Deep Learning in Ultrasound Localization Microscopy: Applications and Perspectives

Ultrasound localization microscopy (ULM) is a novel super-resolution imaging technique that can image the vasculature in vivo at depth with resolution far beyond the conventional limit of diffraction. By relying on the localization and tracking of clinically approved microbubbles injected in the blood stream, ULM can provide not only anatomical visualization but also hemodynamic quantification of the microvasculature. Several deep learning approaches have been proposed to address challenges in ULM including denoising, improving microbubble localization, estimating blood flow velocity, or performing aberration correction. Proposed deep learning methods often outperform their conventional counterparts by improving image quality and reducing processing time. In addition, their robustness to high concentrations of microbubbles can lead to reduced acquisition times in ULM, addressing a major hindrance to ULM clinical application. Herein, we propose a comprehensive review of the diversity of deep learning applications in ULM focusing on approaches assuming a sparse microbubble distribution. We first provide an overview of how existing studies vary in the constitution of their datasets or in the tasks targeted by the deep learning model. We also take a deeper look into the numerous approaches that have been proposed to improve the localization of microbubbles since they differ highly in their formulation of the optimization problem, their evaluation, or their network architectures. We finally discuss the current limitations and challenges of these methods, as well as the promises and potential of deep learning for ULM in the future.

Mitigating Aberration-Induced Noise: A Deep Learning-Based Aberration-to- Aberration Approach

One of the primary sources of suboptimal image quality in ultrasound imaging is phase aberration. It is caused by spatial changes in sound speed over a heterogeneous medium, which disturbs the transmitted waves and prevents coherent summation of echo signals. Obtaining non-aberrated ground truths in real-world scenarios can be extremely challenging, if not impossible. This challenge hinders the performance of deep learning-based techniques due to the domain shift between simulated and experimental data. Here, for the first time, we propose a deep learning-based method that does not require ground truth to correct the phase aberration problem and, as such, can be directly trained on real data. We train a network wherein both the input and target output are randomly aberrated radio frequency (RF) data. Moreover, we demonstrate that a conventional loss function such as mean square error is inadequate for training such a network to achieve optimal performance. Instead, we propose an adaptive mixed loss function that employs both B-mode and RF data, resulting in more efficient convergence and enhanced performance. Finally, we publicly release our dataset, comprising over 180,000 aberrated single plane-wave images (RF data), wherein phase aberrations are modeled as near-field phase screens. Although not utilized in the proposed method, each aberrated image is paired with its corresponding aberration profile and the non-aberrated version, aiming to mitigate the data scarcity problem in developing deep learning-based techniques for phase aberration correction. Source code and trained model are also available along with the dataset at https://code.sonography.ai/main-aaa.

Spatial Ambiguity Correction in Coherence-Based Average Sound Speed Estimation

Sound speed estimation can potentially correct the focusing errors in medical ultrasound. Maximizing the echo spatial coherence as a function of beamforming sound speed is a known technique to estimate the average sound speed. However, beamformation with changing sound speed causes a spatial shift of the echo signals resulting in noise and registration errors in the average sound speed estimates. We show that the spatial shift can be predicted and corrected, leading to superior sound speed estimates. Methods are presented for axial and 2-D location correction. Methods were evaluated using simulations and experimental phantom data. The location correction strategies improved the variance of sound speed estimates and reduced artifacts in the presence of strong backscatter variations. Limitations of the proposed methods and potential improvement strategies were evaluated.

Learning-based sound speed estimation and aberration correction for linear-array photoacoustic imaging

