There is an increased desire for miniature ultrasound probes with small apertures to provide volumetric images at high frame rates for in-body applications. Satisfying these increased requirements makes simultaneous achievement of a good lateral resolution a challenge. As micro-beamforming is often employed to reduce data rate and cable count to acceptable levels, receive processing methods that try to improve spatial resolution will have to compensate the introduced reduction in focusing. Existing beamformers do not realize sufficient improvement and/or have a computational cost that prohibits their use. Here we propose the use of adaptive beamforming by deep learning (ABLE) in combination with training targets generated by a large aperture array, which inherently has better lateral resolution. In addition, we modify ABLE to extend its receptive field across multiple voxels. We illustrate that this method improves lateral resolution both quantitatively and qualitatively, such that image quality is improved compared with that achieved by existing delay-and-sum, coherence factor, filtered-delay-multiplication-and-sum and Eigen-based minimum variance beamformers. We found that only in silica data are required to train the network, making the method easily implementable in practice.

@en

Ultrasound imaging is an attractive imaging modality due to its low-cost and real-time feedback, although it often falls short in image quality compared to MRI and CT imaging. Conventional ultrasound image reconstruction, such as Delay-and-Sum beamforming, is derived from maximum-likelihood estimation. As such, no prior information is exploited in the image formation process, which limits potential image quality. Maximum-a-posteriori (MAP) beamforming aims to overcome this issue, but often relies on rough approximations of the underlying signal statistics. Deep learning based reconstruction methods have demonstrated impressive results over the past years, but often lack interpretability and require vast amounts of data.In this work we present a neural MAP beamforming technique, which efficiently combines deep learning in the MAP beamforming framework. We show that this model-based deep learning approach can achieve high-quality imaging, improving over the state-of-the-art, without compromising the real-time abilities of ultrasound imaging.@en

3-D contrast enhanced ultrasound enables better visualization of inherently 3-D vascular geometries compared to an intersecting plane. Additionally, it would allow the application of motion correction techniques for all directions. Both contrast detection and motion correction work better on high-frame rate data. However high-frame rate 3-D ultrasound imaging with dense matrix arrays is challenging to realize. Sparse arrays alleviate some of the limitations in cable count and data rate that fully populated arrays encounter, but their increased level of secondary lobes negatively impacts image contrast. Meanwhile the use of unfocused transmit beams needed to achieve high-frame rates negatively impacts resolution. Here we propose to use adaptive beamforming by deep learning (ABLE) to improve the image quality of contrast enhanced ultrasound images acquired with a sparse spiral array. We train the neural network on simulated data and evaluate simulated images and in vivo images of an ex ovo chicken embryo. ABLE improved resolution compared to delay-and-sum (DAS) and spatial coherence (SC) beamforming on the simulated and in vivo data. The qualitative improvements persist after histogram matching, indicating that the image quality improvement of the ABLE images was not purely due to dynamic range stretching.

@en

The image quality of ultrasound localization microscopy (ULM) images is driven by the ability to accurately detect and track the location of microbubbles (MBs) in vascular networks. This task becomes increasingly challenging in imaging environments with high MB concentrations and low signal-to-noise ratios, making it difficult to differentiate and localize individual MBs. Recent developments in deep learning (DL) have demonstrated significant improvements over conventional methods but depend on vast amounts of realistic training data with the corresponding ground truth labels, which are difficult to obtain. The alternative, simulated data, in turn, poses challenges in generalizability of the method. In this work, we present a hybrid pipeline for ULM that comprises data generation, localization, and tracking. It combines the current state-of-the-art, utilizing both conventional and DL techniques. We show that using this approach, we can create high-quality velocity maps while being able to generalize well across different domains.

@en