Congyu Liao^1,2, Xiaozhi Cao^1,2, Siddharth Srinivasan Iyer^1,3, Zihan Zhou⁴, Yunsong Liu⁵, Justin Haldar⁵, Mahmut Yurt^1,2, Ting Gong⁶, Zhe Wu⁷, Hongjian He⁴, Jianhui Zhong^4,8, Adam Kerr^1,9, and Kawin Setsompop^1,2
1Department of Radiology, Stanford University, Stanford, CA, United States, ²Department of Electrical Engineering, Stanford University, Stanford, CA, United States, ³Department of Electrical Engineering and Computer Science, Massachusetts Institute of Technology, Cambridge, MA, United States, ⁴Center for Brain Imaging Science and Technology, College of Biomedical Engineering & Instrument Science, Zhejiang University, Hangzhou, China, ⁵Signal and Image Processing Institute, University of Southern California, Los Angeles, CA, United States, ⁶Centre for Medical Imaging Computing, Department of Computer Science, University College London, London, United Kingdom, ⁷Techna Institute, University Health Network, Toronto, ON, Canada, ⁸Department of Imaging Sciences, University of Rochester, Rochester, NY, United States, ⁹Stanford Center for Cognitive and Neurobiological Imaging, Stanford University, Stanford, CA, United States

In this work, we developed ViSTa-MRF, which combined Visualization of Short Transverse relaxation time component (ViSTa) technique with MR Fingerprinting (MRF), to achieve high-fidelity whole-brain myelin-water fraction (MWF) and T₁/T₂/PD mapping at sub-millimeter isotropic resolution on a clinical 3T scanner. To achieve fast acquisition and memory-efficient reconstruction, the ViSTa-MRF sequence leverages an optimized 3D tiny-golden-angle-shuffling (TGAS) spiral-projection acquisition and stochastic subspace reconstruction with optimized k-space diagonal preconditioning. With the proposed ViSTa-MRF method, high-fidelity direct MWF mapping was achieved without a need for multi-compartment fitting.

Agah Karakuzu^1,2, Julien Cohen-Adad^1,3, and Nikola Stikov^1,2
1NeuroPoly Lab, Polytechnique Montreal, Montreal, QC, Canada, ²Montreal Heart Institute, Montreal, QC, Canada, ³Unité de Neuroimagerie Fonctionnelle (UNF), Centre de recherche de l’Institut Universitaire de Gériatrie de Montréal (CRIUGM), Montreal, QC, Canada

Vendor-native implementations can lead to unaccounted variance in quantitative MRI (qMRI). The purpose of this study is to show that fully transparent qMRI workflows coupled with VEndor-NeUtral Sequences (VENUS) improve multicenter reproducibility of quantitative MRI maps across three scanners by two vendors, in phantoms and in-vivo. We developed and deployed a vendor-neutral 3D spoiled gradient-echo (SPGR) sequence on three commercial scanners and showed that vendor-neutral qMRI maps of T1, MTR and MTsat outperform their vendor-native counterparts in terms of reproducibility.

Shohei Fujita^1,2, Borjan Gagoski^3,4, Ken-Pin Hwang⁵, Marcel Warntjes^6,7, Akifumi Hagiwara¹, Issei Fukunaga¹, Wataru Uchida¹, Yuma Nishimura¹, Tomoya Muroi¹, Toshiya Akatsu¹, Akihiro Kasahara², Ryo Sato², Tsuyoshi Ueyama², Christina Andica¹, Koji Kamagata¹, Shiori Amemiya², Hidemasa Takao², Osamu Abe², and Shigeki Aoki^1
1Department of Radiology, Juntendo University, Tokyo, Japan, ²Department of Radiology, The University of Tokyo, Tokyo, Japan, ³Fetal Neonatal Neuroimaging and Developmental Science Center, Boston Children's Hospital, Boston, MA, United States, ⁴Department of Radiology, Harvard Medical School, Boston, MA, United States, ⁵Department of Imaging Physics, The University of Texas MD Anderson Cancer Center, Houston, TX, United States, ⁶SyntheticMR, Linköping, Sweden, ⁷Center for Medical Imaging Science and Visualization (CMIV), Linköping University, Linköping, Sweden

