Sparse Convolutional Beamforming for Ultrasound Imaging

Ultrafast 3D synthetic aperture imaging with Hadamard-encoded aperiodic interval codes and aperiodic sparse arrays with separate transmitters and receivers

3D synthetic aperture (SA) imaging of volumes can be obtained using sparse 2D ultrasound arrays. However, even with just 256 elements, the volumetric imaging rate can be relatively slow due to having to transmit on each element in succession. Hadamard Aperiodic Interval (HAPI) codes can be used to image the full SA dataset in one extended transmit to speed up the synthetic aperture imaging, but their long nature produces large deadzones if the same elements are used as both transmitters and receivers. In this simulation study, we use a 2D Costas sparse array with separate transmitters and receivers to remedy the deadzone problem, and use it with the HAPI-coded imaging scheme to obtain fully transmit-receive focused, wide field-of-view 3D volumes with high-resolution and high SNR at ultrafast volumetric imaging rates of more than 500 volumes per second, almost nine times faster than non-coded SA imaging with the same imaging parameters. We show similar PSF performance compared to non-coded SA, and a 26 dB improvement in SNR with order-256 HAPI codes. We also present cyst simulations showing similar contrast for the HAPI-coded SA method compared to non-coded SA in the context of no noise, and improved contrast in the context of noise.

Branched Convolutional Neural Networks for Receiver Channel Recovery in High-Frame-Rate Sparse-Array Ultrasound Imaging

For high-frame-rate ultrasound imaging, it remains challenging to implement on compact systems as a sparse imaging configuration with limited array channels. One key issue is that the resulting image quality is known to be mediocre not only because unfocused plane-wave excitations are used but also because grating lobes would emerge in sparse-array configurations. In this article, we present the design and use of a new channel recovery framework to infer full-array plane-wave channel datasets for periodically sparse arrays that operate with as few as one-quarter of the full-array aperture. This framework is based on a branched encoder-decoder convolutional neural network (CNN) architecture, which was trained using full-array plane-wave channel data collected from human carotid arteries (59 864 training acquisitions; 5-MHz imaging frequency; 20-MHz sampling rate; plane-wave steering angles between -15° and 15° in 1° increments). Three branched encoder-decoder CNNs were separately trained to recover missing channels after differing degrees of channelwise downsampling (2, 3, and 4 times). The framework's performance was tested on full-array and downsampled plane-wave channel data acquired from an in vitro point target, human carotid arteries, and human brachioradialis muscle. Results show that when inferred full-array plane-wave channel data were used for beamforming, spatial aliasing artifacts in the B-mode images were suppressed for all degrees of channel downsampling. In addition, the image contrast was enhanced compared with B-mode images obtained from beamforming with downsampled channel data. When the recovery framework was implemented on an RTX-2080 GPU, the three investigated degrees of downsampling all achieved the same inference time of 4 ms. Overall, the proposed framework shows promise in enhancing the quality of high-frame-rate ultrasound images generated using a sparse-array imaging setup.

Frame rate improvement in ultrafast coherent plane wave compounding

Coherent plane wave compounding (CPWC), as an ultrafast ultrasound imaging technique, makes a significant breakthrough in frame rate enhancement. However, there exists a compromise between the quality of the final image and the frame rate in CPWC. In this paper, we propose an efficient method to minimize the number of required emissions, and consequently, improve the frame rate, while maintaining the image quality. To this end, we down-sample the angle interval using two specific sampling factors. More precisely, we construct two different subsets, each of which consists of a few numbers of emissions. The optimal values of the angle intervals are achieved based on the beampattern that corresponds to the reference case (that is, the case where all plane waves are used). Finally, in order to keep the image quality comparable with the reference case, we apply some modifications to the image reconstruction procedure. In the proposed algorithm, the Delay-and-Sum beamformed images of two considered subsets are convolved to achieve the final reconstructed image. The obtained results confirm the efficiency of the proposed method in terms of frame rate improvement compared to the reference case. In particular, by using the proposed method, the required emissions in PICMUS data reduce to 16, which is 4.6 times smaller compared to the reference case. Also, the gCNR values of the proposed method and the reference case are obtained as 0.98 and 0.97, respectively, for in-vivo dataset. This demonstrates that the proposed method successfully preserves the quality of the reconstructed image by using much fewer emissions.

A joint method of coherence factor and nonlinear beamforming for synthetic aperture imaging with a ring array

Ultrasound computed tomography (USCT) with a ring array is an emerging diagnostic method for breast cancer. In the literature, synthetic aperture (SA) imaging has employed the delay-and-sum (DAS) beamforming technique for ring-array USCT to obtain isotropic resolution reflection images. However, the images obtained by the conventional DAS beamformer suffer from off-axis clutter and low resolution due to inhomogeneity of the medium and phase distortion. To address these issues, researchers have developed adaptive beamforming methods, such as coherence factor (CF) and convolutional beamforming algorithm (COBA), that improve image quality. In this study, we propose a joint method that combines CF with short-lag COBA (SLCOBA). First, we estimate the average sound speed using CF to address tissue inhomogeneity. Based on the corrected sound speed map, SLCOBA effectively suppresses side lobes and enhances image quality. Numerical results show that the proposed method reduces clutter and noise, improving resolution performance. These findings suggest that the proposed method may be a practical option for breast imaging in inhomogeneous mediums in the future.