Photoacoustic (PA) image reconstruction involves acoustic inversion that necessitates the specification of the speed of sound (SoS) within the medium of propagation. Due to the lack of information on the spatial distribution of the SoS within heterogeneous soft tissue, a homogeneous SoS distribution (such as 1540 m/s) is typically assumed in PA image reconstruction, similar to that of ultrasound (US) imaging. Failure to compensate for the SoS variations leads to aberration artefacts, deteriorating the image quality. Various methods have been proposed to address this issue, but they usually involve complex hardware and/or time-consuming algorithms, hindering clinical translation. In this work, we introduce a deep learning framework for SoS estimation and subsequent aberration correction in a dual-modal PA/US imaging system exploiting a clinical US probe. As the acquired PA and US images were inherently co-registered, the estimated SoS distribution from US channel data using a deep neural network was incorporated for accurate PA image reconstruction. The framework comprised an initial pre-training stage based on digital phantoms, which was further enhanced through transfer learning using physical phantom data and associated SoS maps obtained from measurements. This framework achieved a root mean square error of 10.2 m/s and 15.2 m/s for SoS estimation on digital and physical phantoms, respectively and structural similarity index measures of up to 0.88 for PA reconstructions compared to the conventional approach of 0.69. A maximum of 1.2 times improvement in the signal-to-noise ratio of PA images was further demonstrated with a human volunteer study. Our results show that the proposed framework could be valuable in various clinical and preclinical applications to enhance PA image reconstruction.

Improved Transcranial Plane-Wave Imaging With Learned Speed-of-Sound Maps

Although transcranial ultrasound plane-wave imaging (PWI) has promising clinical application prospects, studies have shown that variable speed-of-sound (SoS) would seriously damage the quality of ultrasound images. The mismatch between the conventional constant velocity assumption and the actual SoS distribution leads to the general blurring of ultrasound images. The optimization scheme for reconstructing transcranial ultrasound image is often solved using iterative methods like full-waveform inversion. These iterative methods are computationally expensive and based on prior magnetic resonance imaging (MRI) or computed tomography (CT) information. In contrast, the multi-stencils fast marching (MSFM) method can produce accurate time travel maps for the skull with heterogeneous acoustic speed. In this study, we first propose a convolutional neural network (CNN) to predict SoS maps of the skull from PWI channel data. Then, use these maps to correct the travel time to reduce transcranial aberration. To validate the performance of the proposed method, numerical, phantom and intact human skull studies were conducted using a linear array transducer (L11-5v, 128 elements, pitch = 0.3 mm). Numerical simulations demonstrate that for point targets, the lateral resolution of MSFM-restored images increased by 65%, and the center position shift decreased by 89%. For the cyst targets, the eccentricity of the fitting ellipse decreased by 75%, and the center position shift decreased by 58%. In the phantom study, the lateral resolution of MSFM-restored images was increased by 49%, and the position shift was reduced by 1.72 mm. This pipeline, termed AutoSoS, thus shows the potential to correct distortions in real-time transcranial ultrasound imaging, as demonstrated by experiments on the intact human skull.

A New Method of Plane-Wave Ultrasound Imaging Based on Reverse Time Migration

Coherent plane-wave compounding technique enables rapid ultrasound imaging with comparable image quality to traditional B-mode imaging that relies on focused beam transmission. However, existing methods assume homogeneity in the imaged medium, neglecting the heterogeneity in sound velocities and densities present in real tissues, resulting in noise reverberation. This study introduces the Reverse Time Migration (RTM) method for ultrasound plane-wave imaging to overcome this limitation, which is combined with a method for estimating the speed of sound in layered media. Simulation results in a homogeneous background demonstrate that RTM reduces side lobes and grating lobes by approximately 30 dB, enhancing the contrast-to-noise ratio by 20% compared to conventional delay and sum (DAS) beamforming. Moreover, RTM achieves superior imaging outcomes with fewer compounding angles. The lateral resolution of the RTM with 5-9 angle compounding is able to achieve the effectiveness of the DAS method with 15-19 angle compounding, and the CNR of the RTM with 11-angle compounding is almost the same as that of the DAS with 21-angle compounding. In a heterogeneous background, experimental simulations and in vitro wire phantom experiments confirm RTM's capability to correct depth imaging, focusing reflected waves on point targets. In vitro porcine tissue experiments enable accurate imaging of layer interfaces by estimating the velocities of multiple layers containing muscle and fat. The proposed imaging procedure optimizes velocity estimation in complex media, compensates for the impact of velocity differences, provides more reliable imaging results.

