Phase Aberration Robust Beamformer for Planewave US Using Self-Supervised Learning

Abstract

Ultrasound (US) is widely used for clinical imaging applications thanks to its real-time and non-invasive nature. However, its lesion detectability is often limited in many applications due to the phase aberration artefact caused by variations in the speed of sound (SoS) within body parts. To address this, here we propose a novel self-supervised 3D CNN that enables phase aberration robust plane-wave imaging. Instead of aiming at estimating the SoS distribution as in conventional methods, our approach is unique in that the network is trained in a selfsupervised manner to robustly generate a high-quality image from various phase aberrated images by modeling the variation in the speed of sound as stochastic. Experimental results using real measurements from tissue-mimicking phantom and in vivo scans confirmed that the proposed method can significantly reduce the phase aberration artifacts and improve the visual quality of deep scans.

Phase Aberration Correction: A Convolutional Neural Network Approach

“…Shen et al utilized a CNN to estimate the aberrated point spread function from beamformed IQ data and subsequently applied the inverse filter to rectify the data [36]. Additionally, there are DL-based beamformers designed to exhibit robustness to the aberration by suppressing off-axis scattering [37] or by mapping images beamformed with randomly perturbed sound speed values to clean images beamformed with a reference sound speed value [38].…”

“…As a result, the aforementioned methods had to rely solely on simulated data for training, leading to a drop in performance when testing on experimental data due to the domain shift problem. Recent studies have recognized the need to eliminate the requirement of ground truths; however, even in such efforts, reconstructed images with a fixed sound speed value of 1540 m/s were still considered clean images [38]. Our contributions can be summarized as follows:…”