Ting Duan¹, Zhen Zhang², Hanyu Jiang², Yidi Chen², Xiaoyong Zhang³, and Bin Song^2
1Radiology, West China Hospital, Sichuan University, Chengdu, China, ²West China Hospital, Sichuan University, Chengdu, China, ³Clinical Science, Philips Healthcare, Chengdu, China, Chengdu, China

Compressed sensing AI-based DWI substantially reduced noise artifacts and improved the signal-to-noise ratio and lesion contrast-to-noise ratio compared with conventional DWI and Compressed sensing based DWI, without any penalty for scan parameters. This technique may prove of value in better diagnostic liver imaging.

Cao Peng¹, Zuojun Wang¹, Chenyang Liu², Tian Li², Edward S. Hui³, and Jing Cai^2
11. Department of Diagnostic Radiology, The University of Hong Kong, Hong Kong, Hong Kong, ²3. Department of Health Technology & Informatics, The Hong Kong Polytechnic University, Hong Kong, Hong Kong, ³The Hong Kong Polytechnic University, Kowloon, Hong Kong

We developed a motion-resolved and free-breathing liver magnetic resonance fingerprinting (MRF) protocol. The deformation maps were obtained from the first singular image of MRF data. The reconstruction method enforced the consistency of the MRF data with the deformation maps by adding the deformation maps to the encoding matrix. MRF was tested on four healthy volunteers. We demonstrated a motion-resolved MRF with a nominal frame rate of 2.5 Hz for free-breathing liver imaging. 

Qingjia Bao¹, Feng Pan², Chongxin Bai³, Kewen Liu³, Zhao Li¹, Peng Sun⁴, Jiazheng Wang⁴, Linkun Zhong⁵, Aodong Xiao⁶, Lian Yang², and Chaoyang Liu^1
1State Key Laboratory of Magnetic Resonance and Atomic and Molecular Physics, Wuhan Institute of Phys, Wuhan, China, ²Department of Radiology, Union Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan, China, ³Wuhan University of Technology School of Information Engineering, Wuhan, China, ⁴Philips Healthcare, Beijing, China, ⁵Wuhan University of Arts and Science, Wuhan, China, ⁶Henan University of Science and Technology, Luoyang, China

We proposed a dual-domain self-supervised motion artifacts disentanglement network (DSMAD-Net) for the liver's gadoxetic acid-enhanced arterial phase images. The motion correction is converted to the image-to-image translation problem by assuming that motion-free images and motion-corrupted images belong to different domains. Specifically, image-to-image translation within the same domain is designed to constrain auto-encoders to learn the feature representation by utilizing the input images as supervision information. Moreover, the cross-domain translation explores the cycle consistency in the absence of paired motion-free and motion-corrupted images. Experimental results demonstrate that our method remarkably removes artifacts in the gadoxetic acid-enhanced arterial phase images.

Bei Liu¹, Huajun She¹, Yufei Zhang¹, Zekang Ding¹, Zhijun Wang¹, and Yiping P. Du^1
1School of Biomedical Engineering, Shanghai Jiao Tong University, Shanghai, China

We propose an algorithm for dMRI reconstruction from highly under-sampled k-space data acquired during free breathing. Stack-of-star GRE radial sequence with self-navigator is used to acquire the data. We explore spatial and temporal redundancy for the reconstruction by using weighted group sparsity, weighted sparsity, and low-rank tensor. Additionally, a tensor total variation is used to ensure spatial and temporal smoothness. By applying a weighting function to the sparsity and group sparsity, the subtle structural sparsity features in the sparse domain can be better utilized. The proposed algorithm has the potential to be used in clinical applications such as MR-guided surgery.

