Effects of Coherence-Based Beamforming on Breast Ultrasound Elastograms

In quasi-static elastography, the operator compresses the tissue under exam with the probe, while the scanner processes the beamformed frames, by means of correlation or signal phase analysis techniques, to estimate both the displacement and the strain. However, the quality of estimates may be influenced by the speckle quality and hence by the implemented beamforming algorithm. Therefore, in this work we investigate the effects of coherence-based beamforming techniques, providing better images than conventional Delay and Sum (DAS), on the estimation of axial/lateral displacement and strain in quasi-static elastography of the breast. Simulations were used to evaluate the error on axial and lateral displacement estimations obtained by 2D normalized cross-correlation (2D- NCC) with polynomial fitting, as compared to the ground truth given by finite-element modeling. Experimental breast phantom acquisitions where instead evaluated in terms of elastographic contrast-to-noise-ratio (CNRe) and non-uniformity (NU) level. Results show that similar performance in axial displacement estimation is achieved by 2D-NCC with all beamformers, while using coherence beamforming provides more accurate estimates of lateral displacement. However, the analysis of CNRe and NU in experimental axial/lateral strain images does not show any significant improvement as compared to DAS when any of these beamformers is employed.