Reverberation Noise Suppression in Ultrasound Channel Signals Using a 3D Fully Convolutional Neural Network

A domain adaptation model for carotid ultrasound: Image harmonization, noise reduction, and impact on cardiovascular risk markers

Deep learning has been used extensively for medical image analysis applications, assuming the training and test data adhere to the same probability distributions. However, a common challenge arises when dealing with medical images generated by different systems or even the same system with varying parameter settings. Such images often contain diverse textures and noise patterns, violating the assumption. Consequently, models trained on data from one machine or setting usually struggle to perform effectively on data from another. To address this issue in ultrasound images, we proposed a Generative Adversarial Network (GAN) based model in this paper. We formulated image harmonization and denoising tasks as an image-to-image translation task, wherein we adapt the texture pattern and reduced noise in Carotid ultrasound images while keeping the image content (the anatomy) unchanged. The performance was evaluated using feature distribution and pixel-space similarity metrics. In addition, blood-to-tissue contrast and influence on computed risk markers (Grey scale median, GSM) were evaluated. The results showed that domain adaptation was achieved in both tasks (histogram correlation 0.920 (0.043) and 0.844 (0.062)), as compared to no adaptation (0.890 (0.077) and 0.707 (0.098)), and that the anatomy of the images was retained (structure similarity index measure e.g. the arterial wall 0.71 (0.09) and 0.80 (0.08)). In addition, the image noise level (contrast) did not change in the image harmonization task (-34.1 (3.8) vs -35.2 (4.1) dB) but was improved in the noise reduction task (-23.5 (3.2) vs -46.7 (18.1) dB). To validate the performance of the proposed model, we compare its results with CycleGAN, the current state-of-the-art model. Our model outperformed CycleGAN in both tasks. Finally, the risk marker GSM was significantly changed in the noise reduction but not in the image harmonization task. We conclude that domain translation models are powerful tools for improving ultrasound image while retaining the underlying anatomy, but downstream calculations of risk markers may be affected.

A narrative review of deep learning in thyroid imaging: current progress and future prospects

Background and Objective

Deep learning (DL) has contributed substantially to the evolution of image analysis by unlocking increased data and computational power. These DL algorithms have further facilitated the growing trend of implementing precision medicine, particularly in areas of diagnosis and therapy. Thyroid imaging, as a routine means to screening for thyroid diseases on large-scale populations, is a massive data source for the development of DL models. Thyroid disease is a global health problem and involves structural and functional changes. The objective of this study was to evaluate the general rules and future directions of DL networks in thyroid medical image analysis through a review of original articles published between 2018 and 2023.

Methods

We searched for English-language articles published between April 2018 and September 2023 in the databases of PubMed, Web of Science, and Google Scholar. The keywords used in the search included artificial intelligence or DL, thyroid diseases, and thyroid nodule or thyroid carcinoma.

Key Content and Findings

The computer vision tasks of DL in thyroid imaging included classification, segmentation, and detection. The current applications of DL in clinical workflow were found to mainly include management of thyroid nodules/carcinoma, risk evaluation of thyroid cancer metastasis, and discrimination of functional thyroid diseases.

Conclusions

DL is expected to enhance the quality of thyroid images and provide greater precision in the assessment of thyroid images. Specifically, DL can increase the diagnostic accuracy of thyroid diseases and better inform clinical decision-making.

Review of Deep Learning Approaches for Interleaved Photoacoustic and Ultrasound (PAUS) Imaging

Photoacoustic (PA) imaging provides optical contrast at relatively large depths within the human body, compared to other optical methods, at ultrasound (US) spatial resolution. By integrating real-time PA and US (PAUS) modalities, PAUS imaging has the potential to become a routine clinical modality bringing the molecular sensitivity of optics to medical US imaging. For applications where the full capabilities of clinical US scanners must be maintained in PAUS, conventional limited view and bandwidth transducers must be used. This approach, however, cannot provide high-quality maps of PA sources, especially vascular structures. Deep learning (DL) using data-driven modeling with minimal human design has been very effective in medical imaging, medical data analysis, and disease diagnosis, and has the potential to overcome many of the technical limitations of current PAUS imaging systems. The primary purpose of this article is to summarize the background and current status of DL applications in PAUS imaging. It also looks beyond current approaches to identify remaining challenges and opportunities for robust translation of PAUS technologies to the clinic.