Mengsi Li¹, Yanjie Hong², Zhixuan Yu², Dandan Zheng², and Jin Wang^1
1Department of Radiology, The Third Affiliated Hospital, Sun Yat-Sen University, Guangzhou, China, ²Shukun (Beijing) Technology Co., Ltd, Beijing, China

The accurate segmentation of liver vascular structure is one of the key components of automated  radiological diagnosis. Unfortunately, accurate vessel segmentation in clinical practice usually relies on the manual delineation by radiologists on each slice, which is extremely tedious and time-consuming. We developed and evaluated a new liver vessel segmentation and reconstruction workflow, which consist of a 3D Res-Net, a 2D Dense-Net and a 3D U-net model, allows for automated extraction of the portal vein, hepatic vein and inferior vena cava on 3D contrast enhanced portal venous phase MR images.

Christian Simonsson^1,2,3, Wolf Claus Bartholomä^2,3, Anna Lindhoff Larsson⁴, Markus Karlsson², Bengt Norén², Gunnar Cedersund^1,3, Per Sandström⁴, Nils Dahlström^2,3, and Peter Lundberg^2,3
1Department of Medical Engineering, Linköping University, Linköping, Sweden, ²Departments of Radiation Physics, Radiology, Department of Medical and Health Sciences, Linköping University, Linköping, Sweden, ³Center for Medical Image Science and Visualization (CMIV), Linköping University, Linköping, Sweden, ⁴Department of Surgery, Department of biomedical and clinical sciences, Linköping University, Linköping, Sweden

In some cases, treatment of severe liver diseases requires resective surgery. This brings serious complications if the remnant tissue fails to match the requirement of liver function. Therefore, a pre-operative risk assessment is vital. However, usually the assessment only investigates global liver function. For a more precise assessment, we investigate the possibility of using DCE-MRI in combination with pharmacokinetic modeling to quantify both global- and regional liver function. Also, we show a novel eight-compartment hepatic model, capable of performing an in-silico resection. This framework might in the future be used as a tool for a more precise pre-operative assessment.

Shuangshuang Xie¹, Caixin Qiu¹, Yajie Sun¹, Jinxia Zhu², Robert Grimm³, and Wen Shen^1
1Tianjin First Central Hospital, Tianjin, China, ²MR Collaboration, Siemens Healthcare Ltd., Beijing, China, ³MR Application development, Siemens Healthcare GmbH, Erlangen, Germany

This study investigated the feasibility of IVIM to evaluate fibrotic liver regeneration and function recovery after partial hepatectomy in a rat model. In fibrotic rats, significant increase in D values and decrease in D* and PF values followed with recovery to pre-PH levels were shown. PF values correlated well with Ki-67 indices, D* and PF values correlated well with ALT, AST, and TBil values. IVIM may be an noninvasive and promising approach to monitor liver regeneration and functional recovery after PH.

Jie Yuan¹, Wenli Tan¹, Songhua Zhan¹, Zhigang Gong¹, and Yongming Dai^2
1Shuguang Hospital Affiliated to Shanghai University of Traditional Chinese Medicine, Shanghai, China, ²United Imaging Healthcare, Shanghai, China

DWI can indirectly reflect the changes of tissue microstructure by detecting the direction and extent of water molecule diffusion. In liver fibrosis, collagen fibers and cell membranes may block the diffusion of water molecules, resulting in changes in the diffusion signal. B-value selection is closely related to the image quality of DWI. Previous studies have found the correlation between tissue water diffusivity and tissue elasticity in the liver. However, there are no widely accepted ADC values on the b-value combination for the stiffness of liver fibrosis. Therefore, the study of b-value combination optimization is of important clinical interest.

