Le Zhang¹, Tzucheng Chao¹, Dinghui Wang¹, Guruprasad Krishnamoorthy², and James Pipe^1
1Radiology, Mayo Clinic, Rochester, MN, United States, ²MR R&D, Philips Healthcare, Rochester, MN, United States

Conventionally, Cartesian pre-scans are collected over several seconds to create accurate coil sensitivity maps (CSM) for SENSE reconstruction of under-sampled MRI data, with body-coil images used for signal normalization. Since body coils have a larger sensitivity region than phased-array coils, these pre-scans typically employ a very large field-of-view (FOV) to avoid aliasing, leading to low-resolution CSM and potential small errors in subsequent reconstructions. This abstract proposes a new CSM pre-scan sequence using a spherically-distributed-spirals trajectory. The acquisition efficiency and incoherent aliasing offered by this trajectory allow for smaller FOV and higher resolution while maintaining similar scanning time as conventional pre-scans.

Bejoy Thomas¹, Jithin S S¹, Ajimi mol Anzar¹, and Santhosh Kannath^2
1Imaging Sciences and Interventional Radiology, Sree Chitra Tirunal Institute for Medical Sciences and Technology, Thiruvananthapuram, India, ²Sree Chitra Tirunal Institute for Medical Sciences and Technology, Thiruvananthapuram, India

Artificial intelligence techniques are widely used in medical imaging and diagnostics. In this retrospective study, a convolutional neural network (CNN) architecture was developed to classify intracranial dural arteriovenous fistula (DAVF) on Susceptibility Weighted Images (SWI). The dataset used was a total of 3965 SWI image slices of DAVF patients and 4380 images of controls. The proposed classifier showed significant accuracy in the diagnosis of DAVF and it could be developed as a computer assisted diagnosis tool to identify unsuspected DAVF in routine MR imaging.

Thomas Meunier¹, Chengran Fang², Maëliss Jallais¹, and Demian Wassermann^1
1INRIA Saclay, Paris, France, ²INRIA Saclay, Palaiseau, France

We extract cytoarchitectural characteristics of brain gray matter from diffusion MRI signals including soma size, neurite signal fraction and water exchange. Our model improves on state-of-the-art in that 1) we extract an invertible system leading to stable parameters estimation, 2) our simulation-based inference approach allows to obtain the full posterior distribution of the parameters given a signal. Our solution is a two-step model. First, a new forward model relates summary statistics of the dMRI signal to different tissue parameters. Then, a likelihood-free inference-based algorithm is applied to invert the model, and returns a full posterior distribution over the parameter space.

Elaine Yuen Fong Kuan^1,2, Viktor Vegh^1,2, Kieran O'Brien³, Amanda Hammond³, Javier Urriola Yaksic¹, and David Reutens^1,2
1Centre of Advanced Imaging, The University of Queensland, Brisbane, Australia, ²ARC Training Centre for Innovation in Biomedical Imaging Technology, The University of Queensland, Brisbane, Australia, ³Siemens Healthcare Pty Ltd, Brisbane, Australia

Common approaches for analyzing task-based fMRI data rely upon the use of regressors, which in some experimental paradigms are difficult to define. A machine learning method is proposed to overcome this challenge. Three machine learning methods with established utility for time series classification were used to classify areas of activation and non-activation in a language fMRI study. Machine learning methods were able to identify the activation regions identified by analyses using the General Linear Model (GLM). Machine learning may be useful for fMRI time series analysis, particularly when regressors required for GLM-based analyses are difficult to define.

Mahta Karimpoor¹, Moss Zhao¹, Brian Mills¹, Marios Georgiadis¹, Dean Tran¹, Maged Goubran², Nicole Mouchawar¹, Sohrab Sami³, Max Wintermark¹, Gerald Grant⁴, David Camarillo⁵, Greg Zaharchuk¹, and Michael Zeineh^1
1Radiology, Stanford University, Stanford, CA, United States, ²Medical Biophysics, University of Toronto, Toronto, ON, Canada, ³Stanford Center for Clinical Research, Stanford, CA, United States, ⁴Neurosurgery, Stanford University, Stanford, CA, United States, ⁵Bioengineering, Stanford University, Stanford, CA, United States

Longitudinal changes (over four years) of cerebral blood flow (CBF) using arterial spin labeling MRI were investigated in a population of high-contact sport football college athletes and were compared to low-contact cohort of volleyball athletes. A linear-mixed-effects model was applied to assess CBF (normalized to the cerebellum) by sport (football vs. volleyball), time from baseline MRI, and the interaction between sport and time. Longitudinal analysis showed a prospective decline in perfusion in football compared to volleyball. Fourteen football players experienced an in-study concussion; in contrast to the longitudinal findings, football players exhibited acutely a mild increase in occipital lobe CBF.