Ultrasound imaging of lung disease and its relationship to histopathology: An experimentally validated simulation approach

Lung ultrasound (LUS) is a widely used technique in clinical lung assessment, yet the relationship between LUS images and the underlying disease remains poorly understood due in part to the complexity of the wave propagation physics in complex tissue/air structures. Establishing a clear link between visual patterns in ultrasound images and underlying lung anatomy could improve the diagnostic accuracy and clinical deployment of LUS. Reverberation that occurs at the lung interface is complex, resulting in images that require interpretation of the artifacts deep in the lungs. These images are not accurate spatial representations of the anatomy due to the almost total reflectivity and high impedance mismatch between aerated lung and chest wall. Here, we develop an approach based on the first principles of wave propagation physics in highly realistic maps of the human chest wall and lung to unveil a relationship between lung disease, tissue structure, and its resulting effects on ultrasound images. It is shown that Fullwave numerical simulations of ultrasound propagation and histology-derived acoustical maps model the multiple scattering physics at the lung interface and reproduce LUS B-mode images that are comparable to clinical images. However, unlike clinical imaging, the underlying tissue structure model is known and controllable. The amount of fluid and connective tissue components in the lung were gradually modified to model disease progression, and the resulting changes in B-mode images and non-imaging reverberation measures were analyzed to explain the relationship between pathological modifications of lung tissue and observed LUS.

Optimal transmit apodization for the maximization of lag-one coherence with applications to aberration delay estimation

Phase aberration is one of the major sources of image degradation in medical ultrasound imaging. One of the earliest and simplest techniques to correct for phase aberration involves nearest-neighbor cross correlation to estimate delays between neighboring receive channels and the compensation of aberration delays in a delay-and-sum beamformer. The main challenge is that neighboring receive channels may not have sufficient signal correlation to accurately estimate the aberration delays. Although algorithms such as the translating transmit aperture or the common midpoint gather are designed to perfectly maximize signal correlations between received signals, these algorithms require the use of different transmit apertures for each received signal. Instead, this work proposes the use of a single globally-applicable transmit apodization function that optimizes the lag-one coherence based on the van Cittert-Zernike theorem. For the application to phase aberration correction, it is shown across 20 different zero-mean Gaussian-random aberrators that the proposed optimal apodization function reduces the estimation error in the aberration delay profile from 22.85% to 15.72%.

Deep Null Space Learning Improves Dataset Recovery for High Frame Rate Synthetic Transmit Aperture Imaging

Synthetic transmit aperture (STA) imaging benefits from the two-way dynamic focusing to achieve optimal lateral resolution and contrast resolution in the full field of view, at the cost of low frame rate (FR) and low signal-to-noise ratio (SNR). In our previous studies, compressed sensing-based STA (CS-STA) and minimal ${l}_{{2}}$ -norm least squares (LS-STA) methods were proposed to recover the complete STA dataset from fewer Hadamard-encoded (HE) plane wave (PW) transmissions. Results demonstrated that, compared with STA imaging, CS/LS-STA can maintain the high resolution of STA in the full field of view and improve the contrast in the deep region with increased FR. However, these methods would introduce errors to the recovered STA datasets and subsequently produce severe artifacts to the beamformed images, especially in the shallow region. Recently, we discovered that the theoretical explanation for the error introduced in the LS-STA-based recovery is that the LS-STA method neglects the null space component of the real STA dataset. To deal with this problem, we propose to train a convolutional neural network under the null space learning framework (CNN-Null) to estimate the missing null space component) for high-accuracy recovery of the STA dataset from fewer HE PW transmissions. The mapping between the low-quality STA dataset (i.e., the range space component of the real STA dataset recovered using the LS-STA method) and the missing null space component of the real STA dataset was learned by the network with the high-quality STA dataset (obtained using full HE STA (HE-STA) imaging) as training labels. The performance of the proposed CNN-Null method was compared with the baseline LS-STA, conventional STA, and HE-STA methods, in terms of the visual quality, the normalized root mean square error (NRMSE), the generalized contrast-to-noise ratio (gCNR), and the lateral full-width at half-maximum (FWHM). The results demonstrate that the proposed method can greatly improve the recovery accuracy of the STA datasets (lower NRMSE) and, therefore, effectively suppress the artifacts presented in the images (especially in the shallow region) obtained using the LS-STA method (with a gCNR improvement of 0.4 in the cross-sectional carotid artery images). In addition, the proposed method can maintain the high lateral resolution of STA with fewer (as low as 16) PW transmissions, as LS-STA does.

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.

