estimation of similar resolution.