Matteo Cencini^1,2, Marta Lancione^2,3, Laura Biagi^1,2, Jan W Kurzawski^1,2, Rosa Pasquariello¹, Graziella Donatelli^2,4, Claudia Dosi^1,5, Chiara Ticci^1,5, Roberta Battini^1,5, Guido Buonincontri^1,2, and Michela Tosetti^1,2
1IRCCS Stella Maris, Pisa, Italy, ²Imago7 Foundation, Pisa, Italy, ³IMT School for Advanced Studies Lucca, Lucca, Italy, ⁴Neuroradiology Unit, Azienda Ospedaliero-Universitaria Pisana, Pisa, Italy, ⁵Department of Clinical and Experimental Medicine, University of Pisa, Pisa, Italy

Magnetic Resonance Fingerprinting (MRF) has been used to obtain Myelin Water Fraction (MWF) estimates in a cohort of developing children by using a three-component signal model. Here, we used a recent approach in which we perform sub-voxel tissue characterization without assumptions on the number and properties of the model components. We then used this signal model to study the myelination process in the developing brain on a 2D MRF dataset. Finally, we measured MWF on a set of subjects acquired with 3D MRF scan and compared the results to the 2D experiment.

Shohei Fujita^1,2, Matteo Cencini^3,4, Guido Buonincontri^3,4, Naoyuki Takei⁵, Rolf F. Schulte⁶, Wataru Uchida¹, Akifumi Hagiwara¹, Koji Kamagata¹, Osamu Abe², Michela Tosetti^3,4, and Shigeki Aoki^1
1Department of Radiology, Juntendo University, Tokyo, Japan, ²Department of Radiology, The University of Tokyo, Tokyo, Japan, ³Imago7 Foundation, Pisa, Italy, ⁴IRCCS Stella Maris, Pisa, Italy, ⁵MR Applications and Workflow, GE Healthcare, Tokyo, Japan, ⁶GE Healthcare, Munich, Germany

Magnetic resonance fingerprinting (MRF) permits simultaneous acquisition of T1 and T2 maps perfectly aligned to the anatomy, allowing morphometry and relaxometry analysis of the brain in a single scan. Here, we examined the reproducibility and repeatability of simultaneous morphology and relaxometry of brain structures in healthy volunteers using three-dimensional MRF at multiple field strengths. Scan-rescan tests of three-dimensional MRF were performed at 1.5 and 3T. The local thickness, volume, T1, and T2 values were calculated for each representative neuroanatomical structure using automatic brain segmentation software. These results can help establish imaging biomarkers using MRF for clinical use.

Thomas O`Reilly¹, Peter Börnert^1,2, Andrew Webb¹, and Kirsten Koolstra^1
1Radiology, Leiden University Medical Center, Leiden, Netherlands, ²Philips Research Hamburg, Hamburg, Germany

In vivo MR relaxation times, their inter-subject variations, and changes in different diseases have not been widely studied at very low magnetic fields (<100 mT). In this work, we implemented a 3D MRF sequence on a 50 mT Halbach permanent magnet system to efficiently measure relaxation times in vivo. We used a short flip angle train and further accelerate the scans by using random undersampling and matrix completion reconstruction. Initial in vivo and phantom data show good agreement with relaxation time values measured with less efficient conventional techniques.

Beomgu Kang¹, Byungjai Kim¹, Hye-Young Heo^2,3, and Hyunwook Park^1
1Department of Electrical Engineering, Korea Advanced Institute of Science and Technology, Daejeon, Korea, Republic of, ²Russell H Morgan Department of Radiology and Radiological Science, Johns Hopkins University, Baltimore, MD, United States, ³F.M. Kirby Research Center for Functional Brain Imaging, Kennedy Krieger Institute, Baltimore, MD, United States

Magnetization transfer contrast MR fingerprinting (MTC-MRF) is a novel quantitative imaging method that simultaneously quantifies free bulk water and semisolid macromolecule parameters using pseudo-randomized scan parameters. Here, we propose a framework for learning-based optimization of the acquisition schedule (LOAS), which optimizes RF saturation-encoded MRF acquisitions with a minimum number of acquisitions for tissue parameter estimation. Unlike the optimization methods based on indirect measurements, the proposed approach can optimize scan parameters by directly computing quantitative errors in tissue parameters.

Siyuan Hu¹, Ignacio Rozada², Rasim Boyacioglu³, Stephen Jordan⁴, Sherry Huang³, Matthias Troyer⁴, Mark Griswold³, Debra McGivney³, and Dan Ma^3
1Biomedical Engineering, Case Western Reserve University, Cleveland, OH, United States, ²1Qbit, Vancouver, BC, Canada, ³Case Western Reserve University, Cleveland, OH, United States, ⁴Microsoft, Redmond, WA, United States

MR Fingerprinting is able to quantify multiple tissue properties simultaneously. Here we propose an advanced MR Fingerprinting optimization framework that computes and minimizes the quantitative random errors, undersampling errors and background phase errors in MRF maps simultaneously in the cost function. The optimization is solved by quantum-inspired algorithms. The proposed framework could provide accelerated MRF scans that are robust to undersampling and system imperfections, and outperform the human-designed sequence on the tradeoff between duration and precision.

