Yuvnik Trada^1,2, Paul Keall², Michael Jameson³, Daniel Moses^4,5, Peter Lin^6,7, Phillip Clap^5,8, Lois Holloway^8,9, Myo Min^10,11, Dion Forstner³, Allan Fowler⁹, and Mark Lee^5,9
1Radiation Oncology, Calvary Mater Newcastle, Waratah, Australia, ²The University of Sydney, Sydney, Australia, ³GenesisCare St Vincents Hospital, Sydney, Australia, ⁴Department of Medical Imaging, Prince of Wales Hospital, Sydney, Australia, ⁵University of New South Wales, Sydney, Australia, ⁶Department of Nuclear Medicine and PET, Sydney, Australia, ⁷Western Sydney University, Sydney, Australia, ⁸Ingham Institute of Applied Medical Research, Liverpool, Australia, ⁹Liverpool Hospital, Liverpool, Australia, ¹⁰Sunshine Coast University Hospital, Sunshine Coast, Australia, ¹¹University of Sunshine Coast, Britinya, Australia

In head and cancer there is a need for a predictive biomarker for patients undergoing radiotherapy(RT) to allow personalisation of treatment.55 patients with HNSCC had DWI imaging performed at baseline, during (week 2, 3, 5 and 6) and post RT (1 and 3 months). Absolute and change in primary tumour ADCmean was correlated to local recurrence.Week 3 was the optimal timepoint for mid-treatment response assessment. Change in week 3 ADCmean of <24.4% predicted local recurrence; corresponding 2 year LRFS of 97% vs 42%.Changes in mid-treatment DWI imaging could be utilised in design of future adaptive clinical trials.

Lixuan Huang¹, Zongxiang Yang¹, Hao Ren², Cheng Tang¹, Huiting Zhang³, Yang Song³, and Liling Long^1
1The First Affliated Hospital of Guangxi Medical University, Nanning, China, ²Guangxi Medical University Kaiyuan Langdong Hospital, Nanning, China, ³Siemens Healthineers, Shanghai, China

This study is to investigate a MRI radiomics model based on pre-treatment texture features to explore its application value in predicting the efficacy of recombinant human endostatin(RHES) combined concurrent chemoradiotherapy for nasopharyngeal carcinoma. The results showed that the AUC and the accuracy   of radiomics-clinics combine model and radiomics model for response assessment in nasopharyngeal carcinoma(NPC) was higher than clinics model. These suggest that pretreatment MRI based radiomics could predict RHES combined concurrent chemoradiotherapy response in NPC, its accuracy was better than clinics model.

Yifan Yuan¹, Qi Yue¹, Xiang Zou¹, Yu Guo¹, Ying-Hua Chu², Yi-Cheng Hsu², Patrick Liebig³, Wenwen Yu⁴, He Wang^4,5, Liang Chen¹, and Ying Mao^1
1Department of Neurosurgery, Huashan Hospital, Fudan University, Shanghai, China, ²Siemens Healthineers Ltd., Shanghai, China, ³Siemens Healthcare GmbH, Erlangen, Germany, ⁴Institute of Science and Technology for Brain-Inspired Intelligence, Fudan University, Shanghai, China, ⁵Human Phenome Institute, Fudan University, Shanghai, China

Magnetic resonance imaging has been applied in intraoperative neuro-navigation and irradiation guidance in glioma diagnosis. The contrast-enhanced MRI or peritumor edema on T2w image is not precise enough to delineate tumor infiltration. MR spectroscopy (MRS) and chemical exchange saturation transfer (CEST) have been used to evaluate tumor metabolism. Studies showed that amino acid PET surpassed conventional MR imaging for tumor delineation. Here, we explored the correlation between MRS and CEST at 7T and chose PET as gold standard to investigate multimodal MR imaging for glioma metabolic characteristics and tumor extent. An exemplary case with stereotactic biopsy proof was shown.

Junjie Wen¹, Hongxi Zhang², Zhipeng Shen³, Xiaohui Ma², Xinchun Chen², Weibo Chen⁴, Dan Wu¹, and Yi Zhang^1
1Key Laboratory for Biomedical Engineering of Ministry of Education, Department of Biomedical Engineering, College of Biomedical Engineering & Instrument Science, Zhejiang University, Hang Zhou, Zhejiang, China, ²Department of Radiology, Children’s Hospital, Zhejiang University School of Medicine, Hang Zhou, Zhejiang, China, ³Department of Neurosurgery, Children’s Hospital, Zhejiang University School of Medicine, Hang Zhou, Zhejiang, China, ⁴Philips Healthcare, Shanghai, China

APT imaging and its derived metric maps were applied to identify molecular subgroups of medulloblastoma for the first time. Thirty-eight newly-diagnosed pediatric patients with medulloblastoma were enrolled in this study and scanned on a 3T scanner. We implemented a radiomic analysis of the APT-related metric maps, with initial regions of interest delineated by an experienced radiologist and then shrunk automatically. After feature extraction and selection, five different classifiers were tested with both single-metric and multi-metric maps. We successfully established predictive models to differentiate the subgroups of medulloblastoma with good accuracy of 0.754.

