Difference Autocorrelation: A Novel Approach to Estimate Shear Wave Speed in the Presence of Compression Waves

TL;DR

This paper introduces the difference autocorrelation estimator, a new method for accurately measuring shear wave speed in elastography despite interference from compression waves and noise, with validation across simulations, ultrasound, and MRI.

Contribution

The study presents the difference autocorrelation estimator, a novel technique that improves shear wave speed estimation accuracy in challenging conditions compared to conventional methods.

Findings

Accurately estimates shear wave speed with less than 2% error in simulations.

Achieves 9% error in ultrasound elastography of a breast phantom.

Attains 19% error in magnetic resonance elastography of a brain phantom.

Abstract

In share wave elastography (SWE), the aim is to measure the velocity of shear waves, however unwanted compression waves and bulk tissue motion pose challenges in evaluating tissue stiffness. Conventional approaches often struggle to discriminate between shear and compression waves, leading to inaccurate shear wave speed (SWS) estimation. In this study, we propose a novel approach known as the difference autocorrelation estimator to accurately estimate reverberant SWS in the presence of compression waves and noise. Methods: The difference autocorrelation estimator, unlike conventional techniques, computes the subtraction of velocity between neighboring particles, effectively minimizing the impact of long wavelength compression waves and other wide-area movements such as those caused by respiration. We evaluated the effectiveness of the integrated difference autocorrelation (IDA) by: (1) using k-Wave simulations of a branching cylinder in a soft background, (2) using ultrasound elastography on a breast phantom, (3) using ultrasound elastography in the human liver-kidney region, and (4) using magnetic resonance elastography (MRE) on a brain phantom. Results: By applying IDA on the unfiltered contaminated wave fields of simulation and elastography experiments, the estimated SWSs are in good agreement with the ground truth values (i.e., less than 2% error for the simulation, 9% error for ultrasound elastography of breast phantom and 19% error for MRE). Conclusion: Our results demonstrate that IDA accurately estimates SWS, revealing the existence of a lesion, even in the presence of strong compression waves. Significance: IDA exhibits consistency in SWS estimation across different modalities and excitation scenarios, highlighting its robustness and potential clinical utility.