 

Cross-vendor techniques enabling the pooling of data among different vendors is desired in multi-center projects; however, few mapping techniques have shown compatibility. In this study, we examined the intra-scanner repeatability and inter-vendor reproducibility of quantitative measurements obtained via a multi-parametric mapping technique, 3D-QALAS. Volunteers underwent scan-rescan sessions on four different 3T systems from three MRI vendors (GE, Philips, and Siemens). T1, T2, and PD values and brain tissue volumes of healthy volunteers derived from 3D-QALAS were robust across scanners from different vendors (inter-vendor coefficient of variation=3.3%). The data in this study will inform future technical developments for 3D-QALAS.
 

Dan Benjamini¹, Mustapha Bouhrara¹, Michal E Komlosh², Diego Iacono³, Daniel P Perl³, David L Brody³, and Peter J Basser^2
1National Institute on Aging, Baltimore, MD, United States, ²National Institute of Child Health and Human Development, Bethesda, MD, United States, ³Uniformed Services University of the Health Sciences, Bethesda, MD, United States

Here we present MRI acquisitions that maximize the amount of information in an image by using a combination of magnetic field profiles to probe relaxation and diffusion mechanisms simultaneously, thus resolving sub-voxel information from ex vivo brain segments of human subjects that sustained traumatic brain injury. We then propose a strategy to refine the ensuing abundance of high-dimensional data, and to sort and recover biologically relevant information into actionable and quantitative injury biomarker images that capture subtle neuropathology.

Gian Franco Piredda^1,2,3, Marion Claudet¹, Tom Hilbert^1,2,3, Manuela Vaneckova⁴, Jan Krasensky⁴, Michaela Andelova⁵, Tomas Uher⁵, Eva Kubala Havrdova⁵, Karolina Vodehnalova⁵, Dana Horakova⁵, Karl Egger⁶, Shan Yang⁶, Riccardo Ludovichetti⁷, Lindsey A. Crowe⁷, Bénédicte M. A. Delattre⁷, Maria Isabel Vargas⁷, Veronica Ravano^1,2,3, Bénédicte Maréchal^1,2,3, Jean-Philippe Thiran^2,3, and Tobias Kober^1,2,3
1Advanced Clinical Imaging Technology, Siemens Healthcare AG, Lausanne, Switzerland, ²Department of Radiology, Lausanne University Hospital and University of Lausanne, Lausanne, Switzerland, ³LTS5, École Polytechnique Fédérale de Lausanne (EPFL), Lausanne, Switzerland, ⁴Department of Radiology, First Faculty of Medicine, Charles University and General University Hospital, Prague, Czech Republic, ⁵Department of Neurology and Center of Clinical Neuroscience, First Faculty of Medicine, Charles University and General University Hospital, Prague, Czech Republic, ⁶Department of Neuroradiology, Medical Center – University of Freiburg, Faculty of Medicine, University of Freiburg, Freiburg, Germany, ⁷Divsion of Neuroradiology, Department of Radiology and Medical Informatics, Hôpitaux Universitaires de Genève (HUG), Geneva, Switzerland

Comparing single-patient data to atlases of normative relaxation times in the brain enables personalized characterization of pathological tissue. However, an accurate comparison is challenging in the cortex due to the gyrification of gray matter and resulting registration problems. This work addresses this problem for parametric T₁ maps that were acquired from 285 healthy subjects. A method aligning inter-subject brain cortices was developed to enable the comparison of cortical T₁ values. We thus established a normative atlas accounting for healthy T₁ values in brain cortical tissues, to be used to detect and characterize pathology-induced T₁ alterations in patients.

