Wave atom based Compressive Sensing and adaptive beamforming in ultrasound imaging

The paper investigates combining Compressive Sensing (CS) with the robust Capon beamformer (RCB) for the purpose of medical ultrasound image formation with a much reduced number of samples compared to those used in current state-of-art ultrasound. The proposed CS algorithm uses wave atom dictionary as a low dimension projection, a Bernouli random matrix as a sensing matrix and a regularized-l₁ optimization technique for recovery. The reconstructed signals are then pre-processed before using the RCB technique augmented with spatial smoothing and diagonal loading. This approach is demonstrated through simulations, wire phantom and in vivo cardiac data with a reduction of up to 1/8 in the processed data rate and ultrasound images of similar perceived quality.