Simona Schiavi¹, Marco Palombo^2,3,4, Domenico Zacà⁵, Francesco Tazza¹, Caterina Lapucci⁶, Lucio Castellan⁷, Mauro Costagli¹, and Matilde Inglese^1,8
1Department of Neuroscience, Rehabilitation, Ophthalmology, Genetics, Maternal and Child Health (DINOGMI), University of Genoa, Genoa, Italy, ²Centre for Medical Image Computing (CMIC), Department of Computer Science, University College London, London, United Kingdom, ³Cardiff University Brain Research Imaging Centre (CUBRIC), School of Psychology, Cardiff University, Cardiff, United Kingdom, ⁴School of Computer Science and Informatics, Cardiff University, Cardiff, United Kingdom, ⁵Siemens Healthcare s.r.l, Milan, Italy, Milan, Italy, ⁶HNSR, IRRCS Ospedale Policlinico San Martino, Genoa, Italy, ⁷Department of Neuroradiology, IRCCS Ospedale Policlinico San Martino, Genoa, Italy, ⁸IRCCS Ospedale Policlinico San Martino, Genoa, Italy

Diffusion MRI is a powerful technique that, thanks to advanced signal modelling like the Soma And Neurite Density Image (SANDI) can probe microstrucutral information of both grey and white matter. However, this model requires multishell acquisitions including b-values that are at least 6 times higher than those used in clinical practice. Here we propose a 10-minute acquisition protocol that enables to acquire such images on a clinical 3T scanner. We show the reproducibility of our approach on five healthy subjects as well as potential clinical impact on two subjects affected by multiple sclerosis.

Asante Ntata¹, Olivier Mougin², Matteo Bastiani¹, Fidel Alfaro Almagro³, Jon Campbell³, Paul S Morgan², Mark Jenkinson^3,4, and Stamatios N Sotiropoulos^1,3
1Sir Peter Mansfield Imaging Centre, School of Medicine, University of Nottingham, Nottingham, United Kingdom, ²Sir Peter Mansfield Imaging Centre, School of Physics, University of Nottingham, Nottingham, United Kingdom, ³Wellcome Centre for Integrative Neuroimaging (WIN - FMRIB), University of Oxford, Oxford, United Kingdom, ⁴Australian Institute for Machine Learning, University of Adelaide, Adelaide, Australia

A key challenge in robustly extracting quantitative information from MRI data is the dependence of derived features on nuisance factors, such as the scanning protocol, hardware and software, which are different between vendors and vary with site. While there exist several harmonisation approaches, what’s missing is objective ways and datasets to compare them. Here we present a novel multi-modal neuroimaging data resource for evaluating and comparing harmonisation approaches based on a “travelling heads” paradigm. We further demonstrate how such a resource can be used to a) map the need for harmonisation for different imaging-derived features, b) evaluate existing harmonisation approaches.

Benjamin Ades-Aron^1,2, Mohammed Elsayed¹, Michael Hoch³, Gregory Lemberskiy¹, Yao Wang², Dmitry S. Novikov¹, Els Fieremans¹, and Timothy M. Shepherd^1
1Center for Advanced Imaging Innovation and Research (CAI2R), Department of Radiology, New York University, school of medicine, New York, NY, United States, ²Electrical and computer engineering, New York University, Tandon school of engineering, Brooklyn, NY, United States, ³Radiology, University of Pennsylviania, Philadelphia, PA, United States

The basal ganglia, thalamus and brainstem are affected by movement disorders and contain key targets for functional neurosurgery. Targeting however is based on indirect coordinates originally derived from pneumoencephalograms! 3D Fast Gray Matter Acquisition T1 Inversion Recovery (FGATIR) can directly visualize potential targeted structures (e.g. dentatorubrothalamic tract), but is signal-starved in clinically-feasible acquisitions. We developed a convolutional neural network to improve FGATIR quality. Expert rater assessment suggested this CNN improved contrast resolution of individual structures and overall clinical image quality of 1-average data to the level of 4-averages. This could further enable investigations of functional neurosurgery for movement disorders.

