Reza Rahmanzadeh^1,2,3, Po-Jui Lu^1,2,3, Muhamed Barakovic^1,2,3, Matthias Weigel^1,2,3, Laura Gaetano⁴, Riccardo Galbusera^1,2,3, Thanh D. Nguyen⁵, Francesco La Rosa ^6,7, Daniel S. Reich⁸, Pascal Sati^8,9, Yi Wang⁵, Meritxell Bach Cuadra^6,7, Ernst-Wilhelm Radue^1,2, Jens Kuhle^1,3, Ludwig Kappos^1,3, Stefano Magon¹⁰, and Cristina Granziera^1,2,3
1Neurologic Clinic and Policlinic, Departments of Medicine, Clinical Research and Biomedical Engineering, University Hospital Basel and University of Basel, Basel, Switzerland, ²Translational Imaging in Neurology (ThINk) Basel, Department of Biomedical Engineering, University Hospital Basel and University of Basel, Basel, Switzerland, ³Research Center for Clinical Neuroimmunology and Neuroscience (RC2NB) Basel, University Hospital Basel and University of Basel, Basel, Switzerland, ⁴Hoffmann-La Roche Ltd., Basel, Switzerland, ⁵Department of Radiology, Weill Cornell Medical College, New York, NY, United States, ⁶Signal Processing Laboratory (LTS5), Ecole Polytechnique Fédérale de Lausanne (EPFL), Laussane, Switzerland, ⁷Radiology Department, Center for Biomedical Imaging (CIBM), Lausanne University and University Hospital, Laussane, Switzerland, ⁸Translational Neuroradiology Section, National Institute of Neurological Disorders and Stroke, NIH, 10 Center Drive MSC 1400, Building 10 Room 5C103, Bethesda, MD, United States, ⁹Department of Neurology, Cedars-Sinai Medical Center, Los Angeles, CA, United States, ¹⁰Pharmaceutical Research and Early Development, Roche Innovation Center Basel, F. Hoffmann-La Roche Ltd., Basel, Switzerland

The differential sensitivity of myelin-sensitive advanced MRIs (aMRIs) to the pathology in various brain lesions and regions in multiple sclerosis (MS) is currently debated. This study aimed to address this issue using myelin water fraction maps (MWF), quantitative susceptibility mapping (QSM) and T1 relaxometry (qT1). Our results show that (i) qT1 is the most sensitive in differentiating white matter and cortical MS lesions from normal-appearing tissue (ii) QSM is best differentiating lesions with various extent of damage (lesions with vs without paramagnetic rim & periventricular vs juxta-cortical lesions) and (iii) MWF outperforms the other aMRI methods in identifying occult normal appearing pathology.

Tianle Cao^1,2, Sen Ma¹, Nan Wang¹, Sara Gharabaghi³, Yibin Xie¹, Zhaoyang Fan^1,4,5, Elliot Hogg⁶, E. Mark Haacke ^3,7,8, Michele Tagliati⁶, Anthony G. Christodoulou^1,2, and Debiao Li^1,2
1Biomedical Imaging Research Institute, Cedars Sinai Medical Center, Los Angeles, CA, United States, ²Department of Bioengineering, University of California, Los Angeles, Los Angeles, CA, United States, ³Magnetic Resonance Innovations, Inc., Bingham Farms, MI, United States, ⁴Department of Radiation Oncology, Keck School of Medicine, University of Southern California, Los Angeles, CA, United States, ⁵Department of Radiology, University of Southern California, Los Angeles, CA, United States, ⁶Department of Neurology, Cedars Sinai Medical Center, Los Angeles, CA, United States, ⁷Department of Radiology, Wayne State University School of Medicine, Detroit, MI, United States, ⁸The MRI Institute for Biomedical Research, Bingham Farms, MI, United States

A new approach for simultaneous quantitative mapping of T1, T2, T2*, and susceptibility was developed. This technique employs hybrid T2 preparation/inversion pulse modules followed by fully flow-compensated multi-echo FLASH readouts. Our method reconstructs images with different T2 preparation times, echo times, and inversion times in the MR Multitasking framework and use them for parameter quantification. Results of both visual comparison and statistical analysis showed that our proposed method agreed well with reference methods while being more time efficient.

