Magnetic resonance elastography (MRE), an advanced MR-based imaging technique, and acoustic radiation force impulse (ARFI), an ultrasound-based imaging technique, have been shown to be accurate for diagnosing nonalcoholic fatty liver disease (NAFLD) fibrosis. However, no head-to-head comparisons between MRE and ARFI for diagnosing NAFLD fibrosis have been performed. We aimed to compare MRE versus ARFI head-to-head for diagnosing fibrosis in well-characterized patients with biopsy-proven NAFLD. Methods This cross-sectional analysis of a prospective cohort involved 125 patients (54.4% female) who underwent MRE, ARFI, and contemporaneous liver biopsies scored using the Nonalcoholic Steatohepatitis Clinical Research Network histological scoring system. MRE versus ARFI’s performances for diagnosing fibrosis were evaluated using area under receiver operating characteristic curves (AUROCs). Results The mean (±SD) age and BMI were 48.9 (±15.4) years and 31.8 (±7.0) kg/m2, respectively. For diagnosing any fibrosis (≥ stage 1), MRE’s AUROC was 0.799 (95% CI, 0.723–0.875), significantly (p=0.012) higher than ARFI’s AUROC of 0.664 (95% CI, 0.568–0.760). In stratified analysis by presence/absence of obesity, MRE was superior to ARFI for diagnosing any fibrosis in obese patients (p<0.001) but not in non-obese patients (p=0.722). MRE’s AUROCs for diagnosing ≥ stages 2, 3, and 4 fibrosis were 0.885 (95% CI, 0.816–0.953), 0.934 (95% CI, 0.863–1.000), and 0.882 (95% CI, 0.729–1.000), and ARFI’s AUROCs were 0.848 (95% CI, 0.776–0.921), 0.896 (95% CI, 0.824–0.968), and 0.862 (95% CI, 0.721–1.000). MRE had higher AUROCs than ARFI for discriminating dichotomized fibrosis stages at all dichotomization cut-points, but the AUROC differences decreased as the cut-points (fibrosis stages) increased. Conclusions MRE is more accurate than ARFI for diagnosing any fibrosis in all NAFLD patients and obese NAFLD patients, although not in non-obese NAFLD patients.

Background & Aims Liver biopsy analysis is the standard method used to diagnose nonalcoholic fatty liver disease (NAFLD). Advanced magnetic resonance imaging is a noninvasive procedure that can accurately diagnose and quantify steatosis, but is expensive. Conventional ultrasound is more accessible but identifies steatosis with low levels of sensitivity, specificity, and quantitative accuracy, and results vary among operators. A new quantitative ultrasound (QUS) technique can identify steatosis in animal models. We assessed the accuracy of QUS in the diagnosis and quantification hepatic steatosis, comparing findings with those from MRI proton density fat fraction (MRI-PDFF) analysis as a reference. Methods We performed a prospective, cross-sectional analysis of a cohort of adults (n=204) with NAFLD (MRI-PDFF≥5%) and without NAFLD (controls). Subjects underwent MRI-PDFF and QUS analyses of the liver on the same day at the University of California, San Diego, from February 2012 through March 2014. QUS parameters and backscatter coefficient (BSC) values were calculated. Patients were randomly assigned to training (n=102; mean age, 51±17 years; mean body mass index, 31±7 kg/m2) and validation (n=102; mean age, 49±17 years; body mass index, 30±6 kg/m2) groups; 69% of patients in each group had NAFLD. Results BSC (range 0.00005–0.25 1/cm-sr) correlated with MRI-PDFF (Spearman’s ρ=0.80; P<.0001). In the training group, the BSC analysis identified patients with NAFLD with an area under the curve value of 0.98 (95% confidence interval, 0.95–1.00; P<.0001). The optimal BSC cutoff value identified patients with NAFLD in the training and validation groups with 93% and 87% sensitivity, 97% and 91% specificity, 86% and 76% negative predictive values, and 99% and 95% positive predictive values, respectively. Conclusions QUS measurements of BSC can accurately diagnose and quantify hepatic steatosis, based on a cross-sectional analysis that used MRI-PDFF as the reference. With further validation, QUS could be an inexpensive, widely available method to screen the general or at-risk population for NAFLD.

Purpose We propose a deep learning‐based approach to breast mass classification in sonography and compare it with the assessment of four experienced radiologists employing breast imaging reporting and data system 4th edition lexicon and assessment protocol. Methods Several transfer learning techniques are employed to develop classifiers based on a set of 882 ultrasound images of breast masses. Additionally, we introduce the concept of a matching layer. The aim of this layer is to rescale pixel intensities of the grayscale ultrasound images and convert those images to red, green, blue (RGB) to more efficiently utilize the discriminative power of the convolutional neural network pretrained on the ImageNet dataset. We present how this conversion can be determined during fine‐tuning using back‐propagation. Next, we compare the performance of the transfer learning techniques with and without the color conversion. To show the usefulness of our approach, we additionally evaluate it using two publicly available datasets. Results Color conversion increased the areas under the receiver operating curve for each transfer learning method. For the better‐performing approach utilizing the fine‐tuning and the matching layer, the area under the curve was equal to 0.936 on a test set of 150 cases. The areas under the curves for the radiologists reading the same set of cases ranged from 0.806 to 0.882. In the case of the two separate datasets, utilizing the proposed approach we achieved areas under the curve of around 0.890. Conclusions The concept of the matching layer is generalizable and can be used to improve the overall performance of the transfer learning techniques using deep convolutional neural networks. When fully developed as a clinical tool, the methods proposed in this paper have the potential to help radiologists with breast mass classification in ultrasound.