On the physics of ultrasound transmission for in-plane needle tracking in guided interventions

Objective.In ultrasound (US) guided interventions, the accurate visualization and tracking of needles is a critical challenge, particularly during in-plane insertions. An inaccurate identification and localization of needles lead to severe inadvertent complications and increased procedure times. This is due to the inherent specular reflections from the needle with directivity depending on the angle of incidence of the US beam, and the needle inclination.Approach.Though several methods have been proposed for improved needle visualization, a detailed study emphasizing the physics of specular reflections resulting from the interaction of transmitted US beam with the needle remains to be explored. In this work, we discuss the properties of specular reflections from planar and spherical wave US transmissions respectively through multi-angle plane wave (PW) and synthetic transmit aperture (STA) techniques for in-plane needle insertion angles between 15°-50°.Main Results.The qualitative and quantitative results from simulations and experiments reveal that the spherical waves enable better visualization and characterization of needles than planar wavefronts. The needle visibility in PW transmissions is severely degraded by the receive aperture weighting during image reconstruction than STA due to greater deviation in reflection directivity. It is also observed that the spherical wave characteristics starts to alter to planar characteristics due to wave divergence at large needle insertion depths.Significance.The study highlights that synergistic transmit-receive imaging schemes addressing the physical properties of reflections from the transmit wavefronts are imperative for the precise imaging of needle interfaces and hence have strong potential in elevating the quality of outcomes from US guided interventional practices.

Ultrasound Signal Processing: From Models to Deep Learning

Medical ultrasound imaging relies heavily on high-quality signal processing to provide reliable and interpretable image reconstructions. Conventionally, reconstruction algorithms have been derived from physical principles. These algorithms rely on assumptions and approximations of the underlying measurement model, limiting image quality in settings where these assumptions break down. Conversely, more sophisticated solutions based on statistical modeling or careful parameter tuning or derived from increased model complexity can be sensitive to different environments. Recently, deep learning-based methods, which are optimized in a data-driven fashion, have gained popularity. These model-agnostic techniques often rely on generic model structures and require vast training data to converge to a robust solution. A relatively new paradigm combines the power of the two: leveraging data-driven deep learning and exploiting domain knowledge. These model-based solutions yield high robustness and require fewer parameters and training data than conventional neural networks. In this work we provide an overview of these techniques from the recent literature and discuss a wide variety of ultrasound applications. We aim to inspire the reader to perform further research in this area and to address the opportunities within the field of ultrasound signal processing. We conclude with a future perspective on model-based deep learning techniques for medical ultrasound.

Generative models for sound field reconstruction

This work examines the use of generative adversarial networks for reconstructing sound fields from experimental data. It is investigated whether generative models, which learn the underlying statistics of a given signal or process, can improve the spatio-temporal reconstruction of a sound field by extending its bandwidth. The problem is significant as acoustic array processing is naturally band limited by the spatial sampling of the sound field (due to the difficulty to satisfy the Nyquist criterion in space domain at high frequencies). In this study, the reconstruction of spatial room impulse responses in a conventional room is tested based on three different generative adversarial models. The results indicate that the models can improve the reconstruction, mostly by recovering some of the sound field energy that would otherwise be lost at high frequencies. There is an encouraging outlook in the use of statistical learning models to overcome the bandwidth limitations of acoustic sensor arrays. The approach can be of interest in other areas, such as computational acoustics, to alleviate the classical computational burden at high frequencies.

A Radon diffraction theorem for plane wave ultrasound imaging

The rising demand on high frame rate ultrasound imaging applications necessitates the development of fast algorithms for plane wave image reconstruction. We introduce a new class of plane wave reconstructions that relies on a relation between receive data and image data in the Radon domain. This relation is derived for arbitrary dimensions and validated on multiple two-dimensional plane wave data sets. We further present a mathematical relation between conventional delay-and-sum and Fourier domain reconstruction methods and the method proposed. Our analysis shows that they all rely on the same physical model with slight variations in certain filtering steps and, therefore, the new Radon domain reconstruction yields similar results as other methods in terms of image quality. However, we show that our method offers a huge potential to improve computation time by reducing the number of applied projections and to improve image quality by introducing nonlinear operations in the Radon domain, e.g., for edge enhancement. As the Radon transform retains both angular and temporal information, the relation also provides new insights on the fundamentals of plane wave imaging that can be leveraged for optimizing acquisition schemes or for developing novel compounding strategies in the future.