Computing Speed-of-Sound From Ultrasound: User-Agnostic Recovery and a New Benchmark

OBJECTIVE

Medical ultrasound is one of the most accessible imaging modalities, but is a challenging modality for quantitative parameters comparison across vendors and sonographers. B-Mode imaging, with limited exceptions, provides a map of tissue boundaries; crucially, it does not provide diagnostically relevant physical quantities of the interior of organ domains.This can be remedied: the raw ultrasound signal carries significantly more information than is present in the B-Mode image. Specifically, the ability to recover speed-of-sound and attenuation maps from the raw ultrasound signal transforms the modality into a tissue-property modality. Deep learning was shown to be a viable tool for recovering speed-of-sound maps. A major hold-back towards deployment is the domain transfer problem, i.e., generalizing from simulations to real data. This is due in part to dependence on the (hard-to-calibrate) system response.

METHODS

We explore a remedy to the problem of operator-dependent effects on the system response by introducing a novel approach utilizing the phase information of the IQ demodulated signal.

RESULTS

We show that the IQ-phase information effectively decouples the operator-dependent system response from the data, significantly improving the stability of speed-of-sound recovery. We also introduce an improvement to the network topology providing faster and improved results to the state-of-the-art. We present the first publicly available benchmark for this problem: a simulated dataset for raw ultrasound plane wave processing.

CONCLUSION

The consideration of the phase of the IQ-signals presents a promising appeal to traversing the transfer learning problem, advancing the goal of real-time speed-of-sound imaging.

Artificial Intelligence-Enhanced Quantitative Ultrasound for Breast Cancer: Pilot Study on Quantitative Parameters and Biopsy Outcomes

Traditional B-mode ultrasound has difficulties distinguishing benign from malignant breast lesions. It appears that Quantitative Ultrasound (QUS) may offer advantages. We examined the QUS imaging system's potential, utilizing parameters like Attenuation Coefficient (AC), Speed of Sound (SoS), Effective Scatterer Diameter (ESD), and Effective Scatterer Concentration (ESC) to enhance diagnostic accuracy. B-mode images and radiofrequency signals were gathered from breast lesions. These parameters were processed and analyzed by a QUS system trained on a simulated acoustic dataset and equipped with an encoder-decoder structure. Fifty-seven patients were enrolled over six months. Biopsies served as the diagnostic ground truth. AC, SoS, and ESD showed significant differences between benign and malignant lesions (p < 0.05), but ESC did not. A logistic regression model was developed, demonstrating an area under the receiver operating characteristic curve of 0.90 (95% CI: 0.78, 0.96) for distinguishing between benign and malignant lesions. In conclusion, the QUS system shows promise in enhancing diagnostic accuracy by leveraging AC, SoS, and ESD. Further studies are needed to validate these findings and optimize the system for clinical use.

Investigating pulse-echo sound speed estimation in breast ultrasound with deep learning

Ultrasound is an adjunct tool to mammography that can quickly and safely aid physicians in diagnosing breast abnormalities. Clinical ultrasound often assumes a constant sound speed to form diagnostic B-mode images. However, the components of breast tissue, such as glandular tissue, fat, and lesions, differ in sound speed. Given a constant sound speed assumption, these differences can degrade the quality of reconstructed images via phase aberration. Sound speed images can be a powerful tool for improving image quality and identifying diseases if properly estimated. To this end, we propose a supervised deep-learning approach for sound speed estimation from analytic ultrasound signals. We develop a large-scale simulated ultrasound dataset that generates representative breast tissue samples by modeling breast gland, skin, and lesions with varying echogenicity and sound speed. We adopt a fully convolutional neural network architecture trained on a simulated dataset to produce an estimated sound speed map. The simulated tissue is interrogated with a plane wave transmit sequence, and the complex-value reconstructed images are used as input for the convolutional network. The network is trained on the sound speed distribution map of the simulated data, and the trained model can estimate sound speed given reconstructed pulse-echo signals. We further incorporate thermal noise augmentation during training to enhance model robustness to artifacts found in real ultrasound data. To highlight the ability of our model to provide accurate sound speed estimations, we evaluate it on simulated, phantom, and in-vivo breast ultrasound data.

