Vishwesh Nath¹, Marco Pizzolato^2,3, Marco Palombo⁴, Noemi Gyori⁴, Kurt G Schilling⁵, Colin Hansen⁶, Qi Yang⁶, Praitayini Kanakaraj⁶, Bennett A Landman⁶, Soumick Chatterjee⁷, Alessandro Sciarra⁷, Max Duennwald⁷, Steffen Oeltze-Jafra⁷, Andreas Nuernberger⁷, Oliver Speck⁷, Tomasz Pieciak ⁸, Marcin Baranek⁸, Kamil Bartocha⁸, Dominika Ciupek⁸, Fabian Bogusz⁸, Azam Hamidinekoo⁹, Maryam Afzali ¹⁰, Harry Lin⁴, Danny C Alexander⁴, Haoyu Lan¹¹, Farshid Sepehrband¹¹, Zifei Liang¹², Tung-Yeh Wu¹³, Ching-Wei Su¹³, Qian-Hua Wu¹³, Zi-You Liu¹³, Yi-Ping Chao¹³, Enes Albay¹⁴, Gozde Unal¹⁴, Dmytro Pylypenko¹³, Xinyu Ye¹³, Fan Zhang¹⁵, and Jana Hutter^16
1NVIDIA Corporation, Bethesda, MD, United States, ²Department of Applied Mathematics and Computer Science, Technical University of Denmark, Kongens Lyngby, Denmark, ³École polytechnique fédérale de Lausanne (EPFL), Lausanne, Switzerland, ⁴CMIC, University College London, London, United Kingdom, ⁵Institute of Imaging Science, Vanterbilt University, Nashville, TN, United States, ⁶Department of Computer Science, Vanterbilt University, Nashville, TN, United States, ⁷Otto von Guericke University, Magdeburg, Germany, ⁸AGH University of Science and Technology, Krakow, Poland, ⁹Institute of Cancer Research, London, United Kingdom, ¹⁰CUBRIC, Cardiff University, Cardiff, United Kingdom, ¹¹University of Southern California, Los Angeles, CA, United States, ¹²NYU Langone, New York, NY, United States, ¹³Center for Biomedical Imaging Research, Department of Biomedical Engineering, Tsinghua University, Beijing, China, ¹⁴Istanbul Technical University, Istanbul, Turkey, ¹⁵Harvard Medical School, Boston, MA, United States, ¹⁶Centre for Medical Engineering, King's College London, London, United Kingdom

Diffusion-weighted magnetic resonance imaging (DW-MRI) is a critical modality that allows characterization of microstructure of the nervous tissue in the human brain. Recent multi-parametric acquisitions expand parameter space to b-values, gradient directions, inversion and echo times. The required long scanning time could be shortened by acquiring at lower resolutions while superesolving the images during post-processing. This work embodies the evaluation of an open challenge where the objective was to upsample multi dimensional data encoding simultaneously T1, T2* and diffusion contrast to the natively acquired voxel resolution from two different down-sampled sets of the data (isotropic down-sampled and anisotropic down-sampled).

Sisi Li¹, Yishi Wang², Zhangxuan Hu³, Zhe Zhang⁴, Bing Wu³, and Hua Guo^1
1Center for Biomedical Imaging Research, Beijing, China, ²Philips Healthcare, Beijing, China, ³GE Healthcare, MR Research China, Beijing, China, ⁴China National Clinical Research Center for Neurological Diseases, Beijing Tiantan Hospital, Capital Medical University, Beijing, China

Diffusion tensor imaging (DTI) holds great potential to aid the diagnosis of spinal cord pathologies. Single shot EPI (SS-EPI) is mostly implemented for DTI but is significantly limited in clinics by susceptibility inhomogeneity-induced distortions. This is especially problematic in the thoracic region for increased inhomogeneities near lungs. To solve this problem, multi-shot EPI and reduced-FOV methods have been proposed. However, these methods cannot remove distortions completely. In this study, we achieved high-fidelity DTI of the thoracic spinal cord using a distortion-free MS-EPI technique, Point-Spread-Function Encoded EPI (PSF-EPI). Both the efficacy of PSF-EPI in distortion correction and quantitative reproducibility were evaluated.

Jan Martin¹, Alexis Reymbaut², Michael Uder³, Frederik Bernd Laun³, and Daniel Topgaard^1
1Lund University, Lund, Sweden, ²Random Walk Imaging AB, Lund, Sweden, ³Institute of Radiology, University Hospital Erlangen, Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU), Erlangen, Germany

Diffusion-relaxation correlation NMR methods have recently received attention from the medical MRI community for their ability to characterize microstructure and local chemical composition in complex tissues containing multiple subvoxel pools of water. We here implement 6D $$$\bf{D}$$$-$$$R_1$$$-$$$R_2$$$ distribution imaging of the human brain using a 20-min acquisition protocol combining EPI signal read-out and tensor-valued diffusion encoding with varying repetition- and echo times. Monte Carlo data inversion yields nonparametric distributions, statistical descriptors, and orientation-resolved diffusion and relaxation properties of white matter fiber bundles that are in good agreement with previous results from less exhaustive 4D and 5D protocols.