A convolutional plane wave model for sound field reconstruction

Spatial sound field interpolation relies on suitable models to conform to available measurements and predict the sound field in the domain of interest. A suitable model can be difficult to determine when the spatial domain of interest is large compared to the wavelength or when spherical and planar wavefronts are present or the sound field is complex, as in the near-field. To span such complex sound fields, the global reconstruction task can be partitioned into local subdomain problems. Previous studies have shown that partitioning approaches rely on sufficient measurements within each domain due to the higher number of model coefficients. This study proposes a joint analysis of all of the local subdomains while enforcing self-similarity between neighbouring partitions. More specifically, the coefficients of local plane wave representations are sought to have spatially smooth magnitudes. A convolutional model of the sound field in terms of plane wave filters is formulated and the inverse reconstruction problem is solved via the alternating direction method of multipliers. The experiments on simulated and measured sound fields suggest that the proposed method retains the flexibility of local models to conform to complex sound fields and also preserves the global structure to reconstruct from fewer measurements.

Minimizing Image Quality Loss After Channel Count Reduction for Plane Wave Ultrasound via Deep Learning Inference

High-frame-rate ultrasound imaging uses unfocused transmissions to insonify an entire imaging view for each transmit event, thereby enabling frame rates over 1000 frames per second (fps). At these high frame rates, it is naturally challenging to realize real-time transfer of channel-domain raw data from the transducer to the system back end. Our work seeks to halve the total data transfer rate by uniformly decimating the receive channel count by 50% and, in turn, doubling the array pitch. We show that despite the reduced channel count and the inevitable use of a sparse array aperture, the resulting beamformed image quality can be maintained by designing a custom convolutional encoder-decoder neural network to infer the radio frequency (RF) data of the nullified channels. This deep learning framework was trained with in vivo human carotid data (5-MHz plane wave imaging, 128 channels, 31 steering angles over a 30° span, and 62 799 frames in total). After training, the network was tested on an in vitro point target scenario that was dissimilar to the training data, in addition to in vivo carotid validation datasets. In the point target phantom image beamformed from inferred channel data, spatial aliasing artifacts attributed to array pitch doubling were found to be reduced by up to 10 dB. For carotid imaging, our proposed approach yielded a lumen-to-tissue contrast that was on average within 3 dB compared to the full-aperture image, whereas without channel data inferencing, the carotid lumen was obscured. When implemented on an RTX-2080 GPU, the inference time to apply the trained network was 4 ms, which favors real-time imaging. Overall, our technique shows that with the help of deep learning, channel data transfer rates can be effectively halved with limited impact on the resulting image quality.

A Low-Complexity and High-Resolution Beamformer for Portable Medical Ultrasound Imaging

So far, researchers have proposed various methods to improve the quality of medical ultrasound imaging. However, in portable medical ultrasound imaging systems, features, such as low cost and low power consumption for battery longevity, are very important. Hence, most of the proposed algorithms have not been proper substitutes for the delay and sum (DAS) algorithm in portable clinical applications due to their high computational complexity and cost. In this article, a new algorithm is presented concentrating on reducing the computational complexity based on a technique that separates the signal from the correlated interferences to overcome the negative characteristics, particularly for portable applications such as high price, high power consumption, and off-axis clutters in the azimuth direction. Also, the proposed algorithm yields a higher contrast compared to that of the DAS algorithm while achieving a similar computation complexity order of O ( n ) similar to the DAS algorithm. Furthermore, the performed simulations confirm that the proposed method is able to achieve a better resolution almost twice as that of the filtered delay multiply and sum (F-DMAS) algorithm with the same sidelobe level.

Deep-Learning Based Adaptive Ultrasound Imaging From Sub-Nyquist Channel Data

Traditional beamforming of medical ultrasound images relies on sampling rates significantly higher than the actual Nyquist rate of the received signals. This results in large amounts of data to store and process, imposing hardware and software challenges on the development of ultrasound machinery and algorithms, and impacting the resulting performance. In light of the capabilities demonstrated by deep learning methods over the past years across a variety of fields, including medical imaging, it is natural to consider their ability to recover high-quality ultrasound images from partial data. Here, we propose an approach for deep-learning-based reconstruction of B-mode images from temporally and spatially sub-sampled channel data. We begin by considering sub-Nyquist sampled data, time-aligned in the frequency domain and transformed back to the time domain. The data are further sampled spatially so that only a subset of the received signals is acquired. The partial data is used to train an encoder-decoder convolutional neural network (CNN), using as targets minimum-variance (MV) beamformed signals that were generated from the original, fully-sampled data. Our approach yields high-quality B-mode images, with up to two times higher resolution than previously proposed reconstruction approaches (NESTA) from compressed data as well as delay-and-sum (DAS) beamforming of the fully-sampled data. In terms of contrast-to- noise ratio (CNR), our results are comparable to MV beamforming of the fully-sampled data, and provide up to 2 dB higher CNR values than DAS and NESTA, thus enabling better and more efficient imaging than what is used in clinical practice today.

