L1-Norm-Regularized Particle Motion Estimation in Ultrasound Shear Wave Elastography

Ultrasound shear wave elastography (SWE) measures tissue elasticity by generating, detecting, and tracking shear waves to reveal clinically important diagnostic information. In SWE, displacement tracking reliability predominantly regulates the accuracy of shear wave particle displacement and overall shear wave speed estimation. However, phase-based techniques, known to produce noisy displacement estimates, are traditionally employed for particle displacement tracking in SWE. This noisy displacement estimation issue can be addressed by leveraging regularized displacement tracking algorithms, yet a regularized technique has the potential to oversmooth displacement estimates. To strike a balance between displacement regularity and irregularity, we incorporated an L1-norm regularization-based displacement tracking technique named L1-SOUL to estimate shear wave particle displacement. When applied to an experimental phantom dataset provided by the IEEE UFFC-JS 2024 Algorithms for Mapping Elastic Modulus (a-MEM) challenge, the proposed technique generates a high-quality shear wave speed (SWS) map with pronounced visual contrast and sharp inclusion edges. The contrast-to-noise ratio (CNR), edge width, and edge localization error associated with the SWS map are 4.249, 3.656 mm, and 0.819 mm, respectively, culminating in securing the best overall rank for the investigated dataset among 18 a-MEM challenge submissions. Results show the capability of L1-SOUL to mitigate the limitations of SWE and expand its clinical applications.