Matthias Weigel^1,2,3, Peter Dechent⁴, Riccardo Galbusera^1,2, Erik Bahn⁵, Ludwig Kappos^1,2, Wolfgang Brück⁵, Christine Stadelmann⁵, and Cristina Granziera^1,2
1Translational Imaging in Neurology (ThINk) Basel, Department of Biomedical Engineering, Faculty of Medicine, University Hospital Basel and University of Basel, Basel, Switzerland, ²Neurological Clinic and Policlinic, MS Center and Research Center for Clinical Neuroimmunology and Neuroscience Basel (RC2NB), University Hospital Basel and University of Basel, Basel, Switzerland, ³Dept. of Radiology, Division of Radiological Physics, University Hospital Basel, Basel, Switzerland, ⁴Department of Cognitive Neurology, MR-Research in Neurology and Psychiatry, University Medical Center Göttingen, Göttingen, Germany, ⁵Institute of Neuropathology, University Medical Center Göttingen, Göttingen, Germany

Balanced steady state free precession (bSSFP) sequences provide the highest signal intensity per unti time, which makes them virtually predestined for ultra-high resolution MR imaging at 3T. Their sensitivity to susceptibility effects and demand for high performance, however, represent two major drawbacks. It will be shown that a carefully chosen protocol, which also includes common phase cycling techniques, will enable artifact-free 200-microns isotropic bSSFP acquisitions of the entire human fixed brain within approximately 27h.

Eberhard Daniel Pracht¹, Markus Cremer², Daniel Loewen¹, Franck Mauconduit³, Aurelien Massire⁴, Sebastian Bludau², Nicolas Boulant³, Katrin Amunts^2,5, and Tony Stoecker^1,6
1German Center for Neurodegenerative Diseases (DZNE), Bonn, Germany, ²Institute for Neuroscience and Medicine, INM-1, Structural and Functional Organisation of the Brain, Forschungszentrum Juelich GmbH, Juelich, Germany, ³University of Paris-Saclay, CEA, CNRS, BAOBAB, NeuroSpin, Gif sur Yvette, Paris-Saclay, France, ⁴Siemens Healthcare SAS, Saint-Denis, France, ⁵Cecile and Oskar Vogt Institute for Brain Research, Heinrich Heine University Duesseldorf, University Hospital, Duesseldorf, Germany, ⁶Department of Physics and Astronomy, University of Bonn, Bonn, Germany

In this work we present a simple set-up for high resolution postmortem imaging at ultra high field strength. The processed images have high, as well as homogeneous gray/white matter contrast over the whole brain. Additionally, during post-processing, the background signal of the fixation liquid is almost completely removed, allowing for an optimal image registration process after histological sectioning.

Sophie Richter¹, Stefan Winzeck^1,2, Marta M Correia³, Evgenios N Kornaropoulos⁴, Anne Manktelow¹, Joanne Outtrim¹, Doris Chatfield¹, Jussi Posti⁵, Olli Tenovuo⁵, Guy B Williams⁶, David K Menon¹, and Virginia F J Newcombe^1
1Division of Anaesthesia, University of Cambridge, Cambridge, United Kingdom, ²BioMedIA Group, Department of Computing, Imperial College, London, United Kingdom, ³MRC Cognition and Brain Sciences Unit, University of Cambridge, Cambridge, United Kingdom, ⁴Diagnostic Radiology, Lund University, Lund, Sweden, ⁵Department of Neurosurgery and Turku Brain Injury Centre, Turku University Hospital and University of Turku, Turku, Finland, ⁶Wolfson Brain Imaging Centre, Department of Clinical Neurosciences, University of Cambridge, Cambridge, United Kingdom

The trend in neuroimaging towards multi-site studies requires validated harmonization approaches to eradicate scanner differences which mask the biological effect of interest. Here, the harmonization algorithm ComBat and its modification for longitudinal data (LongComBat) were compared on a large travelling subject sample (n=23 for structural MRI and n=31 for diffusion tensor MRI).

 

In structural data scanner difference are not apparent in unharmonized data but can be created by harmonization. For DTI data, scanner differences in unharmonized data are large and, both ComBat and LongComBat successfully diminished those in most regions of interest, with LongComBat achieving slightly lower false positive rates.

Rachel L. C. Barrett¹, Robert Dallyn¹, Marie-Stephanie Cahart¹, Richard Stones², Maarten Timmers³, Steven Williams¹, and Flavio Dell'Acqua^2
1Department of Neuroimaging, King's College London, London, United Kingdom, ²Department of Forensics and Neurodevelopmental Sciences, King's College London, London, United Kingdom, ³Janssen Research and Development, Janssen Pharmaceutica NV, Beerse, Belgium

Reliability and test-retest reproducibility of simple and advanced diffusion metrics were evaluated in healthy older adults. Advanced models such as Diffusion Kurtosis Imaging, White Matter Tract Integrity, Neurite Orientation Dispersion and Density Imaging and other multi-compartment methods have greater specificity but lower reliability and reproducibility than simple models such as DTI and spherical harmonic metrics. All models nevertheless had good or acceptable reproducibilty and reliability in certain cases. Particular care may be given when planning clinical research applications with advanced metrics, as greater sample sizes and/or improved data quality may be needed.