Compressed Fourier-Domain Convolutional Beamforming for Sub-Nyquist Ultrasound Imaging

Efficient ultrasound (US) systems that produce high-quality images can improve current clinical diagnosis capabilities by making the imaging process much more affordable and accessible to users. The most common technique for generating B-mode US images is delay-and-sum (DAS) beamforming, where an appropriate delay is introduced to signals sampled and processed at each transducer element. However, sampling rates that are much higher than the Nyquist rate of the signal are required for high-resolution DAS beamforming, leading to large amounts of data, making remote processing of channel data impractical. Moreover, the production of US images that exhibit high resolution and good image contrast requires a large set of transducer elements, which further increases the data size. Previous works suggest methods for reduction in sampling rate and in array size. In this work, we introduce compressed Fourier domain convolutional beamforming, combining Fourier domain beamforming (FDBF), sparse convolutional beamforming, and compressed sensing methods. This allows reducing both the number of array elements and the sampling rate in each element while achieving high-resolution images. Using in vivo data, we demonstrate that the proposed method can generate B-mode images using 142 times less data than DAS. Our results pave the way toward efficient US and demonstrate that high-resolution US images can be produced using sub-Nyquist sampling in time and space.

Acoustic-Field Beamforming-Based Generalized Coherence Factor for Handheld Ultrasound

Handheld ultrasound devices have been widely used for diagnostic applications. The use of the acoustic-field beamforming (AFB) method has been proposed for handheld ultrasound to reduce electricity consumption and avoid battery and unwanted heat issues. However, the image quality, such as the contrast ratio and contrast-to-noise-ratio, are poorer with this technique than with the conventional delay-and-sum method. To address the problems associated with the worse image quality in AFB imaging, in this paper we propose the use of an AFB-based generalized coherence factor (GCF) technique, in which the GCF weighting developed for adaptive beamforming is extended to AFB. Simulation data, experimental results, and in vivo testing verified the efficacy of our proposed AFB-based GCF technique.

Covariance Matrix-Based Statistical Beamforming for Medical Ultrasound Imaging

Medical ultrasound image quality is often limited by clutter, which is the dominant mechanism of image degradation. A variety of beamforming methods have been extensively studied to reduce clutter and, thus, enhance ultrasound image quality. This article introduces a new beamforming approach, called covariance matrix-based statistical beamforming (CMSB), to improve the image contrast and preserve the background speckle pattern while simultaneously achieving a high-resolution performance. In CMSB, adaptive selection of subarray length, diagonal reducing, and mean-to-standard-deviation ratio-based subarray averaging are inherently combined to differentiate and reduce off-axis energy effectively. Moreover, rotary averaging prior to diagonal reducing is introduced to preserve speckle statistics. Simulated, experimental, and in vivo datasets were used to evaluate the imaging performance of the proposed method. The quantitative results indicate that, compared with delay-and-sum (DAS) beamforming, CMSB leads to average improvements of 44.5% and 97.3% in lateral resolution and contrast, respectively, in phantom experiments. Our work shows that CMSB is capable of improving image resolution and contrast while maintaining the speckle reliably. Preliminary in vivo study also demonstrates that the CMSB can enhance image contrast and lesion detection.

Deep Unfolded Recovery of Sub-Nyquist Sampled Ultrasound Images

The most common technique for generating B-mode ultrasound (US) images is delay-and-sum (DAS) beamforming, where the signals received at the transducer array are sampled before an appropriate delay is applied. This necessitates sampling rates exceeding the Nyquist rate and the use of a large number of antenna elements to ensure sufficient image quality. Recently, we proposed methods to reduce the sampling rate and the array size relying on image recovery using iterative algorithms based on compressed sensing (CS) and the finite rate of innovation (FRI) frameworks. Iterative algorithms typically require a large number of iterations, making them difficult to use in real time. In this article, we propose a reconstruction method from sub-Nyquist samples in the time and spatial domain, which is based on unfolding the iterative shrinkage thresholding algorithm (ISTA), resulting in an efficient and interpretable deep network. The inputs to our network are the subsampled beamformed signals after summation and delay in the frequency domain, requiring only a subset of the US signal to be stored for recovery. Our method allows reducing the number of array elements, sampling rate, and computational time while ensuring high-quality imaging performance. Using in vivo data, we demonstrate that the proposed method yields high-quality images while reducing the data volume traditionally used up to 36 times. In terms of image resolution and contrast, our technique outperforms previously suggested methods as well as DAS and minimum-variance (MV) beamforming, paving the way to real-time applicable recovery methods.

Use of Intertransmission Coherence for Haze Artifact Suppression in Cardiovascular Synthetic Aperture Ultrasound Imaging

Synthetic aperture (SA) beamforming is a principal technology of modern medical ultrasound imaging. In that the use of focused transmission provides superior signal-to-noise ratio (SNR) and is promising for cardiovascular diagnosis at the maximum imaging depth of about 160 mm. But there is a pitfall in increasing the frame rate to more than 80 frames per second (frames/s) without image degradation by the haze artifact produced when the transmit foci (SA virtual sources) placed within the imaging field. We hypothesize that the source of this artifact is a grating lobe caused by coarse (decimated) multiple transmission and manifesting in the low brightness region in the accelerated-frame-rate images. We propose an intertransmission coherence factor (ITCF) method suppressing haze artifacts caused by coarse-pitch multiple transmission. The method is expected to suppress the image blurring because the SA grating lobe signal is less coherent than the main lobe signals. We evaluated an ITCF algorithm for suppressing the grating artifact when the transmission pitch is up to four times larger than the normal pitch (equivalent to 160 frames/s). In in-vitro and in-vivo experiments, we demonstrated the strong relation of haze artifact with the grating lobe due to the coarse-pitch transmission. Then, we confirmed that the ITCF method suppresses the haze artifact of a human heart by 15 dB while preserving the spatial resolution. The ITCF method combined with focused transmission SA beamforming is a valid method for getting cardiovascular ultrasound B-mode images without making a compromise in the trade-off relationship between the frame rate and the SNR.

