Chronic liver disease is a substantial cause of mortality and morbidity worldwide, with metabolic dysfunction-associated steatotic liver disease affecting about one-third of the population. Shear-wave elastography (SWE) is a safe and noninvasive US method for quantifying liver fibrosis. It is a widely available, low-cost imaging test that has proven accurate and reliable for diagnosing advanced liver fibrosis, a key determinant of mortality risk among patients with chronic liver disease. Established society guidelines for SWE outline patient selection, preparation, image acquisition, examination reliability characterization, and reporting. However, inconsistencies between guidelines and the complexity of recommendations pose barriers to practical clinical implementation. To address this gap, established best practices for US SWE are reviewed and synthesized, actionable recommendations for streamlining clinical deployment are provided, and common pitfalls are highlighted.

Manual transcription of quantitative medical ultrasound measurements into written reports is time-consuming and error-prone. Current automated transcription approaches are difficult to implement clinically due to data structure heterogeneity across vendors and the need for image header access. We developed and validated a flexible, rapidly deployable artificial intelligence image processing pipeline that extracts and structures quantitative diagnostic data directly from ultrasound images without requiring metadata access or proprietary vendor integration. A dataset of 828 synthetic ultrasound images was generated by creating 36 augmented variations of each of the 23 original phantom images. This dataset was split into training (787 images) and validation sets (41 images) for model development. Two independent test sets were used to evaluate performance: the 23 original phantom images and 105 deidentified clinical images. The Gemma 3 4B Vision model was fine-tuned using Low-Rank Adaptation (LoRA, rank 8, alpha 8) for a single epoch, achieving training loss of 0.397 and validation loss of 0.289. The model was integrated into a three-stage pipeline: (1) image loading across diverse formats; (2) structured reasoning with explicit explanations; and (3) measurement extraction and summary generation. This reasoning-driven design promotes interpretability by exposing reasoning traces. Testing used phantom images simulating shear wave elastography, attenuation coefficient, and common bile duct measurements to mimic in-vivo quantitative imaging. The pipeline achieved 96% accuracy on 23 phantom images and 52.4% accuracy on 105 clinical images. Results demonstrate our vendor-agnostic pipeline delivers interpretable, traceable, and efficient diagnostic data extraction, providing a rapidly deployable solution for quantitative medical imaging.

In patients with Metabolic Dysfunction-associated Steatotic Liver Disease (MASLD), accurate and low-cost non-invasive risk stratification remains a major unmet need. We developed a clinical records-based neural network integrating patient history, routine laboratory tests, and ultrasound imaging features. Here we show that in an internal test set (n = 209), the model achieved receiver operating characteristic area under the curve (ROC-AUC) values of 0.85 vs 0.82 (F ≥2), 0.90 vs 0.86 (F ≥3), 0.96 vs 0.89 (F  = 4) compared with Fibrosis-4 (FIB-4). To address limitations of ROC-AUC, we applied RP-AUC0.5-0.7, a recall-precision metric focused on clinically relevant precision range, showing improved performance over FIB-4. External validation (n = 194) shows reduced liver biopsy failure rate from 86.6% to 50.0% for at-risk metabolic dysfunction-associated steatohepatitis (MASH) prediction. Our work presents a low-cost neural network improving FIB-4, introduces RP-AUC0.5-0.7 for biomarker comparison, and provides a generalizable framework for evaluating screening biomarkers in clinical care and drug trials.

OBJECTIVES:Renal cystic lesions are exceedingly common and typically benign, though accurate diagnostic techniques are required to recognize the subset of malignant lesions that require timely intervention. Ultrasound (US) is the preferred first line test for renal cyst evaluation because it is safe, widely accessible, and low cost. US imaging, however, is limited in individuals with obesity, as subcutaneous fat introduces imaging artifacts (eg, aberration) that obscure features critical for assessing a cyst's malignant potential. METHODS:In this work, we propose a 2-stage image-correction algorithm to improve the resolution, contrast, and potential diagnostic utility of US images of renal cysts. The method combines sound speed correction for beamforming with masking based on the spatial coherence of the beamformed signals. RESULTS:In a pilot cohort of 10 subjects, we observed improved image sharpness and contrast-to-noise ratio in nearly all cases compared to baseline images (mean improvements of approximately 5 and 10%, respectively). Additionally, 2 expert readers preferred nearly universally the corrected images in a blinded review. CONCLUSION:Together, the results from this pilot study suggest that our method has translational potential to improve US image quality and enhance clinical confidence in managing the common and clinically important challenge of renal cysts.

