Fast compressive pulse-echo ultrasound imaging using random incident sound fields

Abstract

In fast pulse-echo ultrasound imaging (UI), the image quality is traded off against the image acquisition rate by reducing the number of sequential wave emissions per image. To alleviate this tradeoff, the concept of compressed sensing (CS) was proposed by the authors in previous studies. CS regularizes the linear inverse scattering problem (ISP) associated with fast pulse-echo UI by postulating the existence of a nearly-sparse representation of the object to be imaged. This representation is obtained by a known linear transform, e.g., the Fourier or a wavelet transform. A central degree of freedom in the regularized ISP is the choice of the incident sound fields. Previous studies focused exclusively on steered plane waves. In this study, we investigate the usage of random incident sound fields to improve the relevant mathematical properties of the scattering operator governing the linear ISP. These sound fields are synthesized by a linear transducer array whose physical elements are excited applying combinations of random time delays and random apodization weights. Using simulated and experimentally obtained radio frequency signals, we demonstrate that these sound fields significantly reduce the recovery errors and improve the rate of convergence for low signal-to-noise ratios.