Nan Wang^1,2, Srinivas Gaddam³, Lixia Wang^1,4, Yibin Xie¹, Zhaoyang Fan¹, Simon Lo³, Stephen Pandol³, Anthony G Christodoulou¹, and Debiao Li^1
1Biomedical Imaging Research Institute, Cedars-Sinai Medical Center, Los Angeles, CA, United States, ²Bioengineering, University of California, Los Angeles, Los Angeles, CA, United States, ³Division of Digestive and Liver Diseases, Cedars-Sinai Medical Center, Los Angeles, CA, United States, ⁴Chaoyang Hospital, Beijing, China

The differentiation of pancreatic ductal adenocarcinoma (PDAC) and chronic pancreatitis (CP) is crucial to the diagnosis and prognosis of PDAC. DCE MRI serves as a promising imaging tool, but still faces several technical challenges. In this work, we evaluated the characterization of PDAC versus CP using the proposed Multitasking DCE technique, which enables free-breathing acquisition, 3D whole-abdomen coverage, high temporal resolution at 500ms, and dynamic T1 mapping throughout all DCE phases. In vivo studies on 16 healthy volunteers, 14 PDAC patients, and 8 CP patients demonstrated the capability of Multitasking DCE in differentiating normal pancreas, PDAC, and CP.

Takayuki Masui¹, Motoyuki Katayama¹, Yuji Iwadate², Naoyuki Takei², Mitsuharu Miyoshi², Masako Sasaki¹, Takahiro Yamada¹, Ty Cashen³, Sagar Mandava⁴, and Kang Wang^5
1Radiology, Seirei Hamamatsu General Hospital, Hamamatsu, Japan, ²Global MR Applications and Workflow, GE Healthcare, Hino, Japan, ³Global MR Applications and Workflow, GE Healthcare, Madison, WI, United States, ⁴GE Healthcare, Tucson, AZ, United States, ⁵GE Healthcare, Waukesha, WI, United States

The feasibility of dynamic Gd-contrast study for evaluation of hepatopancreatobiliary lesions under free breathing was demonstrated. Superb image quality with high temporal resolutions could be obtained using a stack-of-stars k-space trajectory with golden angle ordering a CG-SENSE algorithm that supports parallel imaging and soft-gating for accelerated motion robust imaging. Selective recognition of vasculatures and lesions in the liver and pancreas can be made with this technique, which may be equivalent to fast breath-hold dynamic contrast image in young and old aged population.

Hanyu Jiang^1,2, Bin Song¹, and Mustafa Shadi Rifaat Bashir^2
1Department of Radiology, West China Hospital, Sichuan University, Chengdu, China, ²Department of Radiology, Duke University Medical Center, Durham, NC, United States

We aimed to develop a modified Liver Imaging Reporting and Data System (mLI-RADS) with comparisons against original LI-RADS version 2018 (v2018) for diagnosing hepatocellular carcinoma (HCC) on gadoxetate disodium-enhanced magnetic resonance imaging (EOB-MRI). 1002 hepatic observations in 272 consecutive at-risk patients were prospectively included. Ancillary features were assessed based on inter-rater agreement, prevalence, diagnostic accuracy, and added value to the major features. Compared with the original LI-RADS v2018, mLI-RADS demonstrated superior simplicity, sensitivity and accuracy without substantial loss of specificity; hence should be the preferred diagnostic criteria for HCC in high-risk patients on EOB-MRI.

Wu Zhou¹, Yunling Li¹, Hui Huang¹, Yaoqin Xie², Lijuan Zhang², and Guangyi Wang^3
1School of Medical Information Engineering, Guangzhou University of Chinese Medicine, Guangzhou, China, ²Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen, China, ³Department of Radiology, Guangdong General Hospital, Guangzhou, China

Deep feature derived from data-driven learning has consistently shown to outperform conventional texture features for lesion characterization. However, due to the slice thickness of medical imaging, through-plane has worse resolution than in-plane resolution. Therefore, the performance of deep feature extracted from the through plane slices may be worse, and their contributions to the final characterization may also be very limited. We propose an end-to-end super-resolution and self-attention framework based on generative adversarial network (GAN), in which the through-plane slices with low resolution are enhanced by learning the in-plane slices with high resolution to improve the performance of lesion characterization.