Giorgia Milotta¹, Gastao Cruz², Radhouene Neji², Claudia Prieto², and Rene Botnar^2
1University College London, London, United Kingdom, ²King's College London, London, United Kingdom

Quantitative T₁ and T₂ mapping has shown promising results in the quantification of liver fibrosis and inflammation, whereas proton density fat fraction mapping has been used to quantify the hepatic lipid content in fatty liver diseases. Conventionally multiple sequential 2D breath-held scans are performed to acquire sequential 2D T₁, T₂ and fat fraction maps. However, this approach suffers from limited spatial resolution and coverage and potential misregistration errors. In this work, we sought to develop a free-breathing non-rigid motion corrected 3D sequence for simultaneous and co-registered acquisition of joint liver T₁, T₂ and fat fraction maps for quantitative tissue characterization

Rakshit Dadarwal^1,2 and Susann Boretius^1,2
1Functional Imaging Laboratory, German Primate Center, Göttingen, Germany, ²Georg August Universität Göttingen, Göttingen, Germany

Neuroimaging studies require a species-specific template to promote an exquisite spatial normalization. In this work, we provide a high-resolution multi-contrast MRI template of the cynomolgus macaque (Macaca fascicularis) brain. This template may serve as a standard neuroanatomical template to normalize single subject scans in the stereotaxic space. The incorporation of multiple MRI contrast mechanisms provides excellent contrast between gray matter, white matter, and subcortical structures with fine neuroanatomical details. 

Jiaqi Dou¹, Yajie Wang¹, Huiyu Qiao¹, Zhensen Chen¹, Yuze Li¹, Haikun Qi², Jie Sun³, Dongxiang Xu³, Xihai Zhao¹, and Huijun Chen^1
1Center for Biomedical Imaging Research, School of Medicine, Tsinghua University, Beijing, China, ²School of Biomedical Engineering and Imaging Sciences, King's College London, London, United Kingdom, ³Department of Radiology, University of Washington, Seattle, WA, United States

Carotid atherosclerosis is a leading cause of ischemic stroke worldwide. Multi-contrast magnetic resonance imaging (MRI) is an ideal non-invasive technique to assess carotid plaque. However, its clinical application is limited by long scan time and non-rigid registration between different contrasts. In this study, we generated a set of co-registered T1w, T2w, PDw images, and MR angiogram (MRA) in one scan based on the SIMPLE (SImultaneous T1 and T2 Mapping of carotid PLaquE) sequence. Preliminary experiments on patients validated the feasibility of the generated multi-contrast images for carotid plaque assessment and comparable performance  with conventional sequences.

Xia Ge¹, John A Engelbach¹, Liya Yuan², Sonika Dahiya³, Feng Gao⁴, Keith M Rich², Joseph JH Ackerman^1,5,6,7, and Joel R Garbow^1,5
1Radiology, Washington University in St Louis, St Louis, MO, United States, ²Neurosurgery, Washington University in St Louis, St Louis, MO, United States, ³Neuropathology, Washington University in St Louis, St Louis, MO, United States, ⁴Surgery, Washington University in St Louis, St Louis, MO, United States, ⁵Alvin J Siteman Cancer Center, Washington University in St Louis, St Louis, MO, United States, ⁶Internal Medicine, Washington University in St Louis, St Louis, MO, United States, ⁷Chemistry Department, Washington University in St Louis, St Louis, MO, United States

Malignant brain tumor patients treated with radiation are at risk of developing subsequent treatment effects, including radiation necrosis (RN), which cannot be differentiated from recurrent tumor. We have developed mouse models of RN and of admixed tumor growing in previously irradiated brain (“mixed lesion”) that recapitulate the major histologic characteristics of human RN and recurrent glioma, respectively. These models provide a platform for the development of imaging biomarkers capable of differentiating RN vs. tumor. We demonstrate a multiparametric, clinically translatable ¹H MRI pipeline (R₁, R₁-post-contrast, R₂, ADC, MTR, and DCE_AUC) that shows significant promise for differentiating RN vs. recurrent tumor.

Yimei Lu¹, Qianfeng Wang², and Dengbin Wang^1
1Department of Radiology, Xinhua Hospital, Shanghai Jiao Tong University School of Medicine, shanghai, China, ²Fudan University, Institute of Science and Technology for Brain-Inspired Intelligence, shanghai, China

The relaxation times (including T1, T2, and T1ρ) are tissue-specific parameters , and depend on the physical, chemical and biological characteristics of the tissue. In liver fibrosis, the deposition of macromolecules (such as collagen fibers, proteoglycans, etc.) in the extracellular matrix may affect the movement of free protons, resulting in tissue relaxation times change. Many studies have shown that native T1 and T2 values are related to the degree of liver fibrosis. We investigate the influence of different pathological changes on T1 mapping, T1ρ, and T2 mapping in two vivo animal models of liver fibrosis, with a focus on liver fibrosis.