Transcranial Phase Correction Using Pulse-Echo Ultrasound and Deep Learning: A 2-D Numerical Study

Phase aberration caused by human skulls severely degrades the quality of transcranial ultrasound images, posing a major challenge in the practical application of transcranial ultrasound techniques in adults. Aberration can be corrected if the skull profile (i.e., thickness distribution) and speed of sound (SOS) are known. However, accurately estimating the skull profile and SOS using ultrasound with a physics-based approach is challenging due to the complexity of the interaction between ultrasound and the skull. A deep learning approach is proposed herein to estimate the skull profile and SOS using ultrasound radiofrequency (RF) signals backscattered from the skull. A numerical study was performed to test the approach's feasibility. Realistic numerical skull models were constructed from computed tomography (CT) scans of five ex vivo human skulls in this numerical study. Acoustic simulations were performed on 3595 skull segments to generate array-based ultrasound backscattered signals. A deep learning model was developed and trained to estimate skull thickness and SOS from RF channel data. The trained model was shown to be highly accurate. The mean absolute error (MAE) was 0.15 mm (2% error) for thickness estimation and 13 m/s (0.5% error) for SOS estimation. The Pearson correlation coefficient between the estimated and ground-truth values was 0.99 for thickness and 0.95 for SOS. Aberration correction performed using deep-learning-estimated skull thickness and SOS values yielded significantly improved beam focusing (e.g., narrower beams) and transcranial imaging quality (e.g., improved spatial resolution and reduced artifacts) compared with no aberration correction. The results demonstrate the feasibility of the proposed approach for transcranial phase aberration correction.

Ultrasonic Sound Speed Estimation for Liver Fat Quantification: A Review by the AIUM-RSNA QIBA Pulse-Echo Quantitative Ultrasound Initiative

Non-alcoholic fatty liver disease (NAFLD) is a significant cause of diffuse liver disease, morbidity and mortality worldwide. Early and accurate diagnosis of NALFD is critical to identify patients at risk of disease progression. Liver biopsy is the current gold standard for diagnosis and prognosis. However, a non-invasive diagnostic tool is desired because of the high cost and risk of complications of tissue sampling. Medical ultrasound is a safe, inexpensive and widely available imaging tool for diagnosing NAFLD. Emerging sonographic tools to quantitatively estimate hepatic fat fraction, such as tissue sound speed estimation, are likely to improve diagnostic accuracy, precision and reproducibility compared with existing qualitative and semi-quantitative techniques. Various pulse-echo ultrasound speed of sound estimation methodologies have been investigated, and some have been recently commercialized. We review state-of-the-art in vivo speed of sound estimation techniques, including their advantages, limitations, technical sources of variability, biological confounders and existing commercial implementations. We report the expected range of hepatic speed of sound as a function of liver steatosis and fibrosis that may be encountered in clinical practice. Ongoing efforts seek to quantify sound speed measurement accuracy and precision to inform threshold development around meaningful differences in fat fraction and between sequential measurements.

Aberration correction in diagnostic ultrasound: A review of the prior field and current directions

Medical ultrasound images are reconstructed with simplifying assumptions on wave propagation, with one of the most prominent assumptions being that the imaging medium is composed of a constant sound speed. When the assumption of a constant sound speed are violated, which is true in most in vivoor clinical imaging scenarios, distortion of the transmitted and received ultrasound wavefronts appear and degrade the image quality. This distortion is known as aberration, and the techniques used to correct for the distortion are known as aberration correction techniques. Several models have been proposed to understand and correct for aberration. In this review paper, aberration and aberration correction are explored from the early models and correction techniques, including the near-field phase screen model and its associated correction techniques such as nearest-neighbor cross-correlation, to more recent models and correction techniques that incorporate spatially varying aberration and diffractive effects, such as models and techniques that rely on the estimation of the sound speed distribution in the imaging medium. In addition to historical models, future directions of ultrasound aberration correction are proposed.

