Beamforming With deep learning from a single plane wave RF data

Deep learning approaches for improving ultrasound (US) image reconstruction have proved successful in both experimental and clinical settings. In this paper, we present an autoencoder-based deep learning framework for ultrasound beamforming from a single plane wave RF data. Motivated by U-Net, the network consists of an encoder and a decoder. The network is trained and evaluated with both phantom and \textit{in vivo} datasets. The beamformed images from the proposed network using one RF plane wave reached a 0.86 SSIM score, 37.48 PSNR, 3.84 mean SNR, 9.58 mean CNR, which outperforms the standard delay and sum (DAS) beamforming algorithm.The results demonstrate that the proposed network is capable of generating high quality US images from only one RF plane wave without introducing computational complexity, which offers a feasible beamforming method that can be potentially developed for multiple ultrasound-based tasks.