Layer-Wise Speed-of-Sound Estimation and Beamforming of Segmented Media.

Abstract

Speed-of-sound (SoS) is a fundamental acoustic property of tissues that is essential for ultrasound image beamforming. The SoS can also act as a quantitative biomarker for pathology in tissue. Typically, the beamforming SoS is assumed to be 1540 m/s for human imaging, but it may be imprecise for imaging scenarios with inhomogeneous tissue. In this work, we present a novel framework for multi-layer SoS estimation and propose a SoS-aware (SoSA) beamformer to realize high-quality ultrasound imaging. Our framework consists of three core steps: segmentation of the media layers, sequential estimation of each layer's SoS assuming intra-layer homogeneity, and SoSA beamforming based on the estimated SoS map. We validated our algorithm in vitro, ex vivo, and in vivo in comparison to through-transmission SoS measurements. Across 126 stacked agar phantom experiments (pairwise combinations of six staircase phantoms with different SoS values), the average SoS estimation error of our framework was 4.9 m/s over the top layers and 1.6 m/s over bottom layers. In nine stacked bovine and porcine sample experiments, we achieved an average error of 2.7 m/s with improved point target lateral resolution of 32.5% compared to conventional beamforming with a nominal SoS of 1540 m/s. When indirectly evaluated over five human calves, our algorithm achieved a mean error of 7.9 m/s for the average calf SoS. Also, in human quadriceps imaging scenarios, our proposed framework showed image quality enhancements with improved visibility of the fascicle structure. Overall, our new technique improves.