Excluding Echo Shift Noise in Real-Time Pulse-Echo Speed-of-Sound Imaging

Computed ultrasound tomography in echo mode (CUTE) allows real-time imaging of the tissue speed of sound (SoS) using handheld ultrasound. The SoS is retrieved by inverting a forward model that relates the spatial distribution of the tissue SoS to echo shift maps detected between varying transmit and receive angles. Despite promising results, in vivo SoS maps often show artifacts due to elevated noise in echo shift maps. To minimize artifacts, we propose a technique where an individual SoS map is reconstructed for each echo shift map separately, as opposed to a single SoS map from all echo shift maps simultaneously. The final SoS map is then obtained as a weighted average over all SoS maps. Due to the partial redundancy between different angle combinations, artifacts that appear only in a subset of the individual maps can be excluded via the averaging weights. We investigate this real-time capable technique in simulations using two numerical phantoms, one with a circular inclusion and one with two layers. Our results demonstrate that the SoS maps reconstructed using the proposed technique are equivalent to the ones using simultaneous reconstruction when considering uncorrupted data but show significantly reduced artifact level for data that are corrupted by noise.

Ultrasonic imaging using conditional generative adversarial networks

The Full Matrix Capture (FMC) and Total Focusing Method (TFM) combination is often considered the gold standard in ultrasonic nondestructive testing, however it may be impractical due to the amount of time required to gather and process the FMC, particularly for high cadence inspections. This study proposes replacing conventional FMC acquisition and TFM processing with a single zero-degree plane wave (PW) insonification and a conditional Generative Adversarial Network (cGAN) trained to produce TFM-like images. Three models with different cGAN architectures and loss formulations were tested in different scenarios. Their performances were compared with conventional TFM computed from FMC. The proposed cGANs were able to recreate TFM-like images with the same resolution while improving the contrast in more than 94% of the reconstructions in comparison with conventional TFM reconstructions. Indeed, thanks to the use of a bias in the cGANs' training, the contrast was systematically increased through a reduction of the background noise level and the elimination of some artifacts. Finally, the proposed method led to a reduction of the computation time and file size by a factor of 120 and 75, respectively.

Deep learning for hetero-homo conversion in channel-domain for phase aberration correction in ultrasound imaging

Echo imaging in ultrasound computed tomography (USCT) using the synthetic aperture technique is performed with the assumption that the speed of sound is constant in the system. However, tissue heterogeneity causes a mismatch between the predicted arrival time and the actual arrival time of the echo signal, which will result in phase aberration, leading to the quality degradation of the reconstructed B-mode image. The conventional correction methods that use the correlation of each different channel require the presence of strong point scatterers and involve the problem of local solutions due to excessive correction. In this study, we propose a novel approach to correcting the signal distortion due to sound speed heterogeneity using a deep neural network (DNN). The DNN was trained to convert the distorted radio frequency (RF) inputs for the heterogeneous medium to the distortion-free RF outputs for the homogeneous medium. The network with U-net architecture using ResNet-34 as a backbone was trained using the hetero-homo corresponding channel-domain RF data generated via numerical simulations. The trained network performed phase aberration correction in the channel-domain RF, with the B-mode images reconstructed with the corrected RF demonstrating a higher contrast and an improved resolution compared with uncorrected cases. It was also demonstrated that the DNN model is robust to both varied reflection intensities and varied sound speed heterogeneities. The successful results demonstrated that the proposed DNN-based method is effective for phase aberration correction in US imaging.

A Single-Shot Harmonic Imaging Approach Utilizing Deep Learning for Medical Ultrasound

Tissue harmonic imaging (THI) is an invaluable tool in clinical ultrasound due to its enhanced contrast resolution and reduced reverberation clutter in comparison with fundamental mode imaging. However, harmonic content separation based on high-pass filtering suffers from potential contrast degradation or lower axial resolution due to spectral leakage, whereas nonlinear multipulse harmonic imaging schemes, such as amplitude modulation and pulse inversion, suffer from a reduced frame rate and comparatively higher motion artifacts due to the necessity of at least two pulse echo acquisitions. To address this problem, we propose a deep-learning-based single-shot harmonic imaging technique capable of generating comparable image quality to pulse amplitude modulation methods, yet at a higher frame rate and with fewer motion artifacts. Specifically, an asymmetric convolutional encoder-decoder structure is designed to estimate the combination of the echoes resulting from the half-amplitude transmissions using the echo produced from the full amplitude transmission as input. The echoes were acquired with the checkerboard amplitude modulation technique for training. The model was evaluated across various targets and samples to illustrate generalizability as well as the possibility and impact of transfer learning. Furthermore, for possible interpretability of the network, we investigate if the latent space of the encoder holds information on the nonlinearity parameter of the medium. We demonstrate the ability of the proposed approach to generate harmonic images with a single firing that are comparable to those from a multipulse acquisition.