Christian Simonsson^1,2,3, Markus Karlsson^2,3, Patrik Nasr^3,4, Ralph Sinkus ^5,6, Simone Ignatova⁷, Nils Dahlström^2,3, Mattias Ekstedt^3,4, Stergios Kechagias^3,4, and Peter Lundberg^2,3
1Department of Biomedical Engineering (IMT), Linköping University, Linköping, Sweden, ²Department of Radiation Physics, Radiology, Department of Medical and Health Sciences, Linköping University, Linköping, Sweden, ³Center for Medical Image Science and Visualization (CMIV), Linköping University, Linköping, Sweden, ⁴Department of Gastroenterology and Hepatology, Department of Health, Medicine and Caring Sciences, Linköping University, Linköping, Sweden, ⁵Imaging Sciences & Biomedical Engineering, Kings College, London, United Kingdom, ⁶LVTS, INSERM U1148, Paris, France, ⁷Department of Clinical Pathology and Clinical Genetics, Department of Biomedical and Clinical Sciences, Linköping University, Linköping, Sweden

Several MR-biomarkers characterizing different aspects of NAFLD progression, especially for the early onset, and late stages, are available. However, there is a need to combine and characterize, not only the endpoints but also the intermediate stages of the disease, in particular the early stages of hepatic inflammation. Using an extensive multi-modal MRI protocol, we have investigated correlations between different MR-biomarkers with histopathology measurements of hepatic inflammation, fat accumulation and fibrosis. The long-term aim of this research is to characterize each patient in the early to intermediate stages of disease progression, with a higher degree of precision.

Yexin He¹, Jia Shao¹, Jinxia Guo², Qiang Gao¹, and Cheng Xu^1
1SHANXI PROVINCIAL PEOPLE'S HOSPITAL, Taiyuan, China, ²GE Healthcare, Beijing, China

Gd-EOB-DTPA-enhanced MRI was used to quantitatively assess liver function benefiting the follow-up of patients with chronic liver disease and the appropriate plan for liver cancer treatment. A systematic multi-parametric analysis with non-enhanced and Gd-EOB-DTPA-enhanced T1 weighted and T1 mapping MRI were conducted to assess the liver function in patients with cirrhosis. Independent factor for differentiation of A+N(normal and Child-Pugh A) and B+C (Child-Pugh B and C) were found to be RE_Liver, T1_Native, T1_Post and ΔT1_Liver, while for differentiation of Child-Pugh B and C were found to be ΔT1_Liver and SIR_Bdult/Muscle. Both combined model with these factors obtained better diagnostic performance.

Jiachen Ji¹, Hanyu Wei¹, Yunduo Li¹, Mengmeng Zhang², Changchun Liu², Bo Jin³, Xiaolong Qi⁴, Wen Shen⁵, Jianming Cai², and Rui Li^1
1Biomedical Engineering, Center for Biomedical Imaging Research, Tsinghua Univ., Beijing, China, ²Radiology, Department of Radiology, The Fifth Medical Centre of Chinese PLA General Hospital, Beijing, China, ³First Liver Cirrhosis Diagnosis and Treatment Center, The Fifth Medical Center of Chinese PLA General Hospital, Beijing, China, ⁴CHESS Center, Institute of Portal Hypertension, The First Hospital of Lanzhou University, Lanzhou, China, ⁵Department of Radiology, First Center Hospital of Tianjin, Tianjin, China

Chronic liver disease patients commonly suffer from portal hypertension (PH), which could lead to gastroesophageal varices. Noninvasive methods to diagnose PH and to detect high risk varices are urgently needed. 4D Flow MRI is an advanced technique which could provide visualization and quantification of blood flow. In the study, we observed significant hemodynamic differences in portal vein between PH patients and controls. Similar differences were observed between patients with high risk and low risk varices. The performance of diagnosing PH/detecting high risk varices using the single hemodynamic parameter and the logistic regression model was acknowledged by the ROCs.

Fei Xing¹ and Xiance Zhao^2
1the Third People’s Hospital of Nantong, Nantong, China, ²Philips Healthcare, Shanghai, China

The purpose of this study was to evaluate the category modifications and prognosis of LI-RADS  2,2 and 4 observations in cirrhosis on MRI. We retrospectively analyzed 105 LR-2 , 97 LR-3, and 43 LR-4 observations following-up datas. Follow-up duration and category modifications that progressed, remained stable, or decreased were recorded for each observation. Our study showed that  LR-2, LR-3 and LR-4 lesions in liver cirrhosis patients present with different natural disease courses and outcomes, the lesions with a higher LI-RADS category have a higher cumulative incidence of progression to LR-5/-TIV lesion, thus demonstrating an increasing risk of developing HCC.