Decompensation represents a critical transition in the course of cirrhosis, yet clinicians have limited non-invasive tools to reliably predict its onset. In this study, we propose a novel imaging-based approach that leverages large-scale computer vision models to analyze routine abdominal ultrasound images and extract predictive features beyond those captured by traditional laboratory-based risk scores. Ultrasound is widely available, low cost, and suitable for longitudinal surveillance, making it an attractive modality for scalable risk stratification and long-term follow-up. Our framework integrates automated ultrasound data processing with modern deep learning architectures to identify patients at high risk of decompensation prior to the occurrence of clinical deterioration. This non-invasive strategy offers a practical complement to existing clinical scoring systems and may enable earlier, more proactive management of patients with compensated cirrhosis.

Metabolic dysfunction-associated steatotic liver disease (MASLD) affects 40% of the world's population. Despite an association with increased cardiovascular morbidity, MASLD usually does not lead to liver-related adverse outcomes until fibrosis develops. MASLD with clinically significant fibrosis (histologic fibrosis stage ≤2) is associated with poorer prognosis and helps guide management, including specialty care referral. Among adults with MASLD, those with noncirrhotic clinically significant fibrosis (stage 2 or 3) may be eligible for pharmacotherapy to prevent progression to cirrhosis; those with advanced fibrosis (stage ≤3) usually require intensive monitoring; and those with cirrhosis (stage 4) may require additional supportive therapy and surveillance for gastroesophageal varices or primary liver cancer. Professional societies now recommend selective screening of high-risk patients to detect MASLD with clinically significant or advanced fibrosis. While guideline details vary, the screening population generally includes those with diabetes, other cardiometabolic factors, or incidentally detected steatosis. Screening begins with blood-based tests; second-line assessment, possibly including noninvasive imaging, seeks to verify MASLD with clinically significant or advanced fibrosis, stratify risk, and inform management. This Special Series Review describes current and best practices for screening adults to detect and stratify MASLD, focusing on imaging's critical and evolving role in addressing this public health challenge.

BACKGROUND:Ultrasound-based surveillance for HCC is limited by suboptimal sensitivity and low adherence. Blood-based biomarkers are potential alternatives for routine HCC surveillance. We aimed to define optimal sensitivity, specificity, and cost parameters under which blood-based biomarkers could serve as cost-effective alternatives in patients with cirrhosis. METHODS:We developed a microsimulation model of the natural history of HCC in individuals with compensated cirrhosis. We compared the cost-effectiveness of biannual surveillance using 2 blood-based biomarkers-HES V2.0 and GALAD-versus ultrasound-based surveillance across a range of sensitivity, specificity, test costs, and adherence levels. Outcomes included quality-adjusted life years (QALYs), costs, and incremental cost-effectiveness ratios (ICERs), with a willingness-to-pay threshold of $100,000/QALY. RESULTS:At a $200 test cost, the optimal HES V2.0 performance was 83.8% sensitivity and 62.4% specificity, yielding more QALYs than ultrasound-based surveillance (6790 vs. 6780 per 1000 patients) with an ICER of $27,686/QALY. The optimal GALAD performance was 65.9% sensitivity and 76.4% specificity, yielding a modest QALY increase (6782 vs. 6780 per 1000 patients) with an ICER of $21,374/QALY. HES V2.0 detected more very early-stage HCCs than ultrasound-based surveillance (50.9 vs. 42.4 per 1000 patients) but increased downstream diagnostic imaging, whereas GALAD yielded smaller QALY gains with a moderate increase in diagnostic testing. At specificity ≥80%, neither biomarker was cost-effective, but cost-effectiveness was strongly influenced by adherence and test cost. CONCLUSIONS:Blood-based biomarkers such as HES V2.0 and GALAD can be cost-effective alternatives to ultrasound-based HCC surveillance under defined performance and cost conditions. These findings provide quantitative benchmarks to guide clinical evaluation and inform the development of future blood-based strategies for early HCC detection.