Tunable image quality control of 3-D ultrasound using switchable CycleGAN

In contrast to 2-D ultrasound (US) for uniaxial plane imaging, a 3-D US imaging system can visualize a volume along three axial planes. This allows for a full view of the anatomy, which is useful for gynecological (GYN) and obstetrical (OB) applications. Unfortunately, the 3-D US has an inherent limitation in resolution compared to the 2-D US. In the case of 3-D US with a 3-D mechanical probe, for example, the image quality is comparable along the beam direction, but significant deterioration in image quality is often observed in the other two axial image planes. To address this, here we propose a novel unsupervised deep learning approach to improve 3-D US image quality. In particular, using unmatched high-quality 2-D US images as a reference, we trained a recently proposed switchable CycleGAN architecture so that every mapping plane in 3-D US can learn the image quality of 2-D US images. Thanks to the switchable architecture, our network can also provide real-time control of image enhancement level based on user preference, which is ideal for a user-centric scanner setup. Extensive experiments with clinical evaluation confirm that our method offers significantly improved image quality as well user-friendly flexibility.

Inverse Problem of Ultrasound Beamforming With Denoising-Based Regularized Solutions

During the past few years, inverse problem formulations of ultrasound beamforming have attracted growing interest. They usually pose beamforming as a minimization problem of a fidelity term resulting from the measurement model plus a regularization term that enforces a certain class on the resulting image. Here, we take advantage of alternating direction method of multipliers to propose a flexible framework in which each term is optimized separately. Furthermore, the proposed beamforming formulation is extended to replace the regularization term with a denoising algorithm, based on the recent approaches called plug-and-play (PnP) and regularization by denoising (RED). Such regularizations are shown in this work to better preserve speckle texture, an important feature in ultrasound imaging, than sparsity-based approaches previously proposed in the literature. The efficiency of the proposed methods is evaluated on simulations, real phantoms, and in vivo data available from a plane-wave imaging challenge in medical ultrasound. Furthermore, a comprehensive comparison with existing ultrasound beamforming methods is also provided. These results show that the RED algorithm gives the best image quality in terms of contrast index while preserving the speckle statistics.

Distributed Aberration Correction Techniques Based on Tomographic Sound Speed Estimates

Phase aberration is widely considered a major source of image degradation in medical pulse-echo ultrasound. Traditionally, near-field phase aberration correction techniques are unable to account for distributed aberrations due to a spatially varying speed of sound in the medium, while most distributed aberration correction techniques require the use of point-like sources and are impractical for clinical applications where diffuse scattering is dominant. Here, we present two distributed aberration correction techniques that utilize sound speed estimates from a tomographic sound speed estimator that builds on our previous work with diffuse scattering in layered media. We first characterize the performance of our sound speed estimator and distributed aberration correction techniques in simulations where the scattering in the media is known a priori. Phantom and in vivo experiments further demonstrate the capabilities of the sound speed estimator and the aberration correction techniques. In phantom experiments, point target resolution improves from 0.58 to 0.26 and 0.27 mm, and lesion contrast improves from 17.7 to 23.5 and 25.9 dB, as a result of distributed aberration correction using the eikonal and wavefield correlation techniques, respectively.

Disease-Specific Imaging Utilizing Support Vector Machine Classification of H-Scan Parameters: Assessment of Steatosis in a Rat Model

In medical imaging, quantitative measurements have shown promise in identifying diseases by classifying normal versus pathological parameters from tissues. The support vector machine (SVM) has shown promise as a supervised classification algorithm and has been widely used. However, the classification results typically identify a category of abnormal tissues but do not necessarily differentiate progressive stages of a disease. Moreover, the classification result is typically provided independently as a supplement to medical images, which contributes to an overload of information sources in the clinic. Hence, we propose a new imaging method utilizing the SVM to integrate classification results into medical images. This framework is called disease-specific imaging (DSI) that produces a color overlaid highlight on B-mode ultrasound images indicating the type, location, and severity of pathology from different conditions. In this article, the SVM training was performed to construct hyperplanes that can differentiate normal, fibrosis, steatosis, and pancreatic ductal adenocarcinoma (PDAC) metastases in livers based on ultrasound echoes. Also, cluster centroids for specific diseases define unique disease axes, and the inner product between measured features and any disease axis selected by the SVM quantifies the disease progression. The features were measured from 2794 ultrasound frames using the H-scan analysis, attenuation estimation, and B-mode image analysis. The performance of our proposed DSI method was evaluated for a preclinical model of steatosis ( n = 400 frames). The contribution of each feature was assessed, and the results were compared with ground truth from histology. Moreover, the images generated by our DSI were compared with earlier imaging methods of B-mode, H-scan, and histology. The comparisons demonstrate that DSI images yield higher sensitivity to monitor progressive steatosis than B-mode and H-scan and provide a comparable performance with the histology. For the parameter comparison, DSI and H-scan resulted in similar correlation with histology ( r_s = 0.83 ) but higher than attenuation ( r_s = 0.73 ) and B-mode ( r_s = 0.47 ). Therefore, we conclude that DSI utilizing the SVM applied to steatosis can visually represent the classification results with color highlighting, which can simplify the interpretation of classification compared to the traditional SVM result. We expect that the proposed DSI can be used for any medical imaging modality that can estimate multiple quantitative parameters at high resolution.