DeepUCT: Complex cascaded deep learning network for improved ultrasound tomography

Ultrasound computed tomography is an inexpensive and radiation-free medical imaging technique used to quantify the tissue acoustic properties for advanced clinical diagnosis. Image reconstruction in ultrasound tomography is often modeled as an optimization scheme solved by iterative methods like full-waveform inversion. These iterative methods are computationally expensive, while the optimization problem is ill-posed and nonlinear. To address this problem, we propose to use deep learning to overcome the computational burden and ill-posedness, and achieve near real-time image reconstruction in ultrasound tomography. We aim to directly learn the mapping from the recorded time-series sensor data to a spatial image of acoustical properties. To accomplish this, we develop a deep learning model using two cascaded convolutional neural networks with an encoder–decoder architecture. We achieve a good representation by first extracting the intermediate mapping-knowledge and later utilizing this knowledge to reconstruct the image. This approach is evaluated on synthetic phantoms where simulated ultrasound data are acquired from a ring of transducers surrounding the region of interest. The measurement data is acquired by forward modeling the wave equation using the k-wave toolbox. Our simulation results demonstrate that our proposed deep-learning method is robust to noise and significantly outperforms the state-of-the-art traditional iterative method both quantitatively and qualitatively. Furthermore, our model takes substantially less computational time than the conventional full-wave inversion method.

2D speed of sound mapping using a multilook reflection ultrasound tomography framework

Quantitative ultrasound (QUS) has emerged as a viable tool in diagnosing and staging the onset and progression of various diseases. Within the field of QUS, shear wave elastography (SWE) has emerged as the clinical standard for quantifying and correlating the stiffness of tissue to its underlying pathology. Despite its widespread use, SWE suffers from drawbacks that limit its widespread clinical use; among these are low-frame rates, long settling times, and high sensitivity to operating conditions. Longitudinal speed of sound (SOS) has emerged as a viable alternative to SWE. We propose a framework to obtain 2D sound speed maps using a commercial ultrasound probe. A commercial ultrasound probe is localized in space and used to scan a domain of interest from multiple vantage points; the use of a reflector at the far end of the domain allows us to measure the round trip travel times to and from it. The known locations of the probe and the measured travel times are used to estimate the depth and inclination of the reflector as well as the unknown sound speed map. The use of multiple looks increases the effective aperture of the ultrasound probe and allows for a higher fidelity reconstruction of sound speed maps. We validate the framework using simulated and experimental data and propose a rigorous framework to quantify the uncertainty of the estimated sound speed maps.

Robust Single-Probe Quantitative Ultrasonic Imaging System with a Target-Aware Deep Neural Network

OBJECTIVE

The speed of sound (SoS) has great potential as a quantitative imaging biomarker since it is sensitive to pathological changes in tissues. In this paper, a target-aware deep neural (TAD) network reconstructing an SoS image quantitatively from pulse-echo phase-shift maps gathered from a single conventional ultrasound probe is presented.

METHODS

In the proposed TAD network, the reconstruction process is guided by feature maps created from segmented target images for accuracy and contrast. In addition, the feature extraction process utilizes phase difference information instead of direct pulse-echo radio frequency (RF) data for robust image reconstruction against noise in the pulse-echo data.

RESULTS

The TAD network outperforms the fully convolutional network in root mean square error (RMSE), contrast-to-noise ratio (CNR), and structural similarity index (SSIM) in the presence of nearby reflectors. The measured RMSE and CNR are 5.4 m/s and 22 dB, respectively with the tissue attenuation coefficient of 2 dB/cm/MHz, which are 72% and 13 dB improvement over the state of the art design in RMSE and CNR, respectively. In the in-vivo test, the proposed method classifies the tissues in the neck area using SoS with a p-value below 0.025.

