Search

Excessive liver fat (steatosis) is now the most common cause of chronic liver disease worldwide and is an independent risk factor for cirrhosis and associated complications. Accurate and clinically useful diagnosis, risk stratification, prognostication, and therapy monitoring require accurate and reliable biomarker measurement at acceptable cost. This article describes a joint effort by the American Institute of Ultrasound in Medicine (AIUM) and the RSNA Quantitative Imaging Biomarkers Alliance (QIBA) to develop standards for clinical and technical validation of quantitative biomarkers for liver steatosis. The AIUM Liver Fat Quantification Task Force provides clinical guidance, while the RSNA QIBA Pulse-Echo Quantitative Ultrasound Biomarker Committee develops methods to measure biomarkers and reduce biomarker variability. In this article, the authors present the clinical need for quantitative imaging biomarkers of liver steatosis, review the current state of various imaging modalities, and describe the technical state of the art for three key liver steatosis pulse-echo quantitative US biomarkers: attenuation coefficient, backscatter coefficient, and speed of sound. Lastly, a perspective on current challenges and recommendations for clinical translation for each biomarker is offered.

To develop an ultrasound-based machine learning classifier to diagnose benignity within indeterminate thyroid nodules (ITNs) by fine-needle aspiration, 180 patients with 194 ITNs (Bethesda classes III, IV and V) undergoing surgery over a 5-y study period were analyzed. The data set was randomly divided into training and testing data sets with 155 and 39 ITNs, respectively. All nodules were evaluated by ultrasound using the American College of Radiology Thyroid Imaging Reporting and Data System by manually scoring composition, echogenicity, shape, margin and echogenic foci. Nodule size, participant age and patient sex were recorded. A support vector machine (SVM) model with a cost-sensitive approach was developed using the aforementioned eight parameters with surgical histopathology as the reference standard. Surgical pathology determined 90 (46.4%) ITNs were malignant and 104 (53.6%) were benign. The SVM model classified 14 nodules as benign in the testing data set, of which 13 were correct (sensitivity = 93.8%, specificity = 56.5%). Considering malignancy prevalence by Bethesda group, the negative predictive values of this model for Bethesda III and IV categories were 93.9% and 93. 8%, respectively. The high negative predictive value of the SVM ultrasound-based model suggests a pathway by which surgical excision of Bethesda III and IV ITNs classified as benign may be avoided.

Post-operative urinary retention is a medical condition where patients cannot urinate despite having a full bladder. Ultrasound imaging of the bladder is intermittently used to estimate urine volume for early diagnosis and management of urine retention. Moreover, the use of bladder ultrasound can reduce the need for an indwelling urinary catheter and the risk of catheter-associated urinary tract infection. Wearable ultrasound devices combined with machine-learning based bladder volume estimation algorithms reduce the burdens of nurses in hospital settings and improve outpatient care. However, existing algorithms are memory and computation intensive, thereby demanding the use of expensive GPUs. In this paper, we develop and validate a low-compute memory-efficient deep learning model for accurate bladder region segmentation and urine volume calculation. B-mode ultrasound bladder images of 360 patients were divided into training and validation sets; another 74 patients were used as the test dataset. Our 1-bit quantized models with 4-bits and 6-bits skip connections achieved an accuracy within 3.8% and 2.6%, respectively, of a full precision state-of-the-art neural network (NN) without any floating-point operations and with an 11.5× and 9.0× reduction in memory requirements to fit under 150 kB. The means and standard deviations of the volume estimation errors, relative to estimates from ground-truth clinician annotations, were 5.0±33 ml and 6.8±29 ml, respectively. This lightweight NN can be easily integrated on the wearable ultrasound device for automated and continuous monitoring of urine volume. Our approach can potentially be extended to other clinical applications such as monitoring blood pressure and fetal heart rate.

Elastography has emerged as a preferred non-invasive imaging technique for the clinical assessment of liver fibrosis. Elastography methods provide liver stiffness measurement (LSM) as a surrogate quantitative biomarker for fibrosis burden in chronic liver disease (CLD). Elastography can be performed either with ultrasound or MRI. Currently available ultrasound-based methods include strain elastography, two-dimensional shear wave elastography (2D-SWE), point shear wave elastography (pSWE), and vibration-controlled transient elastography (VCTE). MR Elastography (MRE) is widely available as two-dimensional gradient echo MRE (2D-GRE-MRE) technique. US-based methods provide estimated Young's modulus (eYM) and MRE provides magnitude of the complex shear modulus. MRE and ultrasound methods have proven to be accurate methods for detection of advanced liver fibrosis and cirrhosis. Other clinical applications of elastography include liver decompensation prediction, and differentiation of non-alcoholic steatohepatitis (NASH) from simple steatosis (SS). In this review, we briefly describe the different elastography methods, discuss current clinical applications, and provide an overview of advances in the field of liver elastography.