Ilyes Benslimane¹, Günther Grabner², Simon Hametner³, Thomas Jochmann^1,4, Robert Zivadinov^1,5, and Ferdinand Schweser^1,5
1Department of Neurology, Buffalo Neuroimaging Analysis Center, Buffalo, NY, United States, ²Department of Medical Engineering, Carinthia University of Applied Sciences, Klagenfurt, Austria, ³Department of Biomedical Imaging and Image-Guided Therapy, High Field MR Centre, Medical University of Vienna, Austria, ⁴Department of Computer Science and Automation, Technische Universität Ilmenau, Ilmenau, Germany, ⁵Center for Biomedical Imaging, Clinical and Translational Science Institute at the University at Buffalo, Buffalo, NY, United States

Most quantitative MRI (qMRI) parameters co-depend on multiple tissue properties, limiting their clinical value. Here, we introduce a biophysically constrained autoencoder network that accepts multiparametric MR data and outputs underlying biological tissue parameters. The method requires no ground truth data, no knowledge of the (generally unknown) biophysical model parameters, and a single subject provides sufficient training data. We validated the method using ex vivo iron and myelin stains.  

Christopher E Vaughn^1,2, N Reid Bolding³, Mark A Griswold⁴, and William A Grissom^1,2
1Department of Biomedical Engineering, Vanderbilt University, Nashville, TN, United States, ²VUIIS, Vanderbilt University, Nashville, TN, United States, ³Physics, Case Western Reserve University, Cleveland, OH, United States, ⁴Radiology, Case Western Reserve University, Cleveland, OH, United States

B0 gradients have several problems including cost, bulkiness, noise, and PNS. We propose to use a method called Selective Encoding through Nutation and Fingerprinting (SENF) which is a RF spatial encoding technique that also encodes quantitative information. In this abstract we validate SENF on a 47.5 mT low-field scanner using a high-flip MR Fingerprinting type sequence with a gradient RF coil to perform a 4-voxel 1D reconstruction that can differentiate between three materials (mineral oil, water, and air).

Hampus Olsson¹ and Gunther Helms^1
1Medical Radiation Physics, Lund University, Lund, Sweden

Accuracy in MP2RAGE-based T₁-mapping relies on knowledge of the imperfect inversion efficiency (f_inv) of the inversion pulse. Here we map the inversion efficiency of different adiabatic inversion pulses in human brain at 7T. We identified a mean f_inv=0.87, which is substantially lower than the f_inv=0.96 typically assumed in MP2RAGE. We further observed a lower f_inv in white matter (WM) than in the rest of the cerebrum. Subsequent T₁ estimates differed by approximately 80 ms in WM and 200 ms in gray matter (GM) when assuming either f_inv=0.87 or f_inv=0.96.

Max Hellström¹ and Anders Garpebring^1
1Radiation Sciences, Umeå University, Umeå, Sweden

Tissue parameter estimation often gives noisy parameter maps due to noise in the signal data. In this work, we improve parameter mapping by incorporating noise reduction and uncertainty estimation with Bayesian Deep Image Prior. We implement task-specific loss functions for different applications.  We test our method by estimating T1, T2, and ADC with synthetic- and in-vivo MRI data. Our method results in denoised tissue parameter maps, with associated error estimates. Our method is easy to implement, does not require any training data, and is easy to customize for different applications in parameter mapping.

Dinank Gupta¹, Dave Choi¹, Steven Allen², Tim Hall¹, Zhen Xu¹, and Douglas Noll^1
1University of Michigan, Ann Arbor, MI, United States, ²Brigham Young University, Provo, UT, United States

The accuracy of using MR-Thermometry for pre-treatment targeting for histotripsy was evaluated on ex-vivo tissues. Eight brain samples were treated with an ultrasound array capable of performing both heating and histotripsy treatments. Heating was performed at array focus while MR-Thermometry images were acquired. Subsequently, histotripsy lesions were also made at the focus and post-treatment images were acquired. The location of MR-Thermometry estimated focal-zone was compared with the observed histotripsy lesion location to assess the accuracy of MR-Thermometry targeting. The mean error in estimated focus was about 1mm in all axes, suggesting MR-Thermometry would be suitable for pre-treatment targeting for histotripsy.