CONCLUSION

The proposed TAD network is the most accurate and robust single-probe SoS image reconstruction method reported to date.

SIGNIFICANCE

The accuracy and robustness demonstrated by the proposed SoS imaging method open up the possibilities of wide-spread clinical application of the single-probe SoS imaging system.

Deep Learning in Kidney Ultrasound: Overview, Frontiers, and Challenges

Ultrasonography is a practical imaging technique used in numerous health care settings. It is relatively inexpensive, portable, and safe, and it has dynamic capabilities that make it an invaluable tool for a wide variety of diagnostic and interventional studies. Recently, there has been a revolution in medical imaging using artificial intelligence (AI). A particularly potent form of AI is deep learning, in which the computer learns to recognize pixel or written data on its own without the selection of predetermined features, usually through a specific neural network architecture. Neural networks vary in architecture depending on their task, and key design considerations include the number of layers and complexity, data available, technical requirements, and domain knowledge. Deep learning models offer the potential for promising innovations to workflow, image quality, and vision tasks in sonography. However, there are key limitations and challenges in creating reliable and safe AI models for patients and clinicians.

Localization of Scattering Objects Using Neural Networks

The localization of multiple scattering objects is performed while using scattered waves. An up-to-date approach: neural networks are used to estimate the corresponding locations. In the scattering phenomenon under investigation, we assume known incident plane waves, fully reflecting balls with known diameters and measurement data of the scattered wave on one fixed segment. The training data are constructed while using the simulation package μ-diff in Matlab. The structure of the neural networks, which are widely used for similar purposes, is further developed. A complex locally connected layer is the main compound of the proposed setup. With this and an appropriate preprocessing of the training data set, the number of parameters can be kept at a relatively low level. As a result, using a relatively large training data set, the unknown locations of the objects can be estimated effectively.

Training Variational Networks With Multidomain Simulations: Speed-of-Sound Image Reconstruction

Speed-of-sound (SoS) has been shown as a potential biomarker for breast cancer imaging, successfully differentiating malignant tumors from benign ones. SoS images can be reconstructed from time-of-flight measurements from ultrasound images acquired using conventional handheld ultrasound transducers. Variational networks (VNs) have recently been shown to be a potential learning-based approach for optimizing inverse problems in image reconstruction. Despite earlier promising results, these methods, however, do not generalize well from simulated to acquired data, due to the domain shift. In this work, we present for the first time a VN solution for a pulse-echo SoS image reconstruction problem using diverging waves with conventional transducers and single-sided tissue access. This is made possible by incorporating simulations with varying complexity into training. We use loop unrolling of gradient descent with momentum, with an exponentially weighted loss of outputs at each unrolled iteration in order to regularize the training. We learn norms as activation functions regularized to have smooth forms for robustness to input distribution variations. We evaluate reconstruction quality on the ray-based and full-wave simulations as well as on the tissue-mimicking phantom data, in comparison with a classical iterative [limited-memory Broyden-Fletcher-Goldfarb-Shanno (L-BFGS)] optimization of this image reconstruction problem. We show that the proposed regularization techniques combined with multisource domain training yield substantial improvements in the domain adaptation capabilities of VN, reducing the median root mean squared error (RMSE) by 54% on a wave-based simulation data set compared to the baseline VN. We also show that on data acquired from a tissue-mimicking breast phantom, the proposed VN provides improved reconstruction in 12 ms.

Technology trends and applications of deep learning in ultrasonography: image quality enhancement, diagnostic support, and improving workflow efficiency

In this review of the most recent applications of deep learning to ultrasound imaging, the architectures of deep learning networks are briefly explained for the medical imaging applications of classification, detection, segmentation, and generation. Ultrasonography applications for image processing and diagnosis are then reviewed and summarized, along with some representative imaging studies of the breast, thyroid, heart, kidney, liver, and fetal head. Efforts towards workflow enhancement are also reviewed, with an emphasis on view recognition, scanning guide, image quality assessment, and quantification and measurement. Finally some future prospects are presented regarding image quality enhancement, diagnostic support, and improvements in workflow efficiency, along with remarks on hurdles, benefits, and necessary collaborations.