Hemorrhage is the leading cause of preventable death from trauma, with most fatalities occurring in pre-hospital settings. The current standard for point-of-care hemorrhage detection is the focused assessment with sonography in trauma (FAST) exam, or computed tomography (CT) angiography after hospital arrival. Contrast-enhanced ultrasound (CEUS) has been proposed as a next-generation FAST exam, due to potential for (1) differentiating active vs. inactive bleeding, (2) localizing bleeding sources, and (3) guiding percutaneous and/or endovascular treatments to temporize bleeding for prolonged transport. While these advantages have been reported in human trauma case studies, characterization of the minimum hemorrhage rate detectable by CEUS remains a translational research need. Prior work using conventional B-mode imaging and microvascular CEUS algorithms showed detection of 5 mL/min flow in a generic hemorrhage phantom. Here, we apply nonlinear imaging waveforms (pulse inversion, PI, and amplitude modulation, AM) with an iterative spatiotemporal filter that further enhances contrast signals (fast iterative-shrinkage thresholding algorithm, FISTA) to detect significantly lower flow rates in an anatomically realistic abdominal hemorrhage phantom. We demonstrate detection of bleeding at a minimum flow rate of 0.5 mL/min, comparable to multidetector CT. We achieve this detection limit even with 10% of the standard contrast dose (area under the receiver operating characteristic curve of 0.90 and 0.84 for PI+FISTA and AM+FISTA, respectively, compared to 0.38 for conventional B-mode), suggesting that CEUS examination is feasible throughout contrast wash-in / wash-out. Finally, to bridge towards in vivo characterization, we present a novel porcine liver hemorrhage model tailored to evaluating CEUS capabilities.

Metabolic dysfunction-associated steatotic liver disease (MASLD) affects approximately 30

Renal cystic lesions are common, with approximately 25% of individuals over 40 and 50% over 50 having at least one cyst, the majority of which are benign. However, some lesions can become malignant, necessitating regular imaging to distinguish between benign and malignant cysts. Ultrasound (US) is a preferred imaging modality due to its safety and cost-effectiveness, but its efficacy is often limited in obese patients due to poor image quality resulting from fat attenuation and phase aberration. This can lead to misdiagnoses, as smaller features indicative of pathology can be obscured or made ambiguous. In this study, we employ a two-stage approach to enhance in vivo US images of renal cysts. First, a bulk speed of sound correction is applied to optimize resolution in the delay-and-sum images. Next, we implement a spatial correction based on short-lag spatial coherence of the channel data, computed for the identified optimal speed of sound. Among all subjects (N = 10) with body mass indices ranging from 18 to 38, cystic boundaries were sharpened, and the generalized contrast-to-noise ratio (gCNR) improved by 7.4% [CI (5.5, 9.2)%, p < 0.0001] compared to uncorrected delay-and-sum images. Furthermore, the weighting of the coherence mask can be varied in real time during evaluation to enhance the diagnostic value of the composite image. This approach demonstrates significant potential for rapid clinical translation and for augmenting the critical role of US in monitoring kidney cysts.

Objective: We assessed the diagnostic performance of ultrasound two-dimensional shear wave elastography (US 2D-SWE) to predict clinically significant fibrosis (CSF) in patients with serologic iron overload (SIO) and the subgroup with histologic liver iron overload (LIO). Methods: A single-center retrospective cross-sectional study of adults with SIO (serum ferritin >= 200 ng/mL in females and >= 300 ng/mL in males) and suspected chronic liver disease with nonfocal liver biopsy results and US 2D-SWE exams within 1 year was performed. Histopathological fibrosis stage >= 2 and liver iron >= 2+ was considered CSF and LIO, respectively. Univariate logistic regression to assess prediction of CSF by Young's modulus (YM) and serum ferritin was performed. Sensitivity and specificity were reported at optimal YM threshold determined by the Youden Index. Results: 272 cases were included (211 (77.6%) females, 88 (32.4%) CSF cases) with mean (+/- standard deviation) age of 50.0 (13.6) years. Median YM predicted CSF in patients with SIO (AUC 0.73, 95% confidence intervals (CI) 0.66 -0.80, odds ratio (OR) 1.12), p < 0.001. Optimal YM threshold was 11 kPa (sensitivity 58%, specificity 79%). Subgroup analysis of 47 LIO cases (39 women, mean age 52.5 +/- 11.6 years, 17 (36.2%) CSF) showed that median YM predicted CSF (AUC 0.85, 95% CI 0.73-0.97, OR 1.39), p <0.001. Optimal YM threshold was 11 kPa (sensitivity 77%, specificity 87%). Conclusion: 2D-SWE is a promising, widely available, and noninvasive tool for diagnosing liver fibrosis in iron overload, including when magnetic resonance elastography may be nondiagnostic due to iron-related artifact.

Ultrasound (US) imaging is an indispensable tool for diagnostic imaging, particularly given its cost, safety, and portability profiles compared to other modalities. However, US is challenged in subjects with morphological heterogeneity (e.g., those with overweight or obesity), largely because conventional imaging algorithms do not account for such variation in the beamforming process. Specific knowledge of the these spatial variations enables supplemental corrections of these algorithms, but with added computational complexity. Wavefield correlation imaging (WCI) enables efficient image formation in the spatial frequency domain that, in its canonical formulation, assumes a uniform medium. In this work, we present an extension of WCI to arbitrary known speed-of-sound distributions directly in the image formation process, and demonstrate its feasibility in silico, in vitro, and in vivo. We report resolution improvements of over 30

Simplification of instructions for elastography box placement significantly improved reliability of liver shear-wave elastography measurements, decreased the frequency of noncompliance with few of the quality metrics, and reduced the number of nondiagnostic examinations.