Paul Kennedy^1,2, Octavia Bane^1,2, Stefanie Hectors^1,2,3, Daniel Stocker^1,2, Bradley D Bolster Jr. ⁴, Scott Friedman⁵, Thomas Schiano⁶, Isabel M Fiel⁷, Swan Thung⁷, Aaron Fischman², and Bachir Taouli^1,2
1BioMedical Engineering and Imaging Institute, Icahn School of Medicine at Mount Sinai, New York, NY, United States, ²Department of Diagnostic, Molecular and Interventional Radiology, Icahn School of Medicine at Mount Sinai, New York, NY, United States, ³Department of Radiology, Weill Cornell Medicine, New York, NY, United States, ⁴Siemens Medical Solutions USA, Inc., Salt Lake City, UT, United States, ⁵Division of Liver Diseases, Icahn School of Medicine at Mount Sinai, New York, NY, United States, ⁶Department of Medicine, Icahn School of Medicine at Mount Sinai, New York, NY, United States, ⁷Department of Pathology, Icahn School of Medicine at Mount Sinai, New York, NY, United States

In this study we investigate the ability of MR elastography (MRE) and ultrasound shear wave elastography (SWE) to assess portal hypertension (PH) severity in patients with liver disease and hepatic venous pressure gradient (HVPG) measurement. 3D MRE spleen stiffness correlated with HVPG. 2D and 3D MRE of the spleen were significantly higher in patients with clinically significant PH (CSPH, HVPG>10mmHg) than those with no PH/PH (HVPG>5mmHg). 3D MRE spleen stiffness was significantly elevated in PH/CSPH patients compared to those with no PH and was an excellent predictor of CSPH. MRE spleen stiffness appears sensitive to hemodynamic changes associated with PH.

Shu-Fu Shih^1,2, Tess Armstrong¹, Sevgi Gokce Kafali^1,2, Xiaodong Zhong³, Kara L. Calkins⁴, and Holden H. Wu^1,2
1Radiological Sciences, University of California, Los Angeles, Los Angeles, CA, United States, ²Bioengineering, University of California, Los Angeles, Los Angeles, CA, United States, ³Siemens Healthcare, Los Angeles, CA, United States, ⁴Pediatrics, University of California, Los Angeles, Los Angeles, CA, United States

MRI quantification of hepatic proton-density fat fraction (PDFF) and R₂* enables non-invasive diagnosis and staging of non-alcoholic fatty liver disease (NAFLD) and liver iron overload, respectively. A recent 3D stack-of-radial technique enables free-breathing quantification, but requires scans of 2-4 minutes and motion may affect R₂* accuracy. In this work, we propose an improved free-breathing stack-of-radial technique that combines soft-gating and a sparsity-promoting tensor reconstruction to compensate for motion effects and accelerate the scan to 31 seconds. Data from adult and pediatric NAFLD patients demonstrate good agreement of PDFF and R₂* between the proposed method and the conventional breath-held Cartesian scan.

Zheng Ye¹, Yi Wei¹, Jie Chen¹, and Bin Song^1
1West China Hospital, Sichuan University, Chengdu, China

Hepatocellular carcinoma (HCC) with highly aggressive characteristics is usually actively proliferated and easily relapse, thereby requiring adjuvant therapies before surgery, like preoperative TACE, to improve the patients’ prognosis. Ki-67 labeling index (LI) was reported to be highly correlated with aggressive propensity of HCC, and thus could affect the treatment response of the tumor and prognosis directly. Although most of HCC presented hypointensity on hepatobiliary phase (HBP), the absolute signal intensity and relative contrast enhancement ratio are not the same. In this study, we prospectively investigate the usefulness of HBP quantitative parameters for preoperative prediction of aggressiveness in HCC patients.