Han Dou¹, Xiaoming Wang¹, Yang Zheng¹, and Zhiwei Shen^2
1Department of Radiology, Shengjing Hospital of China Medical University, Shenyang, China, ²Philips healthcare，Beijing，China, Beijing, China

At present, the detection of meningeal diseases is mainly based on MRI enhanced scan sequences, however, the poor image quality of affected the degree of meninges visualization due to the limited concentration of contrast agent staying in the meningeal after administration. In this study, the feasibility of meningeal imaging with multi-echo gradient echo MR sequence (Mffe) was explored, and  optimized the main scan parameters. The optimal imaging parameters were obtained by analyzing the meninge-skull signal intensity ratio (SIR) and contrast-noise ratio (CNR) of healthy volunteers.

Lawrence Ver Hoef^1,2 and Anandh Kilpattu Ramaniharan^1
1Neurology, University of Alabama at Birmingham, Birmingham, AL, United States, ²Neurology, Birmingham VA Medical Center, Birmingham, AL, United States

We present a novel method to objectively quantify a morphological feature known as hippocampal dentation from ultra high resolution 3D T2w images.

Emily Rose Triolo¹, Oleksandr Khegai², Jelle Veraart³, Akbar Alipour², Trey Hedden⁴, Mehmet Kurt^1,2, and Priti Balchandani^2
1Mechanical Engineering, Stevens Institute of Technology, Hoboken, NJ, United States, ²Biomedical Engineering and Imaging Institute, Ichan School of Medicine at Mount Sinai, New York City, NY, United States, ³New York University, New York City, NY, United States, ⁴Ichan School of Medicine at Mount Sinai, New York City, NY, United States

MRE, a technique used to characterize the viscoelasticity of tissues, is typically performed at 3T for the human brain. MRE at 7T has the promise to deliver higher resolution, however, is subject to several artifacts and limitations. Here, we developed and tested an MRE setup and sequence at 7T on one subject at three different resolutions to investigate how both changing resolution and using MP-PCA denoising affects the estimated shear modulus (|G*|). Our pilot study has shown an increase in OSS-SNR after using MP-PCA denoising, and has demonstrated a relationship between resolution, OSS-SNR, and |G*| that requires further investigation.

Sihan celestine Chen¹, Yunfei Zha², Changsheng Liu², Xixiang chen³, Ling ma⁴, and Weiyin vivian Liu^5
1Radiology, Renmin Hospital of Wuhan University and Hubei General Hospital, Wuhan, China, ²Radiology, enmin Hospital of Wuhan University and Hubei General Hospital, Wuhan, China, ³Renmin Hospital of Wuhan University and Hubei General Hospital, Wuhan, China, ⁴He Kang Corporate Management (SH) Co.Ltd, Shanghai, China, Wuhan, China, ⁵Advanced Application specialist, GE Healthcare, Beijing, China, Beijing, China

This study aimed to construct a radiomics-based T2-weighted imaging (T2WI) approach from high-resolution multi-contrast magnetic resonance imaging (hrMRI) in combination with clinical high-risk factors for non-invasive assessment, to differentiate symptomatic carotid atherosclerotic plaques from asymptomatic ones. Radscore showed a better diagnostic performance. The combination model of texture features and clinical data had the best performance in assessment of lesion vulnerability. This study demonstrated that hrMRI radiomics features provided incremental value for assessment of the carotid atherosclerotic vulnerability.

Di Wang¹, Chunmei Li², Yuhui Chen², Pu-Yeh Wu³, and Min Chen^1
1Beijing Hospital, National Center of Gerontology; Institute of Geriatric Me, Beijing, China, ²Beijing Hospital, Beijing, China, ³GE Healthcare, Beijing, China, Beijing, China

Synthetic MRI is a novel method which can provides quantitative relaxation mapping and synthetic contrast-weighted images at the same time. In this study, we adopted this technique to assessed the frailty. We found T2 relaxation of right insular lobe and right limbic subcortical nuclei had significant difference between frailty and prefrailty patients. T1 and T2 relaxation of right limbic subcortical nuclei and right basal ganglia significantly correlated with the short physical performance battery total scores. Overall, our findings suggested the synthetic MR can be considered an effective tool for assessing frailty.