Background:Metabolic dysfunction-associated steatotic liver disease affects 1 in 3 people worldwide. Ultrasound shear wave elastography (SWE) in obese patients, the target population for testing, is hampered by beam attenuation, leading to unreliable liver fibrosis quantification. Purpose:We assess the safety and efficacy of increased push mechanical index (IPMI) above U.S. Food and Drug Administration limits to improve SWE. Materials and methods:This single-center prospective trial (July 2023-April 2024) (NCT05792423) enrolled healthy adults stratified by body mass index (BMI). Participants underwent conventional push pulse (mechanical index [MI] 1.4) and IPMI (MI 2.5) SWE (GE Healthcare LOGIQ E10) performed by 1 of 3 sonographers and serial liver function testing (LFT) before and up to 7 days after imaging. Liver injury was defined as increased serum alanine transaminase (ALT), aspartate aminotransferase (AST), or alkaline phosphatase (ALP) (non-inferiority margins: AST 7.5 U/L, ALT 12 U/L, ALP 17.5 U/L). Secondary endpoints included SWE variability and measurement number. Results:Twenty-two analyzable participants (mean age 39.6 ± 16.3 years; 15 women) had normal BMI (6), overweight (6), class 1 obesity (7), and class 2 obesity (3). Conventional shear wave speed was 1.34 ± 0.21 m/s, and IPMI yielded 1.36 ± 0.20 m/s (velocities ≤ 1.34 m/s indicate minimal or no fibrosis). The mean [95% CI] LFT change from baseline to day 1 was 1) AST: -0.86 [-2.34, 0.61], P = .24, 2) ALT: 0.32 [-1.04, 1.68], P = .63, 3) ALP: 1.73 [-1.02, 4.47], P = .21. The upper 95% CI for all biomarkers met non-inferiority criteria. Mean IPMI interquartile range (IQR) to median ratio decreased 0.019 (29.2% relative reduction) (P = .01) with 0.68 [IQR: 0.0.75] (P = .058), fewer average attempts. Conclusion:IPMI SWE in healthy volunteers did not cause injury and reduced measurement variability. IPMI SWE should be developed to improve examination quality and reliability in obese patients.

OBJECTIVE:Monitoring liver stiffness is essential for managing chronic liver disease, which poses a major public health challenge. Shear wave elastography (SWE), a non-invasive ultrasound-based technique, is commonly used to quantify liver stiffness. However, its performance can be compromised in individuals with higher body mass indices (BMIs) due to increased ultrasound absorption and distortion. Increasing the intensity of the ultrasound push beam could potentially improve signal quality, but regulatory limits currently restrict this due to safety concerns. This pilot study investigated the efficacy of increasing the push pulse mechanical index (MI) from a conventional value of 1.4 to 2.5 toward improving signal quality, and reducing measurement variability and failure rates. METHODS:Healthy volunteers (N=22) stratified by BMI underwent SWE with conventional and increased MI push pulses. The resulting data were processed with conventional SWE algorithms, and the signal and measurement quality of the results were analyzed. RESULTS:We found that the higher MI improved the signal-to-noise ratio by 4.6 dB (p<10-4, 95% confidence interval: 3.4-5.8 dB) and reduced the measurement's coefficient of variation by 13% (p<10-4, 95% confidence interval: 5.8%-20.3%), enhancing the success rate of SWE examinations, especially for subjects with a BMI over 30. Liver function tests before and after the SWE examinations showed no signs of bioeffects or harm based on serum biomarkers. CONCLUSION:These results suggest that increasing the push pulse MI to 2.5 improves the diagnostic utility of SWE, particularly for individuals with a higher BMI, without introducing significant additional risk. This approach could further enhance SWE's vital role in the monitoring of chronic liver disease at a population scale.

