Learning beamforming in ultrasound imaging

Abstract

Medical ultrasound (US) is a widespread imaging modality owing its popularity to cost efficiency, portability, speed, and lack of harmful ionizing radiation. In this paper, we demonstrate that replacing the traditional ultrasound processing pipeline with a datadriven, learnable counterpart leads to significant improvement in image quality. Moreover, we demonstrate that greater improvement can be achieved through a learning-based design of the transmitted beam patterns simultaneously with learning an image reconstruction pipeline. We evaluate our method on an in-vivo first-harmonic cardiac ultrasound dataset acquired from volunteers and demonstrate the significance of the learned pipeline and transmit beam patterns on the image quality when compared to standard transmit and receive beamformers used in high frame-rate US imaging. We believe that the presented methodology provides a fundamentally different perspective on the classical problem of ultrasound beam pattern design.

Deep Learning in Ultrasound Imaging

“…Several other data-driven beamforming methods have recently been proposed. In contrast to [39], these are mostly based on "general-purpose" deep neural networks, such as stacked autoencoders [40], encoder-decoder architectures [41], and fully-convolutional networks that map pre-delayed channel data to beamformed outputs [42]. In the latter, a 29-layer convolutional network was applied to a 3D stack of array response vectors for all lateral positions and a set of depths, to yield a beamformed in-phase and quadrature output for those lateral positions and depths.…”

“…Delay-and-sum beamformers are typically hand-tailored based on knowledge of the array geometry and medium properties, often including specifically designed array apodization schemes that may vary across imaging depth. Interestingly, it is possible to learn the delays and apodizations from paired channel-image data through gradient-descent by dedicated "delay layers" [39]. To show this, unfocused channel data was obtained from echocardiography of six patients for both singleline and multi-line acquisitions.…”

Deep Learning Applied to Beamforming in Synthetic Aperture Ultrasound

“…They propose to train a network that takes MLA/MLT channel data as an input, and reconstructs images such that it mimics the operation of a single-line acquisition (SLA). In [13], they proposed to learn the parameters of the forward model simultaneously with the image reconstruction process, that is, to jointly learn the endto-end transmit and receive beamformers.…”

Artificial Intelligence in Quantitative Ultrasound Imaging: A Review

“…Multi-line acquisition (MLA) has been used in cardiac ultrasound imaging to obtain a high frame rate so as to accurately capture rapid motion. Vedula et al proposed to train two encoder-decoder neural networks for the I and Q signals and apply this technique to the whole transmit-receive pipeline [39]. The unfocused multi-line channel data as input was fed into the network and the output was beamformed image which was as close to the image from single-line acquisition as possible via minimizing the l 1 distance.…”