No abstract

OBJECTIVE The purpose of this study is to explore the diagnostic performance of two investigational quantitative ultrasound (QUS) parameters, attenuation coefficient and backscatter coefficient, in comparison with conventional ultrasound (CUS) and MRI-estimated proton density fat fraction (PDFF) for predicting histology-confirmed steatosis grade in adults with nonalcoholic fatty liver disease (NAFLD). SUBJECTS AND METHODS In this prospectively designed pilot study, 61 adults with histology-confirmed NAFLD were enrolled from September 2012 to February 2014. Subjects underwent QUS, CUS, and MRI examinations within 100 days of clinical-care liver biopsy. QUS parameters (attenuation coefficient and backscatter coefficient) were estimated using a reference phantom technique by two analysts independently. Three-point ordinal CUS scores intended to predict steatosis grade (1, 2, or 3) were generated independently by two radiologists on the basis of QUS features. PDFF was estimated using an advanced chemical shift–based MRI technique. Using histologic examination as the reference standard, ROC analysis was performed. Optimal attenuation coefficient, backscatter coefficient, and PDFF cutoff thresholds were identified, and the accuracy of attenuation coefficient, backscatter coefficient, PDFF, and CUS to predict steatosis grade was determined. Interobserver agreement for attenuation coefficient, backscatter coefficient, and CUS was analyzed. RESULTS CUS had 51.7% grading accuracy. The raw and cross-validated steatosis grading accuracies were 61.7% and 55.0%, respectively, for attenuation coefficient, 68.3% and 68.3% for backscatter coefficient, and 76.7% and 71.3% for MRI-estimated PDFF. Interobserver agreements were 53.3% for CUS (κ = 0.61), 90.0% for attenuation coefficient (κ = 0.87), and 71.7% for backscatter coefficient (κ = 0.82) (p < 0.0001 for all). CONCLUSION Preliminary observations suggest that QUS parameters may be more accurate and provide higher interobserver agreement than CUS for predicting hepatic steatosis grade in patients with NAFLD.

In this work, we propose a deep learning method for breast mass segmentation in ultrasound (US). Variations in breast mass size and image characteristics make the automatic segmentation difficult. To address this issue, we developed a selective kernel (SK) U-Net convolutional neural network. The aim of the SKs was to adjust network’s receptive fields via an attention mechanism, and fuse feature maps extracted with dilated and conventional convolutions. The proposed method was developed and evaluated using US images collected from 882 breast masses. Moreover, we used three datasets of US images collected at different medical centers for testing (893 US images). On our test set of 150 US images, the SK-U-Net achieved mean Dice score of 0.826, and outperformed regular U-Net, Dice score of 0.778. When evaluated on three separate datasets, the proposed method yielded mean Dice scores ranging from 0.646 to 0.780. Additional fine-tuning of our better-performing model with data collected at different centers improved mean Dice scores by ~6%. SK-U-Net utilized both dilated and regular convolutions to process US images. We found strong correlation, Spearman’s rank coefficient of 0.7, between the utilization of dilated convolutions and breast mass size in the case of network’s expansion path. Our study shows the usefulness of deep learning methods for breast mass segmentation. SK-U-Net implementation and pre-trained weights can be found at github.com/mbyr/bus_seg .

N onalcoholic fatty liver disease (NAFLD) is the most common chronic liver disease worldwide, affecting approximately 25% of the human population (1). NAFLD covers a spectrum of liver abnormalities ranging from simple steatosis to nonalcoholic steatohepatitis. Hepatic steatosis, characterized by the accumulation of fat droplets within hepatocytes, can progress to nonalcoholic steatohepatitis, fibrosis, cirrhosis, and even hepatocellular carcinoma (1,2). Early detection and treatment may halt or reverse NAFLD progression (2). Liver biopsy remains the reference standard for diagnosing NALFD and grading hepatic steatosis. However, biopsy is costly, invasive, and inappropriate for screening.There is a critical need to develop noninvasive imaging methods to assess hepatic steatosis. Several modalities have been investigated (3-8), among which MRI and conventional (qualitative) US have the advantage of involving no ionizing radiation. Confounder-corrected chemical shiftencoded MRI can measure the proton density fat fraction (PDFF), a leading method for noninvasive quantification of hepatic steatosis (4,5). However, chemical shift-encoded MRI is not routinely accessible. Conventional US is widely

Liver fibrosis is a histological hallmark of most chronic liver diseases, which can progress to cirrhosis and liver failure, and predisposes to hepatocellular carcinoma. Accurate diagnosis of liver fibrosis is necessary for prognosis, risk stratification, and treatment decision‐making. Liver biopsy, the reference standard for assessing liver fibrosis, is invasive, costly, and impractical for surveillance and treatment response monitoring. Elastography offers a noninvasive, objective, and quantitative alternative to liver biopsy. This article discusses the need for noninvasive assessment of liver fibrosis and reviews the comparative advantages and limitations of ultrasound and magnetic resonance elastography techniques with respect to their basic concepts, acquisition, processing, and diagnostic performance. Variations in clinical contexts of use and common pitfalls associated with each technique are considered. In addition, current challenges and future directions to improve the diagnostic accuracy and clinical utility of elastography techniques are discussed. Level of Evidence: 5 Technical Efficacy Stage: 2 J. Magn. Reson. Imaging 2020;51:25–42.