Yanyan Xie¹, Lijuan Zhang², Guangyi Wang³, and Wu Zhou^1
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

The pathological grade and microvascular invasion (MVI) of hepatocellular carcinoma (HCC) are two key factors related to the patient's prognosis. Previous studies usually predict these two factors separately based on medical images. In this study, we propose an end-to-end multi-task deep learning network to simultaneously predict the MVI and grading information. Specifically, we are the first to demonstrate that these two tasks are related and can promote each other in the framework of multi-task deep learning. Experimental results of HCC in Contrast-enhanced MR demonstrate the effectiveness of the proposed method, outperforming the single task learning.

Jung Hee Son¹, Yedaun Lee¹, Ho-Joon Lee¹, Jisook Yi¹, Joonsung Lee², and Ersin Bayram^3
1Radiology, Heaundae Paik Hospital, Busan, Korea, Republic of, ²GE Healthcare, Seoul, Korea, Republic of, ³GE Healthcare, Houston, TX, United States

In this study, we evaluate image quality of high resolution contrast-enhanced T1-weighted MR enterography with deep learning reconstruction. Use of DLRecon for near-isotropic CE-T1WI MRE provides improved image quality, and increased SNR.

Juan Li¹, cheng jingliang¹, and Lin Liangjie ^2
1Department of MRI, The First Affiliated Hospital of Zhengzhou University, Zhengzhou, China, ²Philips Healthcare, Beijing, China

Rectal mucinous adenocarcinoma (MC) is a typical subtype of rectal adenocarcinomas, which has a poor prognosis and it is not sensitive to neoadjuvant chemoradiotherapy thus associated with significantly different individualized treatments as those for rectal common adenocarcinoma (AC). The amide proton transfer (APT) weighted and intravoxel incoherent motion (IVIM) imaging were investigated in this study for differential diagnosis between MC and AC. Results indicated that APT signal intensity and the D value by IVIM can be used in discriminating between MC and AC, slightly inferior to the performance of apparent diffusion coefficient by conventional diffusion weighted imaging.

Shohei Fujita^1,2, Katsuhiro Sano¹, Gastao Cruz³, Yuki Fukumura⁴, Hideo Kawasaki¹, Issei Fukunaga¹, Masami Yoneyama⁵, Koji Kamagata¹, Osamu Abe², Kenichi Ikejima⁶, Rene Botnar³, Claudia Prieto³, and Shigeki Aoki^1
1Department of Radiology, Juntendo University, Tokyo, Japan, ²Department of Radiology, The University of Tokyo, Tokyo, Japan, ³Department of Biomedical Engineering, School of Biomedical Engineering and Imaging Sciences, King’s College London, London, United Kingdom, ⁴Department of Human Pathology, Juntendo University, Tokyo, Japan, ⁵MR Clinical Science, Philips Japan, Tokyo, Japan, ⁶Department of Gastroenterology, Juntendo University, Tokyo, Japan

Magnetic resonance fingerprinting (MRF) has been proposed for simultaneous T1, T2, T2*, and fat-fraction mapping of the liver, which is suitable for the diagnosis and monitoring of liver diseases. We evaluated the repeatability of liver MRF measurements and correlated it with reference standard imaging in patients with diffuse liver disease. Furthermore, we validated MRF-derived measurements against histological grading of liver biopsies. Liver MRF provided repeatable multi-parametric maps with high agreement with measurements of separate conventional quantitative mapping as well as histological grading of liver biopsies. This approach may enable objective and non-invasive hepatic tissue characterization in a single breath-hold scan.