Jin Li¹, Xiangrong Li¹, and Huiting Zhang^2
1The First Affiliated Hospital of Guangxi Medical University, Nanning, China, ²MR Scientific Marketing, Siemens Healthineers, Wuhan, China

We explored the applied value of multi-mode MR imaging technology, including diffusion kurtosis imaging (DKI), arterial spin labeling (ASL) and ¹H-MRS, in glioma tumor invasion boundary. Results showed mean kurtosis from DKI and apparent diffusion coefficient had the highest diagnostic efficiency between high-grade (HGG) and low-grade gliomas (LGG). There were significant differences in CHO/Cr and NAA/Cr from ¹H-MRS between the solid area and the proximal peritumoral edema area (PEA), and CHO/Cr and NAA/Cr in different PEA of LGG or HGG. PEA within 1cm around the solid area is recommended to be resected to decrease recurrence rate.

ChengLong Deng^1,2, BingChao Wu^1,2, QingJun Wang³, QingLei Shi⁴, Bei Guan^1,2, Dacheng Qu⁵, ChenXi Li^1,2, DaoGuang Zan^1,2, XiaoLin Chen^1,2, and YongJi Wang^1,2
1Collaborative Innovation Center, Institute of Software Chinese Academy of Sciences, Beijing, China, ²University of Chinese Academy of Sciences, Beijing, China, ³Department of Radiology, The 6th Medical Center of Chinese PLA General Hospital, Beijing, China, ⁴MR Scientific Marketing, Siemens Healthcare, Beijing, China, ⁵School of Computer Science and Technology, Beijing Institute of Technology, Beijing, China

In this paper, a multi-stage ensemble learning based on the majority voting mechanism was designed to leverage the contradiction between an insufficient number of thyroid MRI and well-trained deep learning models that accurately predicted the pathology of thyroid micronodules. And its clinical applicability value was also assessed in terms of micronodule risk stratification and optimal regimen selection on high b-value (2000 s/mm²) diffusion-weighted images. Experimental results proved that our model had the capability of effectively distinguishing benign and malignant micronodules on small-dataset thyroid DWI images.

Minxiong Zhou¹, Huiting Zhang², Ankang Gao³, Shaoyu Wang², Jie Bai³, Guang Yang⁴, Jingliang Cheng³, and Xu Yan^2
1Shanghai University of Medicine & Health Sciences, Shanghai, China, ²MR Scientific Marketing, Siemens Healthineers, Shanghai, China, ³Department of MRI, The First Affiliated Hospital of Zhengzhou University, Zhengzhou, China, ⁴Shanghai Key Laboratory of Magnetic Resonance, School of Physics and Electronic Science, Shanghai Key Laboratory of Magnetic Resonance, School of Physics and Electronic Science,East China Normal University, Shanghai, China

There are many advanced diffusion models available for neural and body researches, while their acquisitions are different and  normally difficult to be joint applied and compared in one study. In this study, we tried to calculate 7 diffusion models, including 4 neural and 3 cancer related models, using data from one acquisition, and applied it for glioma. Two diffusion acquisition strategy were evaluated, and the performance of these models in glioma grading was also compared. The result showed that both acquisition strategies could generate high quality quantitative parameters for all models, which showed significant differences between high- and low-grade tumor.

Bin Bo¹, Tianyao Wang², Ziyu Meng¹, Rong Guo^3,4, Yudu Li^3,4, Yibo Zhao^3,4, Xin Yu⁵, Zhi-Pei Liang^3,4, and Yao Li^1
1School of Biomedical Engineering, Shanghai Jiao Tong University, Shanghai, China, ²Radiology Department, The Fifth People's Hospital of Shanghai, Fudan University, Shanghai, China, ³Beckman Institute for Advanced Science and Technology, University of Illinois at Urbana-Champaign, Urbana, IL, United States, ⁴Department of Electrical and Computer Engineering, University of Illinois at Urbana-Champaign, Urbana, IL, United States, ⁵Department of Biomedical Engineering, Case Western Reserve University, Cleveland, OH, United States

Conventional structural MRI has limited specificity in defining the extent and grade of glioma. MRSI complements MRI in tumor tissue characterization with neurometabolic fingerprints. In this study, we investigated the use of high-resolution 3D MRSI for glioma grading and evaluation of patient overall survival. Our results showed that the neurometabolic biomarkers could differentiate low-grade from high-grade gliomas and provide prognostic value for overall survival of newly diagnosed glioma patients.