Noninvasive estimation of local speed of sound by pulse-echo ultrasound in a rat model of nonalcoholic fatty liver

Objective. Speed of sound has previously been demonstrated to correlate with fat concentration in the liver. However, estimating speed of sound in the liver noninvasively can be biased by the speed of sound of the tissue layers overlying the liver. Here, we demonstrate a noninvasive local speed of sound estimator, which is based on a layered media assumption, that can accurately capture the speed of sound in the liver. We validate the estimator using an obese Zucker rat model of non-alcoholic fatty liver disease and correlate the local speed of sound with liver steatosis.Approach.We estimated the local and global average speed of sound noninvasively in 4 lean Zucker rats fed a normal diet and 16 obese Zucker rats fed a high fat diet for up to 8 weeks. The ground truth speed of sound and fat concentration were measured from the excised liver using established techniques.Main Results. The noninvasive, local speed of sound estimates of the livers were similar in value to their corresponding 'ground truth' measurements, having a slope ± standard error of the regression of 0.82 ± 0.15 (R²= 0.74 andp< 0.001). Measurement of the noninvasive global average speed of sound did not reliably capture the 'ground truth' speed of sound in the liver, having a slope of 0.35 ± 0.07 (R²= 0.74 andp< 0.001). Decreasing local speed of sound was observed with increasing hepatic fat accumulation (approximately -1.7 m s^-1per 1% increase in hepatic fat) and histopathology steatosis grading (approximately -10 to -13 m s^-1per unit increase in steatosis grade). Local speed of sound estimates were highly correlated with steatosis grade, having Pearson and Spearman correlation coefficients both ranging from -0.87 to -0.78. In addition, a lobe-dependent speed of sound in the liver was observed by theex vivomeasurements, with speed of sound differences of up to 25 m s^-1(p< 0.003) observed between lobes in the liver of the same animal.Significance.The findings of this study suggest that local speed of sound estimation has the potential to be used to predict or assist in the measurement of hepatic fat concentration and that the global average speed of sound should be avoided in hepatic fat estimation due to significant bias in the speed of sound estimate.

Reverberation Clutter Suppression Using 2-D Spatial Coherence Analysis

Diffuse reverberation clutter often significantly degrades the visibility of abdominal structures. Reverberation clutter acts as a temporally stationary haze that originates from the multiple scattering within the subcutaneous layers and has a narrow spatial correlation length. We recently presented an adaptive beamforming technique, Lag-one Spatial Coherence Adaptive Normalization (LoSCAN), which can recover the contrast suppressed by incoherent noise. LoSCAN successfully suppressed reverberation clutter in numerous clinical examples. However, reverberation clutter is a 3-D phenomenon and can often exhibit a finite partial correlation between receive channels. Due to a strict noise-incoherence assumption, LoSCAN does not eliminate correlated reverberation clutter. This work presents a 2-D matrix array-based LoSCAN method and evaluates matrix-LoSCAN-based strategies to suppress partially correlated reverberation clutter. We validated the proposed matrix LoSCAN method using Field II simulations of a 64×64 symmetric 2-D array. We show that a subaperture beamforming (SAB) method tuned to the direction of noise correlation is an effective method to enhance LoSCAN's performance. We evaluated the efficacy of the proposed methods using fundamental and harmonic channel data acquired from the liver of two healthy volunteers using a 64×16 custom 2-D array. Compared to azimuthal LoSCAN, the proposed approach increased the contrast by up to 5.5 dB and the generalized contrast-to-noise ratio (gCNR) by up to 0.07. We also present analytic models to understand the impact of partially correlated reverberation clutter on LoSCAN images and explain the proposed methods' mechanism of image quality improvement.

Upstream Machine Learning in Radiology

Machine learning (ML) and Artificial intelligence (AI) has the potential to dramatically improve radiology practice at multiple stages of the imaging pipeline. Most of the attention has been garnered by applications focused on improving the end of the pipeline: image interpretation. However, this article reviews how AI/ML can be applied to improve upstream components of the imaging pipeline, including exam modality selection, hardware design, exam protocol selection, data acquisition, image reconstruction, and image processing. A breadth of applications and their potential for impact is shown across multiple imaging modalities, including ultrasound, computed tomography, and MRI.