Mehdi Sadighi¹, Mert Şişman¹, and B. Murat Eyüboğlu^1
1Electrical and Electronics Eng., Middle East Technical University (METU), Ankara, Turkey

In this study, a Diffusion-Weighted (DW) Spin Echo (SE) based pulse sequence with current injection is proposed to combine data acquisitions of the Diffusion Tensor ($$$\overline{\overline{D}}$$$), current-induced magnetic flux density $$$B_z$$$ and the magnetohydrodynamic (MHD) flow imaging. The current density distribution ($$$\overline{J}$$$) can be estimated from the measured $$$B_z$$$  using the MRCDI method and the conductivity tensor can be reconstructed from the acquired $$$\overline{\overline{D}}$$$ and the estimated $$$\overline{J}$$$ using the DT-MREIT method. The acquired results of an experimental phantom with anisotropic diffusion using DW-SE pulse sequence shows that the proposed method could provide multi-contrast imaging based on multiple physical properties.

Chantal Tax^1,2, Hugo Larochelle³, Joao P. De Almeida Martins⁴, Jana Hutter⁵, Derek K. Jones², Maxime Chamberland², and Maxime Descoteaux^6
1Image Sciences Institute, University Medical Center Utrecht, Utrecht, Netherlands, ²CUBRIC, Cardiff University, Cardiff, United Kingdom, ³Google Brain, Montreal, QC, Canada, ⁴Department of Radiology and Nuclear Medicine, St. Olav's University Hospital, Trondheim, Norway, ⁵Centre for Medical Engineering, King's College London, London, United Kingdom, ⁶SCIL, University of Sherbrooke, Sherbrooke, QC, Canada

Multi-contrast MRI provides a comprehensive picture of tissue microstructure, but the high dimensionality of the parameter space increases scan time. In this work, we present a data-driven approach to multi-contrast MRI experiment design using concrete autoencoders. Concrete autoencoders simultaneously perform measurement subset-selection and learn a prediction of the full set of measurements. This approach was evaluated on two multi-contrast databases encoding diffusion, relaxation, and susceptibility. The results showed similar patterns of measurement-subset selection and mean-squared errors across different training sets. The increasing availability of public multi-contrast MRI databases can further push data-driven approaches in providing recommendations for experiment design.

Mahmut Yurt^1,2, Salman Ul Hassan Dar^1,2, Berk Tinaz^1,2,3, Muzaffer Ozbey^1,2, Yilmaz Korkmaz^1,2, and Tolga Çukur^1,2,4
1Department of Electrical and Electronics Engineering, Bilkent University, Ankara, Turkey, ²National Magnetic Resonance Research Center, Bilkent University, Ankara, Turkey, ³Department of Electrical and Computer Engineering, University of Southern California, Los Angeles, CA, United States, ⁴Neuroscience Program, Aysel Sabuncu Brain Research Center, Bilkent University, Ankara, Turkey

Current approaches for synthetic multi-contrast MRI involve deep networks trained to synthesize target-contrast images from source-contrast images in fully-supervised protocols. Yet, their performance is undesirably circumscribed to training sets of costly fully-sampled source-target images. For practically advanced multi-contrast MRI synthesis accelerated across the k-space and contrast sets, we propose a semi-supervised generative model that can be trained to synthesize fully-sampled images using only undersampled ground-truths by introducing a selective loss function expressed only on the acquired k-space coefficients randomized across training subjects. Demonstrations on multi-contrast brain images indicate that the proposed model maintains equivalent performance to the gold-standard fully-supervised model.

Jing Liu¹, Angela Jakary¹, Duan Xu¹, and Janine Lupo^1
1Radiology and Biomedical Imaging, University of California San Francisco, San Francisco, CA, United States

Multi-contrast anatomical brain MR images are widely used for tissue characterization and lesion detection, by applying multiple sequences. We developed a single continuous data acquisition applied with multiple inversion pulses, allowing multi-contrast imaging with high data acquisition efficiency. In order to achieve accurate tissue segmentation based on the multi-contrast images, imaging parameters are optimized based on simulated signal evolutions for achieving good image quality and accurate tissue characterization. A motion compensation approach for 3D whole-brain, multi-contrast MRI, was developed by exploiting the flexible data sampling strategy that allows for arbitrary combination of data collected throughout acquisition time and inversion times.

