Ultrasound deep beamforming using a multiconstrained hybrid generative adversarial network

Zixia Zhou,Yi Guo,Yuanyuan Wang

DOI: https://doi.org/10.1016/j.media.2021.102086

IF: 10.9

2021-07-01

Medical Image Analysis

Abstract:Ultrasound beamforming is a principal factor in high-quality ultrasound imaging. The conventional delay-and-sum (DAS) beamformer generates images with high computational speed but low spatial resolution; thus, many adaptive beamforming methods have been introduced to improve image qualities. However, these adaptive beamforming methods suffer from high computational complexity, which limits their practical applications. Hence, an advanced beamformer that can overcome spatiotemporal resolution bottlenecks is eagerly awaited. In this paper, we propose a novel deep-learning-based algorithm, called the multiconstrained hybrid generative adversarial network (MC-HGAN) beamformer that rapidly achieves high-quality ultrasound imaging. The MC-HGAN beamformer directly establishes a one-shot mapping between the radio frequency signals and the reconstructed ultrasound images through a hybrid generative adversarial network (GAN) model. Through two specific branches, the hybrid GAN model extracts both radio frequency-based and image-based features and integrates them through a fusion module. We also introduce a multiconstrained training strategy to provide comprehensive guidance for the network by invoking intermediates to co-constrain the training process. Moreover, our beamformer is designed to adapt to various ultrasonic emission modes, which improves its generalizability for clinical applications. We conducted experiments on a variety of datasets scanned by line-scan and plane wave emission modes and evaluated the results with both similarity-based and ultrasound-specific metrics. The comparisons demonstrate that the MC-HGAN beamformer generates ultrasound images whose quality is higher than that of images generated by other deep learning-based methods and shows very high robustness in different clinical datasets. This technology also shows great potential in real-time imaging.

engineering, biomedical,computer science, interdisciplinary applications, artificial intelligence,radiology, nuclear medicine & medical imaging

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to reduce the computational complexity while achieving high - quality image generation in ultrasonic imaging. Specifically, although the traditional Delay - and - Sum (DAS) method has a fast calculation speed, the spatial resolution of the generated image is low; while the adaptive beamforming method can improve the image quality, but the computational complexity is too high, which limits its practical application. Therefore, researchers urgently need an advanced beamformer that can overcome the bottleneck of spatial - temporal resolution and maintain low computational complexity.

For this purpose, this paper proposes a new deep - learning algorithm based on the Multiconstrained Hybrid Generative Adversarial Network (MC - HGAN), called the MC - HGAN beamformer. This method quickly realizes high - quality ultrasonic imaging by directly establishing a single - shot mapping between radio - frequency signals and reconstructed ultrasonic images. The MC - HGAN beamformer extracts radio - frequency signals and image features through two specific branches and integrates these features through a fusion module. In addition, the researchers also introduced a multi - constraint training strategy, which jointly constrains the training process by calling intermediate results to provide comprehensive guidance for the network. This beamformer is designed to be adaptable to different ultrasonic emission modes, which improves its versatility in clinical applications.

In general, this paper aims to develop an efficient and high - quality ultrasonic beamforming method to meet the high requirements for imaging quality and speed in practical applications.