Coherent Plane Wave Compounding Combined With Tensor Completion Applied for Ultrafast Imaging

To solve the problem of resolution and contrast in plane wave imaging (PWI), coherent plane wave compounding (CPWC) was introduced, in which scanning was performed at different angles, which can achieve the desired image quality by combining the images obtained from PWI at different angles. However, the application of this idea reduces the frame rate in proportion to the number of plane waves (PWs) or angles, so that in this modality, when dealing with some applications such as shear wave imaging (SWI) and strain imaging, there is always a compromise between the frame rate and the image quality. Tensor completion (TC) is a powerful technique to recover missing information of a low-rank tensor from limited observations based on rank minimization. In this article, we present an idea based on TC to make this compromise lighter; in other words, with a smaller number of angles, we can achieve the desired quality of the output image. To evaluate the proposed idea, plane wave imaging challenge in medical ultrasound (PICMUS) datasets was used, which were recorded at 75 different angles. The results of the resolution evaluation showed that using 20% of the coherent PWs and reconstructing other 80% by TC, compared with the situation of using only 20% of the coherent PWs provided a resolution improvement of 14.97% and 17.4% in the simulated and experimental point targets, respectively. Also, the results of the contrast investigation showed that the contrast ratio (CR) improved by 72.6%, 62.9%, and 111.4% in the simulated cyst target data, experimental cyst targets, and in vivo carotid cross section, respectively. The results confirmed that using 20% of the coherent PWs and reconstructing other 80% by TC, the image quality is very close to that obtained by considering all 75 angles, so that the difference in resolution is less than 2% and the difference in contrast to noise ratio (CNR) is less than 5 dB. Therefore, with this idea, it can be said that less compromise is needed; in other words, despite having a higher frame rate, an acceptable quality can be achieved.

Self-supervised learning for accelerated 3D high-resolution ultrasound imaging

PURPOSE

Ultrasound (US) imaging has been widely used in diagnosis, image-guided intervention, and therapy, where high-quality three-dimensional (3D) images are highly desired from sparsely acquired two-dimensional (2D) images. This study aims to develop a deep learning-based algorithm to reconstruct high-resolution (HR) 3D US images only reliant on the acquired sparsely distributed 2D images.

METHODS

We propose a self-supervised learning framework using cycle-consistent generative adversarial network (cycleGAN), where two independent cycleGAN models are trained with paired original US images and two sets of low-resolution (LR) US images, respectively. The two sets of LR US images are obtained through down-sampling the original US images along the two axes, respectively. In US imaging, in-plane spatial resolution is generally much higher than through-plane resolution. By learning the mapping from down-sampled in-plane LR images to original HR US images, cycleGAN can generate through-plane HR images from original sparely distributed 2D images. Finally, HR 3D US images are reconstructed by combining the generated 2D images from the two cycleGAN models.

RESULTS

The proposed method was assessed on two different datasets. One is automatic breast ultrasound (ABUS) images from 70 breast cancer patients, the other is collected from 45 prostate cancer patients. By applying a spatial resolution enhancement factor of 3 to the breast cases, our proposed method achieved the mean absolute error (MAE) value of 0.90 ± 0.15, the peak signal-to-noise ratio (PSNR) value of 37.88 ± 0.88 dB, and the visual information fidelity (VIF) value of 0.69 ± 0.01, which significantly outperforms bicubic interpolation. Similar performances have been achieved using the enhancement factor of 5 in these breast cases and using the enhancement factors of 5 and 10 in the prostate cases.

CONCLUSIONS

We have proposed and investigated a new deep learning-based algorithm for reconstructing HR 3D US images from sparely acquired 2D images. Significant improvement on through-plane resolution has been achieved by only using the acquired 2D images without any external atlas images. Its self-supervision capability could accelerate HR US imaging.

Sparse Convolutional Beamforming for 3-D Ultrafast Ultrasound Imaging

Real-time 3-D ultrasound (US) provides a complete visualization of inner body organs and blood vasculature, crucial for diagnosis and treatment of diverse diseases. However, 3-D systems require massive hardware due to the huge number of transducer elements and consequent data size. This increases cost significantly and limit both frame rate and image quality, thus preventing the 3-D US from being common practice in clinics worldwide. A recent study presented a technique called sparse convolutional beamforming algorithm (SCOBA), which obtains improved image quality while allowing notable element reduction in the context of 2-D focused imaging. In this article, we build upon previous work and introduce a nonlinear beamformer for 3-D imaging, called COBA-3D, consisting of 2-D spatial convolution of the in-phase and quadrature received signals. The proposed technique considers diverging-wave transmission and achieves improved image resolution and contrast compared with standard delay-and-sum beamforming while enabling a high frame rate. Incorporating 2-D sparse arrays into our method creates SCOBA-3D: a sparse beamformer that offers significant element reduction and, thus, allows performing 3-D imaging with the resources typically available for 2-D setups. To create 2-D thinned arrays, we present a scalable and systematic way to design 2-D fractal sparse arrays. The proposed framework paves the way for affordable ultrafast US devices that perform high-quality 3-D imaging, as demonstrated using phantom and ex-vivo data.