Hailong Li^1,2, Lili He^1,2,3, Jonathan Dudley^2,4, Thomas Maloney^2,4, Elanchezhian Somasundaram⁴, Samuel L. Brady^4,5, Nehal A. Parikh ^1,3, and Jonathan R. Dillman^2,4,5
1The Perinatal Institute and Section of Neonatology, Perinatal and Pulmonary Biology, Cincinnati Children's Hospital Medical Center, Cincinnati, OH, United States, ²Imaging Research Center, Cincinnati Children's Hospital Medical Center, Cincinnati, OH, United States, ³Department of Pediatrics, University of Cincinnati College of Medicine, Cincinnati, OH, United States, ⁴Department of Radiology, Cincinnati Children's Hospital Medical Center, Cincinnati, OH, United States, ⁵Department of Radiology, University of Cincinnati College of Medicine, Cincinnati, OH, United States

Detection and monitoring of chronic liver diseases is typically assessed using a combination of clinical history, physical examination, laboratory testing, biopsy with histopathologic assessment, and imaging. The aim of this study is to develop a deep transfer learning model (DeepLiverNet) to categorically classify the severity of liver stiffening (no/mild vs. moderate/severe) using both anatomic T2-weighted MR images and clinical data. The DeepLiverNet model achieved accuracies of 88.0% and 80.0% on the risk stratification of liver stiffness in internal and external validation datasets, respectively. This demonstrates that a deep learning model may provide a means for stratifying liver stiffness without elastography.

Justin Yu¹, Anshuman Panda¹, Kelly Tung-Smith¹, Robert Nelson², Akira Kawashima¹, Ming Yang¹, Chen Lin³, Aiden McGirr¹, Sophia Fasani¹, Kristina Flicek¹, Sukhdeep Singh¹, and Alvin Silva^1
1Radiology, Mayo Clinic Arizona, Phoenix, AZ, United States, ²National Institutes of Health, Phoenix, AZ, United States, ³Mayo Clinic Florida, Jacksonville, FL, United States

The variability of stiffness data from GRE and SE EPI MRE sequences is tested on phantoms and in-vivo on 1.5T and 3T MRI scanners from two vendors. Large variability was observed in the phantom measurements for the GRE and EPI sequences on one of the vendor’s scanner, ranging from 20-33% difference. Stiffness measurements were very similar (within 10%) between sequences on the other vendor’s scanner. Similar results were found in several clinical subjects who had GRE and SE EPI sequences performed.

Rencheng Zheng¹, Tian Qiu², Nannan Shi², Yuxin Shi², Weibo Chen³, Chengyan Wang⁴, and He Wang^4,5
1Fudan University, Shanghai, China, ²ShanghaiPublic Health Clinical Center, Shanghai, China, ³Philips Healthcare, Shanghai, China, ⁴Human Phenome Institute, Fudan University, Shanghai, China, ⁵Institute of Science and Technology for Brain-Inspired Intelligence, Fudan University, Shanghai, China

This study proposed an automatic hepatic fibrosis staging model based on transfer learning segmentation and radiomics analysis for hepatitis B patients. The automatic liver ROI extarction Time dimension features of multi DCE phases were included in feature set which played an important role in the classification. The proposed model exhibited a superior performance in significant fibrosis, advanced fibrosis and cirrhosis classification.

Jitka Starekova¹, Ruiyang Zhao^1,2, Timothy J Colgan¹, Kevin M Johnson^1,2, Jennifer L Rehm³, Scott B Reeder^1,2,4,5,6, and Diego Hernando^1,2
1Department of Radiology, University of Wisconsin, Madison, Madison, WI, United States, ²Department of Medical Physics, University of Wisconsin, Madison, Madison, WI, United States, ³Department of Pediatrics, University of Wisconsin, Madison, Madison, WI, United States, ⁴Department of Biomedical Engineering, University of Wisconsin, Madison, Madison, WI, United States, ⁵Department of Medicine, University of Wisconsin, Madison, Madison, WI, United States, ⁶Department of Emergency Medicine, University of Wisconsin, Madison, Madison, WI, United States