Feng Cheng¹, Yong Chen², and Pew-Thian Yap^3
1Department of Computer Science, University of North Carolina, Chapel Hill, NC, United States, ²Case Western Reserve University, Cleveland, OH, United States, ³Department of Radiology and Biomedical Research Imaging Center, University of North Carolina, Chapel Hill, NC, United States

We accelerate submillimeter 3D MRF using a dual-domain deep learning reconstruction approach that utilizes a graph convolutional network for k-space and a U-Net for image space acceleration. Our preliminary results show that a total of 16x acceleration can be achieved, reducing the acquisition time for whole-brain-coverage at 0.8 mm isotropic resolution to less than 5 mins.

Carlos Velasco¹, Gastao Cruz¹, René M. Botnar¹, and Claudia Prieto^1
1School of Biomedical Engineering and Imaging Sciences, King's College London, London, United Kingdom

Quantitative T₁, T₂, T₂^* and fat fraction maps are promising imaging biomarkers for the assessment and follow-up of liver disease, while T_1⍴ mapping has been reported to be a valuable tool for contrast-free assessment of liver fibrosis. However, these multiple scans are usually performed sequentially during separate breath-holds, leading to long scan times and potentially mis-registered maps. In this study we propose an 8-echo T₁, T₂ and T_1⍴ prepared liver MRF sequence that allows for quantitative T₁, T₂, T₂^*, T_1⍴ and FF liver tissue characterization in a single breath-hold scan. The proposed approach has been investigated in healthy subjects.

Aurélien Delphin¹, Fabien Boux^1,2, Clément Brossard¹, Jan M Warnking¹, Benjamin Lemasson¹, Emmanuel Luc Barbier¹, and Thomas Christen^1
1Univ. Grenoble Alpes, Inserm, U1216, Grenoble Institut Neurosciences, GIN, 38000, Grenoble, France, ²Univ. Grenoble Alpes, Inria, CNRS, G-INP, 38000, Grenoble, France

MR vascular fingerprinting aims at mapping cerebral vascular properties. We propose to improve the method on two levels: (1) by testing new acquisitions patterns using a Monte-Carlo based method  that assesses the encoding capacity of MRF sequences; and (2) by testing new geometrical models that represent vascular networks during dictionary simulations. We obtained results suggesting that new MRF-type sequences can be tailored for vascular exploration and should be tested in vivo. We also showed the clear influence of geometry in the simulations and the possibility to include realistic vascular networks.

Xiaozhi Cao^1,2, Congyu Liao^1,2, Siddharth Srinivasan Iyer^3,4, Gilad Liberman³, Zijing Dong^3,4, Ting Gong⁵, Zihan Zhou⁵, Hongjian He⁵, Jianhui Zhong^5,6, and Berkin Bilgic^3,7
1Department of Rdiology, Stanford university, Stanford, CA, United States, ²Department of Electrical Engineering, Stanford university, Stanford, CA, United States, ³Athinoula A. Martinos Center for Biomedical Imaging, Massachusetts General Hospital, Charlestown, MA, United States, ⁴Department of Electrical Engineering and Computer Science, MIT, Cambridge, MA, United States, ⁵Center for Brain Imaging Science and Technology, Department of Biomedical Engineering, Zhejiang University, Hangzhou, China, ⁶Department of Imaging Sciences, University of Rochester, Rochester, NY, United States, ⁷Department of Radiology, Harvard Medical School, Cambridge, MA, United States

To improve the quality and speed of 3D MRF, we applied spatiotemporal subspace reconstruction to 3D MRF and further modified its spiral-projection spatiotemporal encoding scheme. When compared to conventional sliding-window iNUFFT reconstruction, the subspace reconstruction provided markedly improved quantitative maps, with lower artifacts and higher SNR. The optimized spiral-projection encoding scheme, which was designed to increase spatiotemporal incoherency, was also validated to be more robust to artifacts, particularly at high accelerations. The proposed method enables high-quality whole-brain T₁, T₂, and proton density mapping with 1-mm isotropic resolution in 2 minutes and 0.8-mm isotropic resolution in ~4minutes.

Mark Stephan Widmaier¹, Song-I Lim², and Lijing Xin^2
1Laboratory for Functional and Metabolic Imaging, CIBM, EPFL, Lausanne, Switzerland, ²Animal Imaging and Technology, CIBM, EPFL, Lausanne, Switzerland

 This study shows preliminary results of the feasibility using MRS-FP to measure the relaxation parameters of γ-ATP and PCr as well as the chemical exchange rate k_CK. All results were in good agreement with literature values enabling a fast and robust 31 P multi parameter estimation in vivo human brain at 7T .