Steatosis is the most common cause of chronic liver disease worldwide. Ultrasound helps diagnose steatosis due to its wide availability, non-invasiveness, and safety. The hepatorenal index (HRI) is a clinically used biomarker for diagnosing steatosis. To study the automated calculation of HRI from ultrasound images, we identified a retrospective cohort undergoing abdominal ultrasound with 1320 images, one image per patient. The patients were divided into training (n=1166) and test (n=154) sets. The test set consists of 48 patients without steatosis (S0) and 106 patients with steatosis (S1-S3). An automated method was proposed to calculate HRI using four steps (a) segment the liver and the renal cortex, (b) remove pixels associated with ducts, vessels, masses, cysts, and shadowing artifacts, (c) place two regions of interest (ROIs), one each in the liver and the renal cortex, and (d) calculate the HRI as the ratio of the mean pixel value of the liver ROI to the renal cortex ROI. We explored several sets of rules for ROI placement. The method minimizing the Euclidean distance between the two ROIs, the best performer, had an area under the ROC (AUROC) curve of 0.87 compared to physicians with an AUROC of 0.82 (p=0.159). In addition, the Euclidean distance constraint showed no dependence on depth for ROI placement. The proposed automated method can be translated into clinical care without modifying existing imaging protocols to optimize the diagnostic capability of ultrasound for hepatic steatosis and assist with triaging patients.

Ultrasound imaging makes fundamental assumptions about the interrogated medium, including the propagation speed of the waves. For many clinical applications, assuming a constant speed of sound (SoS, typically 1540 m/s) is sufficiently accurate to yield diagnostically useful images. However, when the medium is appreciably heterogeneous (e.g., trans-skull, or through deep layers of subcutaneous fat), this approximation fails and the distortion of the wavefronts yields a degraded image (i.e., there is aberration). Correcting these distortions for a known SoS distribution is typically intensive computationally. Here, we demonstrate the utility of the heterogeneous angular spectrum method (HASM), a spectral method developed for passive cavitation imaging, to recover efficiently the appropriate apodization and delays in transmit and receive toward corrected images. Compared with a straight ray computation (i.e., no refraction) with mean 5% SoS variation, HASM was roughly tenfold faster, and computed delays matched to within (−84 ± 89) ns [compared with (−660 ± 312) ns error for 1540 m/s assumption]. Furthermore, transmit corrections for focused or plane-wave sequences may be obtained similarly with the method for a specified depth. HASM, thus, has promise toward real-time image enhancement and improved diagnostic utility. [Work Supported by the ASA FV Hunt Postdoctoral Fellowship.]

OBJECTIVES—: To quantify the bias of shear wave speed (SWS) measurements between different commercial ultrasonic shear elasticity systems and a magnetic resonance elastography (MRE) system in elastic and viscoelastic phantoms. METHODS—: Two elastic phantoms, representing healthy through fibrotic liver, were measured with 5 different ultrasound platforms, and 3 viscoelastic phantoms, representing healthy through fibrotic liver tissue, were measured with 12 different ultrasound platforms. Measurements were performed with different systems at different sites, at 3 focal depths, and with different appraisers. The SWS bias across the systems was quantified as a function of the system, site, focal depth, and appraiser. A single MRE research system was also used to characterize these phantoms using discrete frequencies from 60 to 500 Hz. RESULTS—: The SWS from different systems had mean difference 95% confidence intervals of ±0.145 m/s (±9.6%) across both elastic phantoms and ± 0.340 m/s (±15.3%) across the viscoelastic phantoms. The focal depth and appraiser were less significant sources of SWS variability than the system and site. Magnetic resonance elastography best matched the ultrasonic SWS in the viscoelastic phantoms using a 140 Hz source but had a − 0.27 ± 0.027-m/s (−12.2% ± 1.2%) bias when using the clinically implemented 60-Hz vibration source. CONCLUSIONS—: Shear wave speed reconstruction across different manufacturer systems is more consistent in elastic than viscoelastic phantoms, with a mean difference bias of < ±10% in all cases. Magnetic resonance elastographic measurements in the elastic and viscoelastic phantoms best match the ultrasound systems with a 140-Hz excitation but have a significant negative bias operating at 60 Hz. This study establishes a foundation for meaningful comparison of SWS measurements made with different platforms.

