Background Nonalcoholic fatty liver disease (NAFLD) is rapidly increasing worldwide. It is subdivided into nonalcoholic fatty liver (NAFL) and the more aggressive form, nonalcoholic steatohepatitis (NASH), which carries a higher risk of developing fibrosis and cirrhosis. There is currently no reliable non‐invasive method for differentiating NASH from NAFL. Purpose To investigate the ability of magnetic resonance imaging (MRI)‐based imaging biomarkers to diagnose NASH and moderate fibrosis as well as assess their repeatability. Study Type Prospective. Subjects Sixty‐eight participants (41% women) with biopsy‐proven NAFLD (53 NASH and 15 NAFL). Thirty participants underwent a second MRI in order to assess repeatability. Field Strength/Sequence 3.0 T; MR elastography ( MRE ) (a spin‐echo echo‐planar imaging [ SE‐EPI ] sequence with motion‐encoding gradients), MR proton density fat fraction ( PDFF ) and R2 * mapping (a multi‐echo three‐dimensional gradient‐echo sequence), T1 mapping (a single‐point saturation‐recovery technique), and diffusion‐weighted imaging ( SE‐EPI sequence). Assessment Quantitative MRI measurements were obtained and assessed alone and in combination with biochemical markers (cytokeratin‐18 [CK18] M30, alanine transaminase [ALT], and aspartate transaminase [AST]) using logistic regression models. Models that could differentiate between NASH and NAFL and between moderate to advanced fibrosis (F2–4) and no or mild fibrosis (F0–1), based on the histopathological results, were identified. Statistical Tests Independent samples t ‐test, Pearson's chi‐squared test, area under the receiver operating characteristic curve (AUROC), Spearman's correlation, intra‐individual coefficient of variation, and intraclass correlation coefficient (ICC). Statistical significance was set at P < 0.05. Results There was a significant difference between the NASH and NAFL groups with liver stiffness assessed with MRE, CK18 M30, and ALT, with an AUROC of 0.74, 0.76, and 0.70, respectively. Both MRE and PDFF contributed significantly to a bivariate model for diagnosing NASH (AUROC = 0.84). MRE could significantly differentiate between F2–4 and F0–1 (AUROC = 0.74). A model combining MRE with AST improved the diagnosis of F2–4 (AUROC = 0.83). The ICC for repeatability was 0.94 and 0.99 for MRE and PDFF, respectively. Data Conclusion MRE can potentially diagnose NASH and differentiate between fibrosis stages. Combining MRE with PDFF improves the diagnosis of NASH. Level of Evidence 2 Technical Efficacy Stage 2

PURPOSE:An approach for the automated segmentation of visceral adipose tissue (VAT) and subcutaneous adipose tissue (SAT) in multicenter water-fat MRI scans of the abdomen was investigated, using 2 different neural network architectures.METHODS:The 2 fully convolutional network architectures U-Net and V-Net were trained, evaluated, and compared using the water-fat MRI data. Data of the study Tellus with 90 scans from a single center was used for a 10-fold cross-validation in which the most successful configuration for both networks was determined. These configurations were then tested on 20 scans of the multicenter study beta-cell function in JUvenile Diabetes and Obesity (BetaJudo), which involved a different study population and scanning device.RESULTS:The U-Net outperformed the used implementation of the V-Net in both cross-validation and testing. In cross-validation, the U-Net reached average dice scores of 0.988 (VAT) and 0.992 (SAT). The average of the absolute quantification errors amount to 0.67% (VAT) and 0.39% (SAT). On the multicenter test data, the U-Net performs only slightly worse, with average dice scores of 0.970 (VAT) and 0.987 (SAT) and quantification errors of 2.80% (VAT) and 1.65% (SAT).CONCLUSION:The segmentations generated by the U-Net allow for reliable quantification and could therefore be viable for high-quality automated measurements of VAT and SAT in large-scale studies with minimal need for human intervention. The high performance on the multicenter test data furthermore shows the robustness of this approach for data of different patient demographics and imaging centers, as long as a consistent imaging protocol is used.

Computed Tomography (CT) allows detailed studies of body composition and its association with metabolic and cardiovascular disease. The purpose of this work was to develop and validate automated and manual image processing techniques for detailed and efficient analysis of body composition from CT data. The study comprised 107 subjects examined in the Swedish CArdioPulmonary BioImage Study (SCAPIS) using a 3-slice CT protocol covering liver, abdomen, and thighs. Algorithms were developed for automated assessment of liver attenuation, visceral (VAT) and subcutaneous (SAT) abdominal adipose tissue, thigh muscles, subcutaneous, subfascial (SFAT) and intermuscular adipose tissue. These were validated using manual reference measurements. SFAT was studied in selected subjects were the fascia lata could be visually identified (approx. 5%). In addition, precision of manual measurements of intra- (IPAT) and retroperitoneal adipose tissue (RPAT) and deep- and superficial SAT was evaluated using repeated measurements. Automated measurements correlated strongly to manual reference measurements. The SFAT depot showed the weakest correlation (r = 0.744). Automated VAT and SAT measurements were slightly, but significantly overestimated (≤4.6%, p ≤ 0.001). Manual segmentation of abdominal sub-depots showed high repeatability (CV ≤ 8.1%, r ≥ 0.930). We conclude that the low dose CT-scanning and automated analysis makes the setup suitable for large-scale studies.

Magnetic Resonance Imaging(MRI) has developed as a widespread technique to examine various body parts and diagnose a wide range of diseases. MRI can often be superior to other imaging techniques su ...