Jiahao Hu^1,2,3, Yilong Liu^1,2, Zheyuan Yi^1,2,3, Yujiao Zhao^1,2, Fei Chen³, and Ed X. Wu^1,2
1Laboratory of Biomedical Imaging and Signal Processing, The University of Hong Kong, Hong Kong, China, ²Department of Electrical and Electronic Engineering, The University of Hong Kong, Hong Kong, China, ³Department of Electrical and Electronic Engineering, Southern University of Science and Technology, Shenzhen, China

Multi-contrast MRI offers us images with complementary diagnostic information. Despite the dramatic difference in contrast, multi-contrast images often share highly correlated structure information. A deep learning (DL) based strategy is proposed to denoise multi-contrast MR images with flexible noise-levels using residual U-Net. This method utilizes the structural similarities across contrasts by simultaneously denoising multiple contrasts while existing single-contrast MRI denoising methods neglect the analogous structure information. The proposed method outperforms BM3D in terms of better noise reduction and details preservation. More importantly, we introduce a noise-level map that can be manually set to fit the different noise levels.

Daichi Murayama¹, Takayuki sakai¹, Masami Yoneyama², and Shigehiro Ochi^1
1Radiology, Eastern Chiba Medical Center, Chiba, Japan, ²Philips Japan, Tokyo, Japan

We propose a new sequence called MIXTURE (Multi-Interleaved X-prepared TSE with inTUitive RElaxometry).  MIXTURE based is a 3D TSE that can set arbitrary echo times using the T2 preparation pulses, and enables several image contrasts (such as PDW, T2W) and T2 mapping by acquiring at least two echo time images. MIXTURE could simultaneously provide morphology, pathology and T2 quantitative lumbar imaging in one single scan. This sequence is very promising to evaluate the degeneration of nerve roots and intervertebral disc for 3D T2 mapping and improve the ability of diagnostic imaging around the lumbar for 3D multi contrast MRI.

Haoxiang Li^1,2, Jing Chen¹, Yuanyuan Liu¹, Hairong Zheng¹, Dong Liang¹, and Yanjie Zhu^1
1Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen, China, ²Shenzhen College of Advanced Technology, University of Chinese Academy of Sciences, Shenzhen, China

Magnetic Resonance (MR) parametric mapping like $$$T_1$$$ , $$$T_2$$$, proton density is a powerful tool for biology tissue characterization, which is useful for clinical application such as diagnosis of pathologies including Alzheimer’s disease and multiple sclerosis¹, evaluation of myocardial fibrosis² and assessment of knee cartilage damage³. However the long scan time makes it challenging for practical clinical application. The purpose of this study was to develop a deep learning based method for accelerated MR parametric mapping with good performance at high acceleration rate both by reducing the contrast number and undersampling the k-space data.

Naoyuki Takei¹, Shohei Fujita^2,3, Issei Fukunaga², Mitsuharu Miyoshi¹, Shigeki Aoki², Suchandrima Banerjee⁴, and Tetsuya Wakayama^1
1MR Applications and Workflow, GE Heatlcare, Tokyo, Japan, ²Juntendo University School of Medicine, Tokyo, Japan, ³The University of Tokyo Graduate School of Medicine, Tokyo, Japan, ⁴MR Applications and Workflow, GE Heatlcare, Menlo Park, CA, United States

To aim for accelerated brain routine MRI, a novel 3D multi-contrast imaging technique using the hybrid acquisition of FSE and GRE was proposed. The basic image contrasts commonly used in clinical practice such as T1 weighted, T2 weighted and FLAIR were acquired simultaneously. The comparison with conventional 3D imagings was performed in term of image contrast and scan time to demonstrate the proof of concept. The proposed technique potentially gives a new perspective of the use of 3D acquisition strategy.

Zhixing Wang¹, Xue Feng¹, John Mugler², and Craig Meyer^1
1Biomedical Engineering, University of Virginia, Charlottesville, VA, United States, ²Radiology & Medical Imaging, University of Virginia, Charlottesville, VA, United States

This study describes a new approach to obtain T2-weighted and fluid-attenuated inversion recovery (FLAIR) images simultaneously in a short scan time. In this technique, the Cartesian readout is replaced by an annular spiral-ring segmentation, and a 180°(y)-90°(x) preparation RF pulse at the end of the echo train is used for driven inversion, rather than a single 180° inversion pulse as used for conventional FLAIR. Both the T2-weighted and FLAIR images can be acquired in a single scan, with higher scan efficiency and similar image quality, when compared with Cartesian TSE and Cartesian FLAIR.