Ying-Hua Chu¹, Yifan Yuan², Xueying Zhao³, He Wang^3,4, Thorsten Feiweier⁵, and Yi-Cheng Hsu^1
1Siemens Healthineers Ltd., Shanghai, China, ²Department of Neurosurgery, Huashan Hospital, Fudan University, Shanghai, China, ³Institute of Science and Technology for Brain-Inspired Intelligence, Fudan University, Shanghai, China, ⁴Human Phenome Institute, Fudan University, Shanghai, China, ⁵Siemens Healthcare GmbH, Erlangen, Germany

We used MRI-cytometry to estimate the mean cell diameter and intracellular volume of tumors of the meningioma subtype classification. We found cell size and intracellular volume to be very different between the three meningioma subtypes in this preliminary study. Our results reflected high collagen volume among the extracellular space in fibrous meningioma. We plan to scan more patients in the future to find a useful biomarker using MRI-cytometry.

Yifan Liu¹, Shasha Bao¹, Nan Xu¹, Xiaobin Guo¹, Chengde Liao¹, Zhongping Zhang², and Tengfei Ke^1
1Dept. of Radiology, Yunnan Cancer Hospital, Kunming, China, ²Philips Healthcare China, Guangzhou, China

We aimed to differentiate between T1 stage nasopharyngeal carcinoma (NPC) and nasopharyngeal lymphoid hyperplasia using multiple b-values Turbo-spin-echo DWI acquisition. Turbo-spin-echo DWI were adopted to minimize susceptibility artifacts in nasopharynx. Mono-exponential and bi-exponential models were both used to generate ADC, D, D* and f maps. We found that T1 stage NPC had significant lower ADC and D, higher D* than nasopharyngeal lymphoid hyperplasia. Furthermore, D has best diagnostic efficacy with highest area under curve in ROC analysis, indicating that it's a potential imaging biomarker for accurate early diagnosis of T1 stage NPC.

Chao Luo¹, Haixia Li², Kan Deng², and Haojiang Li^1
1Sun Yat-sen University Cancer Center, Guangzhou, China, ²Philips Healthcare, Guangzhou, China

This study aimed to identify magnetic resonance (MR) imaging markers associated with the overall survival (OS) of patients with nasopharyngeal carcinoma (NPC) and establish a random survival forest (RSF) model, which is a time-related machine learning model for survival analysis, to predict their survival.

Wenguang Liu¹, Yigang Pei¹, Simin Xie¹, Xiao Wang¹, Yu Bai¹, Gaofeng Zhou¹, Weiyin Vivian Liu², and Wenzheng Li^1
1Department of Radiology, XiangYa Hospital, Central South University, Changsha, China, ²GE Healthcare, Beijing, China

To predict the treatment response for NPC patients is very important. This study aimed to explore the diagnostic performance of 3D APT imaging on predicting the treatment response of NPC patients. A total of 40 patients and 14 healthy volunteers were identified in this study. Result show that the APT and MT values have good and excellent intra- and inter-observer agreement. It is significant difference for MT values between residual and non-residual group. The result suggests that the pre-treatment MT values can be used to predict treatment response of NPC with chemoradiotherapy, and better than the APT and ADC value.

Chenxi Liu¹, Shengzhong Wang², Xiaocheng Wei³, Gang Xiao¹, YUchuan Hu¹, and Guangbin Cui^1
1Tangdu Hospital, Air Force Medical University, Xi'an, China, ²Shaanxi University of Traditional Chinese Medicine, Xi'an, China, ³GE Healthcare, Beijing, China

The pathological staging system of Thymic epithelial tumors (TETs) which called Masaoka staging, is an important prognostic factor, and its impact on survival is much greater than that of WHO classification. However, as the Masaoka stage can only be established postoperatively, preoperative imaging has important implications concerning treatment strategy. This study analyzes the differences of T1 values and ADC values in different regions, different types and stages of TETs tumors, and establishes a reliable noninvasive imaging technology to distinguish the risk classification and stage of tumors.