Background & Aims The vitronectin receptor integrin αvβ3 drives fibrogenic activation of hepatic stellate cells (HSCs). Molecular imaging targeting the integrin αvβ3 could provide a non-invasive method for evaluating the expression and the function of the integrin αvβ3 on activated HSCs (aHSCs) in the injured liver. In this study, we sought to compare differences in the uptake of [18F]-Alfatide between normal and injured liver to evaluate its utility for assessment of hepatic fibrogenesis. Methods PET with [18F]-Alfatide, non-enhanced CT, histopathology, immunofluorescence staining, immunoblotting and gene analysis were performed to evaluate and quantify hepatic integrin αvβ3 levels and liver fibrosis progression in mouse models of fibrosis (carbon tetrachloride [CCl4] and bile duct ligation [BDL]). The liver AUC divided by the blood AUC over 30 min was used as an integrin αvβ3–PET index to quantify fibrosis progression. Ex vivo analysis of frozen liver tissue from patients with fibrosis and cirrhosis verified the animal findings. Results Fibrotic mouse livers showed enhanced [18F]-Alfatide uptake and retention compared to control livers. The radiotracer was demonstrated to bind specifically with integrin αvβ3, which is mainly expressed on aHSCs. Autoradiography and histopathology confirmed the PET imaging results. Further, the mRNA and protein level of integrin αvβ3 and its signaling complex were higher in CCl4 and BDL models than controls. The results obtained from analyses on human fibrotic liver sections supported the animal findings. Conclusions Imaging hepatic integrin αvβ3 with PET and [18F]-Alfatide offers a potential non-invasive method for monitoring the progression of liver fibrosis. Lay summary Integrin αvβ3 expression on activated hepatic stellate cells (aHSCs) is associated with HSC proliferation during hepatic fibrogenesis. Herein, we show that a radioactive tracer, [18F]-Alfatide, binds to integrin αvβ3 with high affinity and specificity. [18F]-Alfatide could thus be used as a non-invasive imaging biomarker to track hepatic fibrosis progression.

Background Angiotensin-converting enzyme 2, a target of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), demonstrates its highest surface expression in the lung, small bowel, and vasculature, suggesting abdominal viscera may be susceptible to injury. Purpose To report abdominal imaging findings in patients with coronavirus disease 2019. Materials and Methods In this retrospective cross-sectional study, patients consecutively admitted to a single quaternary care center from March 27 to April 10, 2020, who tested positive for SARS-CoV-2 were included. Abdominal imaging studies performed in these patients were reviewed, and salient findings were recorded. Medical records were reviewed for clinical data. Univariable analysis and logistic regression were performed. Results A total of 412 patients (average age, 57 years; range, 18 to >90 years; 241 men, 171 women) were evaluated. A total of 224 abdominal imaging studies were performed (radiography, n = 137; US, n = 44; CT, n = 42; MRI, n = 1) in 134 patients (33%). Abdominal imaging was associated with age (odds ratio [OR], 1.03 per year of increase; P = .001) and intensive care unit (ICU) admission (OR, 17.3; P < .001). Bowel-wall abnormalities were seen on 31% of CT images (13 of 42) and were associated with ICU admission (OR, 15.5; P = .01). Bowel findings included pneumatosis or portal venous gas, seen on 20% of CT images obtained in patients in the ICU (four of 20). Surgical correlation (n = 4) revealed unusual yellow discoloration of the bowel (n = 3) and bowel infarction (n = 2). Pathologic findings revealed ischemic enteritis with patchy necrosis and fibrin thrombi in arterioles (n = 2). Right upper quadrant US examinations were mostly performed because of liver laboratory findings (87%, 32 of 37), and 54% (20 of 37) revealed a dilated sludge-filled gallbladder, suggestive of bile stasis. Patients with a cholecystostomy tube placed (n = 4) had negative bacterial cultures. Conclusion Bowel abnormalities and gallbladder bile stasis were common findings on abdominal images of patients with coronavirus disease 2019. Patients who underwent laparotomy often had ischemia, possibly due to small-vessel thrombosis. © RSNA, 2020.