Rakshit Dadarwal^1,2 and Susann Boretius^1,2
1Functional Imaging Laboratory, German Primate Center, Göttingen, Germany, ²Georg August Universität Göttingen, Göttingen, Germany

Automatic brain tissue segmentations mostly rely on T1 weighted images (T1w) which provide an excellent grey matter – to -white matter contrast but T1w lacks sufficient contrast for major subcortical nuclei. Particularly in non-human primates, these segmentation approaches encounter their limits. By merging T1w with Quantitative Susceptibility Mapping, a unique brain tissue contrast could be generated. TQ-SILiCON maps allowed for an improved subcortical nuclei classification while retaining the excellent white matter delineation of T1w. The approach works equally well for humans and non-human primates.

Holly Doig¹, Maaike van den Boomen^1,2,3, Erin Connors¹, Joan Kim¹, Jaume Coll-Font^1,3, Robert A. Eder¹, Shi Chen¹, Yoshiko Iwamoto¹, Kyrre E. Emblem⁴, Kawin Setsompop³, Niek H.J. Prakken², Ronald J.H. Borra^2,5, and Christopher T. Nguyen^1,3
1Cardiovascular Research Center, Massachusetts General Hospital, Boston, MA, United States, ²Department of Radiology, University Medical Center Groningen, Groningen, Netherlands, ³A.A. Martinos Center for Biomedical Imaging, Massachusetts General Hospital, Boston, MA, United States, ⁴Department of Diagnostic Physics, Oslo University Hospital, Oslo, Norway, ⁵Department of Nuclear Medicine and Molecular Imaging, University Medical Center Groningen, Groningen, Netherlands

Myocardial infarction (MI) and ischemic reperfusion (IR) injury are a common western world problem and are aimed to be better understood by imaging techniques to enable non-invasive longitudinal follow up. Novel MRI based techniques, such as cardiac DWI, BOLD and VAI, are emerging and detailed assessment and comparison with current golden standard and histology is needed to determine their role in treatment planning and revalidation of MI patients. Here we show that specifically BOLD can be used as an early marker in the acute MI phase.

Weijian Huang¹, Yulon Qi², Qiang He³, Ting Ma⁴, Xin Liu¹, Guanxun Cheng², Hairong Zheng¹, and Shanshan Wang^1
1Paul C Lauterbur Research Center, Shenzhen Institutes of Advanced Technology Chinese Academy of Sciences, Shenzhen, China, ²Radiology department, Peking University Shenzhen Hospital, shenzhen, China, ³United Imaging Research Institute of Innovative Medical Equipment, Shenzhen, China, ⁴Pengcheng Laboratory, Shenzhen, China

Stroke is a leading cause of death and disability worldwide. However, the misalignment between multi-contrast MR images bring difficulties in identifying the lesions. We propose an automatic framework including affine and deformation transformation for multi-contrast stroke images registration. In the framework, a new inverse operation is proposed to maintain the topology of images and a background suppression loss function is designed to optimize background predictions. The method achieves the best Dice score of 0.826 compared to 5 state-of-the-art methods. Moreover, our method is about 17 times faster than the most competitive method SyN when testing on  a same CPU. 

Thomas Campbell Arnold¹, Samantha By², Hadrien Dyvorne², Rafael O'Halloran², Farzana Sayani³, Lisa M. Desiderio⁴, Brian Litt^1,5, and Joel M. Stein^4
1Bioengineering, University of Pennsylvania, Philadelphia, PA, United States, ²Hyperfine Research, Guilford, CT, United States, ³Medicine, Perelman School of Medicine, Philadelphia, PA, United States, ⁴Radiology, Perelman School of Medicine, Philadelphia, PA, United States, ⁵Neurology, Perelman School of Medicine, Philadelphia, PA, United States

MRI provides superior imaging for diverse clinical applications, but cost and other factors limit availability in various healthcare and lower resource settings. Low-field strength units promise to expand access but involve trade-offs including reduced signal, longer acquisitions, and uncertain benefit of contrast agents. Here we characterize T1 and T2 properties of ferumoxytol, an iron oxide agent with prolonged blood pool phase and higher R1 and R2 values than gadolinium, on a 64mT portable system. We demonstrate enhancement across a range of concentrations in phantoms and visualization of cerebral vasculature in patients receiving the agent for iron-deficiency anemia.