Spatially-Weighted Inverse Formulation for Enhanced Ultrasound Attenuation Imaging

Abstract

Quantitative ultrasound imaging is a modality that aims to characterize intrinsic tissue properties like attenuation and backscatter coefficients. The regularized spectral log difference (RSLD-TV) method, which introduced total variation regularization to stabilize attenuation imaging, is able to generate accurate attenuation maps. However, the performance of the method degrades when significant changes of backscatter amplitude occur. Variations of the method were introduced involving a weighted approach to backscatter regularization (RSLD-SWTV), which however is not effective when changes in both attenuation and backscatter are present.The present work introduces a novel variation of RSLD (RSLD-WFR) that combines an L1-norm prior for the backscatter term with spatially-varying weights for both the fidelity and regularization terms derived from an initial estimation of the changes in backscatter. A comparative analysis is performed with simulated, phantom and clinical data against RSLD-TV and RSLD-SWTV. The proposed method reduced the root mean square error by factors of 3 and 2 in simulations and phantom data, while more than doubling the contrast-to-noise ratio. The results in vivo showcase not only an increase in imaging quality but also a significant reduction in estimation bias.