In shear wave elastography (SWE), speckle tracking is employed to obtain the particle velocity from IQ or RF data. Conventional phase shift speckle tracking methods in SWE compute phase changes between adjacent firings, enabling estimation of local particle velocity. Although computationally efficient, these methods are highly sensitive to noise, as they rely on only one adjacent firing for each sample or patch. Deep learning methods have improved particle velocity estimation, with recent approaches using 3D convolutions over limited firings. However, these multi-resolution architectures are restricted to short temporal windows, limiting their ability to model long-range dependencies and generalize across varying conditions. To address this, we introduce Lorast, a long-range spatiotemporal velocity tracking network. This network disentangles spatiotemporal feature extraction by explicitly computing the phase difference between firings as a form of known operator inside the network architecture. Spatial features are extracted using 2D convolutions, while temporal dependencies are modeled with Mamba, a recent sequence modeling architecture outperforming transformers. This separation eliminates the need to learn joint spatiotemporal representations. We evaluate the performance of the compared methods on simulation data, and experimental phantom data.

BACKGROUND:Hospitals struggle to predict critical outcomes. Traditional early warning systems, like NEWS and MEWS, rely on static variables and fixed thresholds, limiting their adaptability, accuracy, and personalization. METHODS:We previously developed the Enhanced Transformer for Health Outcome Simulation (ETHOS), an artificial intelligence (AI) model that tokenizes patient health timelines (PHTs) from electronic health records and uses transformer-based architectures to predict future PHTs. ETHOS is a versatile framework for developing a wide range of applications. In this work, we develop the Adaptive Risk Estimation System (ARES) that leverages ETHOS to compute dynamic, personalized risk probabilities for clinician-defined critical events. ARES also features a personalized explainability module that highlights key clinical factors influencing risk estimates. We evaluated ARES using the MIMIC-IV v2.2 dataset, together with its emergency department extension, and benchmarked performance against both classical early warning systems and contemporary machine learning models. RESULTS:The entire dataset was tokenized, resulting in 285,622 PHTs (63% with at least 1 hospital admission), comprising over 357 million tokens. ETHOS outperformed benchmark models in predicting hospital admissions, intensive care unit admissions, and prolonged stays, achieving superior area under the curve scores. Its risk estimates were robust across demographic subgroups, with calibration curves confirming model reliability. The explainability module provided valuable insights into patient-specific risk factors. CONCLUSIONS:ARES, powered by ETHOS, advances predictive health care AI by delivering dynamic, real-time, personalized risk estimation with patient-specific explainability. Although our results are promising, the clinical impact remains uncertain. Demonstrating ARES's true utility in real-world settings will be the focus of our future work.

The shear wave elastography (SWE) provides quantitative markers for tissue characterization by measuring the shear wave speed (SWS), which reflects tissue stiffness. SWE uses an acoustic radiation force pulse sequence to generate shear waves that propagate laterally through tissue with transient displacements. These waves travel perpendicular to the applied force, and their displacements are tracked using high-frame-rate ultrasound. Estimating the SWS map involves two main steps: speckle tracking and SWS estimation. Speckle tracking calculates particle velocity by measuring RF/IQ data displacement between adjacent firings, while SWS estimation methods typically compare particle velocity profiles of samples that are laterally a few millimeters apart. Deep learning (DL) methods have gained attention for SWS estimation, often relying on supervised training using simulated data. However, these methods may struggle with real-world data, which can differ significantly from the simulated training data, potentially leading to artifacts in the estimated SWS map. To address this challenge, we propose a physics-inspired learning approach that utilizes real data without known SWS values. Our method employs an adaptive unsupervised loss function, allowing the network to train with the real noisy data to minimize the artifacts and improve the robustness. We validate our approach using experimental phantom data and in vivo liver data from two human subjects, demonstrating enhanced accuracy and reliability in SWS estimation compared with conventional and supervised methods. This hybrid approach leverages the strengths of both data-driven and physics-inspired learning, offering a promising solution for more accurate and robust SWS mapping in clinical applications.

Point-of-care US (POCUS), also known as focused US, targeted US, clinical US, bedside US, or emergency US, is a rapidly evolving, dynamic imaging tool that aids in rapid diagnoses and informed bedside decision-making. POCUS devices are portable, affordable, and considered easy to use for nonradiologists. POCUS has several advantages over diagnostic US, including improved accessibility, real-time assessment, and provision of immediate results. This makes POCUS useful to guide bedside procedures, such as vascular access for central and peripheral line placement. However, advancing POCUS across a broad health care network requires collaboration among the POCUS stakeholders and clinical departments within the existing multidisciplinary ecosystem. This collaboration should support nonradiology colleagues to ensure quality patient care and streamlined operations while ensuring POCUS imaging availability for comparison. Creating a comprehensive POCUS program in a large hospital setting requires policies and guidelines like those of other clinical imaging programs. Leadership, training programs, credentialing bodies, documentation, proper image storage, and quality assurance are essential for any clinical imaging program, including POCUS. This review article presents a framework aimed to standardize POCUS practice across a health care system, thereby enhancing both diagnostic accuracy and patient outcomes.