Variational Formulation of Unsupervised Deep Learning for Ultrasound Image Artifact Removal

Recently, deep learning approaches have been successfully used for ultrasound (US) image artifact removal. However, paired high-quality images for supervised training are difficult to obtain in many practical situations. Inspired by the recent theory of unsupervised learning using optimal transport driven CycleGAN (OT-CycleGAN), here, we investigate the applicability of unsupervised deep learning for US artifact removal problems without matched reference data. Two types of OT-CycleGAN approaches are employed: one with the partial knowledge of the image degradation physics and the other with the lack of such knowledge. Various US artifact removal problems are then addressed using the two types of OT-CycleGAN. Experimental results for various unsupervised US artifact removal tasks confirmed that our unsupervised learning method delivers results comparable to supervised learning in many practical applications.

3D synthetic aperture imaging with a therapeutic spherical random phased array for transcostal applications

Experimental validation of a synthetic aperture imaging technique using a therapeutic random phased array is described, demonstrating the dual nature of imaging and therapy of such an array. The transducer is capable of generating both continuous wave high intensity beams for ablating the tumor and low intensity ultrasound pulses to image the target area. Pulse-echo data is collected from the elements of the phased array to obtain B-mode images of the targets. Since therapeutic arrays are optimized for therapy only with concave apertures having low f-number and large directive elements often coarsely sampled, imaging can not be performed using conventional beamforming. We show that synthetic aperture imaging is capable of processing the acquired RF data to obtain images of the field of interest. Simulations were performed to compare different synthetic aperture imaging techniques to identify the best algorithm in terms of spatial resolution. Experimental validation was performed using a 1 MHz, 256-elements, spherical random phased array with 130 mm radius of curvature. The array was integrated with a research ultrasound scanner via custom connectors to acquire raw RF data for variety of targets. Imaging was implemented using synthetic aperture beamforming to produce images of a rib phantom and ex vivo ribs. The array was shown to resolve spherical targets within ±15 mm of either side of the axis in the focal plane and obtain 3D images of the rib phantom up to ±40 mm of either side of the central axis and at a depth of 3-9 cm from the array surface. The lateral and axial full width half maximum was 1.15 mm and 2.75 mm, respectively. This study was undertaken to emphasize that both therapy and image guidance with a therapeutic random phased array is possible and such a system has the potential to address some major limitations in the existing high intensity focused ultrasound (HIFU) systems. The 3D images obtained with a therapeutic array can be used to identify and locate strong scattering objects aiding to image guidance and treatment planning of the HIFU procedure.

Advances in ultrasonography: image formation and quality assessment

Delay-and-sum (DAS) beamforming is widely used for generation of B-mode images from echo signals obtained with an array probe composed of transducer elements. However, the resolution and contrast achieved with DAS beamforming are determined by the physical specifications of the array, e.g., size and pitch of elements. To overcome this limitation, adaptive imaging methods have recently been explored extensively thanks to the dissemination of digital and programmable ultrasound systems. On the other hand, it is also important to evaluate the performance of such adaptive imaging methods quantitatively to validate whether the modification of the image characteristics resulting from the developed method is appropriate. Since many adaptive imaging methods have been developed and they often alter image characteristics, attempts have also been made to update the methods for quantitative assessment of image quality. This article provides a review of recent developments in adaptive imaging and image quality assessment.

Learning Sub-Sampling and Signal Recovery With Applications in Ultrasound Imaging

Limitations on bandwidth and power consumption impose strict bounds on data rates of diagnostic imaging systems. Consequently, the design of suitable (i.e. task- and data-aware) compression and reconstruction techniques has attracted considerable attention in recent years. Compressed sensing emerged as a popular framework for sparse signal reconstruction from a small set of compressed measurements. However, typical compressed sensing designs measure a (non)linearly weighted combination of all input signal elements, which poses practical challenges. These designs are also not necessarily task-optimal. In addition, real-time recovery is hampered by the iterative and time-consuming nature of sparse recovery algorithms. Recently, deep learning methods have shown promise for fast recovery from compressed measurements, but the design of adequate and practical sensing strategies remains a challenge. Here, we propose a deep learning solution termed Deep Probabilistic Sub-sampling (DPS), that enables joint optimization of a task-adaptive sub-sampling pattern and a subsequent neural task model in an end-to-end fashion. Once learned, the task-based sub-sampling patterns are fixed and straightforwardly implementable, e.g. by non-uniform analog-to-digital conversion, sparse array design, or slow-time ultrasound pulsing schemes. The effectiveness of our framework is demonstrated in-silico for sparse signal recovery from partial Fourier measurements, and in-vivo for both anatomical image and tissue-motion (Doppler) reconstruction from sub-sampled medical ultrasound imaging data.