Chemical shift-encoded (CSE)-MRI enables accurate and precise quantification of proton density fat-fraction (PDFF) in the liver. Widely used 3D multi-echo spoiled gradient echo (SGRE) CSE-MRI requires reliable breath-holding to avoid motion-related artifacts. This is a major limitation for children, the elderly, and sick patients. Free-breathing 2D sequential CSE-MRI is motion-robust, however, suffers from low signal-to-noise-ratio (SNR). To overcome these limitations, we combined variable flip angle (VFA) 2D acquisitions and nonlocal means (NLM) motion-corrected averaging. In this prospective study, free-breathing multi-repetition VFA-NLM demonstrated high SNR and reduced artifacts compared to the conventional 3D-SGRE, while preserving accuracy of PDFF quantification.

Yurui Qian¹, Jian Hou¹, Yixiang Wang¹, Vincent Wong², Queenie Chan³, Weibo Chen⁴, Min Deng¹, Franklin Au¹, Anthony Chan⁵, Winnie Chu¹, and Weitian Chen^1
1Department of Imaging and Interventional Radiology, Chinese University of Hong Kong, Hong Kong, Hong Kong, ²Department of Medicine & Therapeutics, Chinese University of Hong Kong, Hong Kong, Hong Kong, ³Philips Healthcare, Hong Kong, Hong Kong, ⁴Philips Healthcare, Shanghai, China, ⁵Department of Anatomical and Cellular Pathology, Chinese University of Hong Kong, Hong Kong, Hong Kong

MRI is widely used as a non-invasive method to diagnose and monitor liver diseases. For certain quantitative MRI techniques, liver iron content may affect the measurement. In this work, we investigated the influence of liver iron content on several quantitative MRI methods, including macromolecular proton fraction, T1rho and intravoxel incoherent motion.

Huanhuan Chong¹, Li Yang¹, Yangli Yu¹, Dijia Wu², Chun Yang¹, and Mengsu Zeng^1
1Department of Radiology, Shanghai Institute of Medical Imaging, Zhongshan Hospital of Fudan University, Shanghai, China, ²Shanghai United Imaging Intelligence Co., Ltd, Shanghai, China

This study aims to construct a preoperative MVI prediction model in solitary HCC derived from gadoxetic acid-enhanced magnetic resonance imaging, and to further investigate its latent association with clinical indexes, imaging features and radiomics signatures based on diverse sequences and volumetric interests (VOIs) of tumor. The conclusion indicated that serum α-fetoprotein, total bilirubin, higher value of tumor margin smoothness (prefer to non-smooth margin), non-intact capsule enhancement and peritumoral enhancement are independent and significant predictors for MVI, and the final nomogram incorporating clinical, imaging and the optimal radiomics model based on VOI_entire_5mm_10mm_liver achieves satisfactory prediction for MVI in solitary HCC.

Mao-Tong LIU¹, Xue-Qin ZHANG¹, Jian LU¹, and Wei-Bo CHEN^2
1Third Affiliated Hospital of Nantong University & Nantong Third People's Hospital, Nan Tong, China, ²Philips Healthcare Shanghai, China, Shang Hai, China

The purpose of this study was to identify whether hepatocyte fraction that based on Gadoxetic Acid–enhanced MRI is useful for the assessment of liver function. Firstly, T1 mapping imaging was performed before and 20 minutes after Gd-EOB-DTPA administration, The following parameters are then obtained from the images: pre- and postcontrast T1 values of the liver (T1pre and T1post), increase in the T1 relaxation rate (Δ R1), rate of the decrease of the T1 relaxation time (Δ T1), hepatocyte fraction (HeF), and uptake coefficient (K). Our study showed that hepatocyte fraction can be used to evaluate the liver function of patients with hepatitis B cirrhosis, K value has high diagnostic efficiency.