To develop a contrast-enhanced ultrasound (CEUS)–based model for differentiating cirrhotic liver lesions and for active surveillance of hepatocellular carcinoma (HCC). Patients with focal liver lesions (FLLs) with biopsy/resection-proven pathology and pre-procedure CEUS were enrolled from our institution between January 2011 and November 2014. Univariable and multivariable regression models were constructed using qualitative CEUS features and/or contrast arrival time ratio (CATR). The optimism-adjusted Harrell’s generalized concordance index (CH) was used to quantify the discriminatory ability of each CEUS feature and model. A total of 149 patients (113 men and 36 women) with 162 FLLs were enrolled with mean age 53.4 ± 12.7 years. A 0.1-unit reduction in CATR was associated with a 68% increase in the odds of having a higher nodule ranking (RN

Non-compressible internal hemorrhage (NCIH) is the most common cause of death in acute non-penetrating trauma. NCIH management requires accurate hematoma localization and evaluation for ongoing bleeding for risk stratification. The current standard point-of-care diagnostic tool, the focused assessment with sonography for trauma (FAST), detects free fluid in body cavities with conventional B-mode imaging. The FAST does not assess whether bleeding is ongoing, at which location(s), and to what extent. Here, we propose contrast-enhanced ultrasound (CEUS) techniques to better identify, localize, and quantify hemorrhage. We designed and fabricated a custom hemorrhage-mimicking phantom, comprising a perforated vessel and cavity to simulate active bleeding. Lumason contrast agents (UCAs) were introduced at clinically relevant concentrations (3.5×108 bubbles/ml). Conventional and contrast pulse sequence images were captured, and post-processed with bubble localization techniques (SVD clutter filter and bubble localization). The results showed contrast pulse sequences enabled a 2.2-fold increase in the number of microbubbles detected compared with conventional CEUS imaging, over a range of flow rates, concentrations, and localization processing parameters. Additionally, particle velocimetry enabled mapping of dynamic flow within the simulated bleeding site. Our findings indicate that CEUS combined with advanced image processing may enhance visualization of hemodynamics and improve non-invasive, real-time detection of active bleeding.

Effective masking policies to prevent the spread of airborne infections depend on public access to masks with high filtration efficacy. However, poor face-fit is almost universally present in pleated multilayer disposable face masks, severely limiting both individual and community respiratory protection. We developed a set of simple mask modifications to mass-manufactured disposable masks, the most common type of mask used by the public, that dramatically improves both their personalized fit and performance in a low-cost and scalable manner. These modifications comprise a user-moldable full mask periphery wire, integrated earloop tension adjusters, and an inner flange to trap respiratory droplets. We demonstrate that these simple design changes improves quantitative fit factor by 320%, triples the level of protection against aerosolized droplets, and approaches the model efficacy of N95 respirators in preventing the community spread of COVID-19, for an estimated additional cost of less than 5 cents per mask with automated production.TeaserA personalizable, low-cost mask improves facial fit, reduces user exposure, and decreases spread of contagious aerosols.

OBJECTIVE: Data engineering is the foundation of effective machine learning model development and research. The accuracy and clinical utility of machine learning models fundamentally depend on the quality of the data used for model development. This article aims to provide radiologists and radiology researchers with an understanding of the core elements of data preparation for machine learning research. We cover key concepts from an engineering perspective, including databases, data integrity, and characteristics of data suitable for machine learning projects, and from a clinical perspective, including the HIPAA, patient consent, avoidance of bias, and ethical concerns related to the potential to magnify health disparities. The focus of this article is women’s imaging; nonetheless, the principles described apply to all domains of medical imaging. CONCLUSION: Machine learning research is inherently interdisciplinary: effective collaboration is critical for success. In medical imaging, radiologists possess knowledge essential for data engineers to develop useful datasets for machine learning model development.

To evaluate the value of pre-stenting and early post-stenting (

We developed a method to create customizable phantoms suitable for endocavitary imaging and interventional research, based on the fabrication of an acrylic phantom mold, and development of a phantom matrix composed of gelatin, agar, graphite particles, and propanol. Our phantom was mechanically stable, easily fabricated, and highly adjustable, and its ultrasound (US) and magnetic resonance imaging (MRI) scans showed the qualification for the procedure guidance compared with the human prostate image using the same US system. To test the feasibility of the phantom for the research, the seeds placement guided by MRI/US fusion was performed, and the overall test error (distance from the seed center to the virtual lesion center in olives) was 2.59 ± 0.59 mm. We have created a simple, low-cost, configurable, gelatin-based phantom and tested its feasibility for simulating endorectal interventional US procedures. The design of the phantom mold and matrix is likely to be useful to the broader medical training community, and the preliminary data from the experiment of MRI/US-guided seeds placement showed its potential to test the clinical hypothesis in US research.