Adaptive Ultrasound Beamforming Using Deep Learning

Biomedical imaging is unequivocally dependent on the ability to reconstruct interpretable and high-quality images from acquired sensor data. This reconstruction process is pivotal across many applications, spanning from magnetic resonance imaging to ultrasound imaging. While advanced data-adaptive reconstruction methods can recover much higher image quality than traditional approaches, their implementation often poses a high computational burden. In ultrasound imaging, this burden is significant, especially when striving for low-cost systems, and has motivated the development of high-resolution and high-contrast adaptive beamforming methods. Here we show that deep neural networks, that adopt the algorithmic structure and constraints of adaptive signal processing techniques, can efficiently learn to perform fast high-quality ultrasound beamforming using very little training data. We apply our technique to two distinct ultrasound acquisition strategies (plane wave, and synthetic aperture), and demonstrate that high image quality can be maintained when measuring at low data-rates, using undersampled array designs. Beyond biomedical imaging, we expect that the proposed deep learning based adaptive processing framework can benefit a variety of array and signal processing applications, in particular when data-efficiency and robustness are of importance.

CohereNet: A Deep Learning Architecture for Ultrasound Spatial Correlation Estimation and Coherence-Based Beamforming

Deep fully connected networks are often considered "universal approximators" that are capable of learning any function. In this article, we utilize this particular property of deep neural networks (DNNs) to estimate normalized cross correlation as a function of spatial lag (i.e., spatial coherence functions) for applications in coherence-based beamforming, specifically short-lag spatial coherence (SLSC) beamforming. We detail the composition, assess the performance, and evaluate the computational efficiency of CohereNet, our custom fully connected DNN, which was trained to estimate the spatial coherence functions of in vivo breast data from 18 unique patients. CohereNet performance was evaluated on in vivo breast data from three additional patients who were not included during training, as well as data from in vivo liver and tissue mimicking phantoms scanned with a variety of ultrasound transducer array geometries and two different ultrasound systems. The mean correlation between the SLSC images computed on a central processing unit (CPU) and the corresponding DNN SLSC images created with CohereNet was 0.93 across the entire test set. The DNN SLSC approach was up to 3.4 times faster than the CPU SLSC approach, with similar computational speed, less variability in computational times, and improved image quality compared with a graphical processing unit (GPU)-based SLSC approach. These results are promising for the application of deep learning to estimate correlation functions derived from ultrasound data in multiple areas of ultrasound imaging and beamforming (e.g., speckle tracking, elastography, and blood flow estimation), possibly replacing GPU-based approaches in low-power, remote, and synchronization-dependent applications.

Subsampling Approaches for Compressed Sensing with Ultrasound Arrays in Non-Destructive Testing

Full Matrix Capture is a multi-channel data acquisition method which enables flexible, high resolution imaging using ultrasound arrays. However, the measurement time and data volume are increased considerably. Both of these costs can be circumvented via compressed sensing, which exploits prior knowledge of the underlying model and its sparsity to reduce the amount of data needed to produce a high resolution image. In order to design compression matrices that are physically realizable without sophisticated hardware constraints, structured subsampling patterns are designed and evaluated in this work. The design is based on the analysis of the Cramér–Rao Bound of a single scatterer in a homogeneous, isotropic medium. A numerical comparison of the point spread functions obtained with different compression matrices and the Fast Iterative Shrinkage/Thresholding Algorithm shows that the best performance is achieved when each transmit event can use a different subset of receiving elements and each receiving element uses a different section of the echo signal spectrum. Such a design has the advantage of outperforming other structured patterns to the extent that suboptimal selection matrices provide a good performance and can be efficiently computed with greedy approaches.

Compressed Sensing Approach for Reducing the Number of Receive Elements in Synthetic Transmit Aperture Imaging

Recently, researchers have shown an increased interest in ultrasound imaging methods alternate to conventional focused beamforming (CFB). One such approach is based on the synthetic aperture (SA) scheme; more popular are the ones based on synthetic transmit aperture (STA) schemes with a single-element transmit or multielement STA (MSTA). However, one of the main challenges in translating such methods to low-cost ultrasound systems is the tradeoffs among image quality, frame rate, and complexity of the system. These schemes use all the transducer elements during receive, which dictates a corresponding number of parallel receive channels, thus increasing the complexity of the system. A considerable amount of literature has been published on compressed sensing (CS) for SA imaging. Such studies are aimed at reducing the number of transmissions in SA but still recover images of acceptable quality at high frame rate and fail to address the complexity due to full-aperture receive. In this work, we adopt a CS framework to MSTA, with a motivation to reduce the number of receive elements and data. The CS recovery performance was assessed for the simulation data, tissue-mimicking phantom data, and an example in vivo biceps data. It was found that in spite of using 50% receive elements and overall using only 12.5% of the data, the images recovered using CS were comparable to those of reference full-aperture case in terms of estimated lateral resolution, contrast-to-noise ratio, and structural similarity indices. Thus, the proposed CS framework provides some fresh insights into translating the MSTA imaging method to affordable ultrasound scanners.

Super-resolution photoacoustic and ultrasound imaging with sparse arrays

It has previously been demonstrated that model-based reconstruction methods relying on a priori knowledge of the imaging point spread function (PSF) coupled to sparsity priors on the object to image can provide super-resolution in photoacoustic (PA) or in ultrasound (US) imaging. Here, we experimentally show that such reconstruction also leads to super-resolution in both PA and US imaging with arrays having much less elements than used conventionally (sparse arrays). As a proof of concept, we obtained super-resolution PA and US cross-sectional images of microfluidic channels with only 8 elements of a 128-elements linear array using a reconstruction approach based on a linear propagation forward model and assuming sparsity of the imaged structure. Although the microchannels appear indistinguishable in the conventional delay-and-sum images obtained with all the 128 transducer elements, the applied sparsity-constrained model-based reconstruction provides super-resolution with down to only 8 elements. We also report simulation results showing that the minimal number of transducer elements required to obtain a correct reconstruction is fundamentally limited by the signal-to-noise ratio. The proposed method can be straigthforwardly applied to any transducer geometry, including 2D sparse arrays for 3D super-resolution PA and US imaging.

Strategic Undersampling and Recovery Using Compressed Sensing for Enhancing Ultrasound Image Quality

In conventional focused beamforming (CFB), there is a known tradeoff between the active aperture size of the ultrasound transducer array and the resulting image quality. Increasing the size of the active aperture leads to an increase in the image quality of the ultrasound system at the expense of increased system cost. An alternate approach is to get rid of the requirement of having consecutive active receive elements and instead place them in a random order in a larger aperture. This, in turn, creates an undersampled situation where there are only M active elements placed in a larger aperture, which can accommodate N consecutive receive elements (with ). It is possible to formulate and solve the above-mentioned undersampling situation using a compressed sensing (CS) approach. In our previous work, we had proposed Gaussian undersampling strategy for reducing the number of active receive elements. In this work, we introduce a novel framework, namely Gaussian undersampling-based CS framework (GAUCS) with wave atoms as a sparsifying basis for CFB imaging method. The performance of the proposed method is validated using simulation and in vitro phantom data. Without an increase in the active elements, it is found that the proposed GAUCS framework improved the lateral resolution (LR) and image contrast by 27% and 1.5 times, respectively, while using 16 active elements and by 39% and 1.1 times, respectively, while using 32 active elements. Thus, the GAUCS framework can play a significant role in improving the performance, especially, of affordable point-of-care ultrasound systems.

A dynamic generalized coherence factor for side lobe suppression in ultrasound imaging

Coherence-based weighting techniques have been widely studied to weight beamsummed data to improve image quality in ultrasound imaging. Although generalized coherence factor (GCF) enhances the robustness of coherence factor (CF) with preserved speckle pattern by including some incoherent components, the side lobe suppression performance is insufficient due to constant cut-off frequency M₀. To address this problem, we introduced in this paper a dynamic GCF method, referred to as DGCF-C, based on the amplitude standard deviation and the convolution output of aperture data. The cut-off frequency is adaptively selected for GCF at each imaging point using the amplitude standard deviation of aperture data. Moreover, the convolution output of aperture data is used to calculate the dynamic GCF. The proposed method is evaluated in simulation and tissue-mimicking phantom studies. The image quality was analyzed in terms of resolution, contrast ratio (CR), generalized contrast-to-noise ratio (GCNR), speckle signal-to-noise ratio (sSNR), and signal-to-noise ratio (SNR). The results demonstrate that DGCF-C (M_max=2) achieves mean resolution improvements of 35.1% in simulation, and 32.6% in experiment, compared with GCF (M₀=1). Moreover, DGCF-C (M_max=4) outperforms GCF (M₀=2) with an average GCNR improvement of 13.5% and an average sSNR improvement of 15.2%, which indicates the better-preservation of speckle.

High-Resolution Fast Ultrasound Imaging With Adaptive-Lag Filtered Delay-Multiply-and-Sum Beamforming and Multiline Acquisition

Multiline acquisition (MLA) is a well-established method for a high-frame-rate cardiac ultrasound imaging, which is commonly used in conjunction with delay-and-sum (DAS) beamforming. The block-like artifacts that occur secondary to the use of MLA can be reduced using interpolation of the data acquired from adjacent transmitted beams-a method called synthetic transmit beamforming (STB). A recently proposed filtered delay-multiply-and-sum (F-DMAS) is a novel beamforming method, based on modified autocorrelation of the aperture data, which provides superior contrast resolution compared to the DAS beamforming. In this study, we demonstrate that a combination of the F-DMAS with the STB compensated MLA results in superior contrast as compared to both DAS beamformed STB and DAS beamformed single-line acquisition. Moreover, we propose a novel formulation for adaptive-lag F-DMAS that outperforms both DAS and F-DMAS in terms of contrast and lateral resolutions. The results are demonstrated in tissue-mimicking phantom and in human cardiac data.