We evaluated variation in fibrosis staging caused by depth, pre-load force and measurement off-axis distance on different ultrasound shear wave elastography (SWE) systems prospectively in 20 patients with diffuse liver disease. Shear wave speed (SWS) was measured with transient elastography, acoustic radiation force impulse (ARFI) and 2-D shear wave elastography (SWE). ARFI and 2-D-SWE measurements were obtained at different depths (3, 5 and 7 cm), with different pre-load forces (4, 7 and 10N and variable) and at 0, 2 and 4cm off the central axis of the transducer. A single, blinded pathologist staged fibrosis using the METAVIR system (F0-F4). Area under the receiver operating characteristic curve was charted to differentiate significant fibrosis (F ≥ 2). Depth was the only factor found to influence ARFI-derived values; no acquisition factors were found to affect 2-D-SWE SWS values. ARFI and 2-D-SWE for diagnosis of significant fibrosis at a depth of 7cm along the central axis had good diagnostic performance (areas under the receiver operating characteristic curve: 0.92 and 0.82, respectively), comparable to that of transient elastography. Further investigation of this finding will likely be of interest.

In this work, we propose accurate binary Depthwise Separable Convolutional Neural Networks (DSCNNs) for medical image segmentation. The networks are binarized by learning the distribution of weights and activations, and by using parameter-free skip connections in their encoder and decoder structure. We design full precision DSCNNs based on a symmetric encoder-decoder, feature pyramid network with an asymmetric decoder, and spatial pyramid pooling with atrous convolutions strategies for image segmentation. The DSCNNs have 14 X and 8 X fewer number of model parameters and operations, respectively, than standard segmentation networks. The trained full precision DSCNNs are used as baselines to achieve accurate binary DSCNNs. The networks are trained on two medical ultrasound datasets, a public fetal skull dataset and a privileged bladder dataset. The accuracy of the binary DSCNNs are within a 3% drop from the full precision networks on both the medical datasets.

Medical ultrasound (US) is an indispensable medical imaging modality. It is nonionizing, portable, and can provide real-time visualization of tissue dynamics. In particular, US imaging of the heart, known as echocardiography, has been routinely used in the diagnosis, management, and treatment of patients with heart disease. However, US is highly operator dependent, thus suffering from high intra- and intervariability. Leveraging recent advances in artificial intelligence, machine learning has been applied to circumvent existent limitations of US-based imaging. In this chapter, we highlight the recent progress and discuss remaining challenges as well as future opportunities. The focus is on cardiovascular applications, but common themes and differences for noncardiovascular applications are also summarized.

OBJECTIVE This study aims to confirm the prevalence of incidental cervical extension of normal thymus in children and adolescents undergoing neck ultrasound and describe the ultrasound appearance to minimize future misdiagnosis. MATERIALS AND METHODS This retrospective study was conducted in a single institution. Thyroid and lower neck ultrasound images of the consecutive pediatric subjects between January 1, 2011 and September 30, 2017 were independently reviewed by 2 radiologists for the presence of cervical thymus. When identified on sonographic images, cervical thymus was described on the basis of echogenicity, location, and shape. RESULTS In 278 consecutive cases, the 2 reviewers identified 105 (37.8%) and 103 (37.1%) cases respectively as having sonographically visible tissue in the expected location of cervical extension of the thymus. The internal echotexture was variable with 38.1% of cases being hypoechoic, 37.1% mixed, and 24.8% hyperechoic. Cervical extension of the thymus was most commonly (65.0%) to the left of the trachea or (30.9%) bilateral/anterior to the trachea; isolated right paratracheal thymus was uncommon. Thymic shape was variable: quadrilateral (30.9%), oval (29.9%), triangular (25.8%), and other (13.4%). The logistic regression model including age, gender, and BMI z-scores showed that, when controlled for sex and BMI z-scores, younger age was a predictor for the presence of cervical thymic extension (p

Bone stress injuries (BSI) are caused by repetitive loading patterns that do not allow adequate time for bone remodeling. These injuries are most prevalent among endurance athletes and military personnel, particularly military recruits. Treatment for BSI usually includes some level of activity modification and, with prescribed recovery times of weeks to months, these injuries impose significant time lost from training or duties. The gold standard method for diagnosing and grading stress injuries is MRI. Ultrasound would be an attractive alternative or supplement for BSI diagnosis and recovery monitoring, since it can be applied at point-of-care locations and repeat scans are safe and relatively low cost. We describe the Image Registered Quantitative Ultrasound (IR-QUS) technique, aimed at evaluating BSI on the basis of surface roughness as an indicator of rapid bone remodeling. IR-QUS collects angle-diverse acoustic measurements on a focal point and detects relative increases in diffusive versus specular scattering. We demonstrate the sensitivity of this method to characterize surface roughness on the scale of tens of microns, and show the feasibility of detecting a localized field of fractures on an ex vivo intact bone sample. Experimental results suggest reliable detection of roughened surfaces with a diffuse-to-specular energy metric that increases by 10 dB relative to smooth regions.