Seung-Yi Lee¹, Briana Meyer¹, Shekar Kurpad², and Matthew Budde^2
1Biophysics, Medical College of Wisconsin, Milwaukee, WI, United States, ²Neurosurgery, Medical College of Wisconsin, Milwaukee, WI, United States

Sagittal diffusion and perfusion magnetic resonance imaging can provide unique contrast relevant to changes in the acute spinal cord injury. However, sagittal diffusion imaging with echo planar imaging often results in severe motion and susceptibility artifacts. To improve the quality of imaging, we propose a higher-order diffusion preparation combined with a fast spin echo readout. The proposed method demonstrates high quality imaging free from artifacts. Further optimization achieved more accurate diffusivity measurements independent from direction of the cord. Lastly, we show matching diffusion and perfusion maps of the injured cord, highlighting clear spatial differences in microstructure and vasculature injuries.

Hong-Hsi Lee¹, Els Fieremans¹, Jiangyang Zhang¹, and Dmitry S Novikov^1
1New York University School of Medicine, New York, NY, United States

Non-Cartesian MRI enables efficient coverage in k-space and is leveraged to accelerate acquisitions of images in multiple contrasts. However, denoising such data is non-trivial, since the noise statistics is neither independent nor normally distributed in reconstructed images. Here, we propose a random-matrix-theory-based denoising and noise-mapping pipeline applicable to MRI of any non-Cartesian k-space sampling. We demonstrate the denoising pipeline on diffusion MRI data, including a numerical phantom and ex vivo mouse brain data in radial trajectories. The proposed pipeline robustly estimates the noise level, removes the noise, and corrects the bias in parametric maps of diffusion and kurtosis metrics.

Shreyas Fadnavis¹, Joshua Batson², and Eleftherios Garyfallidis^3
1Intelligent Systems Engineering, Indiana University Bloomington, Bloomington, IN, United States, ²Chan Zuckerberg Biohub, San Francisco, CA, United States, ³Indiana University Bloomington, Bloomington, IN, United States

Diffusion MRI (dMRI) is a promising tool for evaluating the spinal cord in health and disease, however low SNR can impede accurate, repeatable, quantitative measurements. Here, we apply a recently proposed denoiser, Patch2Self, that strictly suppresses statistically independent random fluctuations in the signal originating from various sources of noise. Typical spinal cord dMRI scans have a smaller number of gradient directions (10-20) making PCA based 4D denoisers (require at least 30) inapplicable. Using self-supervised learning, Patch2Self addresses these issues which we quantitatively show with an improvement in repeatability and conspicuity of pathology in the spinal cord.

Geraline Vis¹, Markus Nilsson¹, and Filip Szczepankiewicz^1
1Diagnostic Radiology, Clinical Sciences Lund, Lund University, Lund, Sweden

Strong spherical diffusion encoding enables visualization of isotropically restricted regions in the human brain, believed to resemble densely packed cells. As this forces imaging in a low SNR regime, visualization has been limited to a low resolution to avoid deleterious signal bias caused by the noise floor. In this work, we propose a novel method based on super-resolution reconstruction to enable high-resolution visualization of isotropically restricted diffusion in human brain in vivo. We show that our method is superior over conventional methods with acquisition times that are compatible with clinical routine.

Alard Roebroeck¹, Benedikt A Poser¹, and Francisco J Fritz^2
1Department of Cognitive Neuroscience, Faculty of Psychology & Neuroscience, Maastricht University, Maastricht, Netherlands, ²Department of Systems Neurosciences, University Medical Center Hamburg-Eppendorf, Hamburg, Germany, Hamburg, Germany

We compare MESMERISED and standard PGSE for their signal-to-noise efficiency in b-value = 7000 s/mm² (b7k) dMRI at 7T and their effectiveness in supporting microstructure modeling. At current state-of-the-art gradient performance, MESMERISED’s SAR and gradient duty cycle efficiency allows it to outperform PGSE for 7T high-b dMRI and provide highly efficient and effective microstructure modeling. Its added capacity to explore diffusion times, super-accelerate quantitative T₁ mapping, and also perform qT₂ and B₁+ mapping, makes it highly useful for quantitative multi-contrast and diffusion MRI.

Merry Mani¹, Vincent Magnotta², and Mathews Jacob^2
1Radiology, University of Iowa, Iowa City, IA, United States, ²University of Iowa, Iowa City, IA, United States

We propose a new acceleration and reconstruction method for under-sampled multi-shot multi-shell dMRI. The method makes use of incoherent under-sampling in the joint k-q domain to achieve high acceleration. We develop a new model-based reconstruction that jointly recovers missing q-space points, by utilizing a q-space manifold prior that is pre-learned using deep learning.  The proposed method is shown to accurately recover the DWIs from 8-fold accelerated multi-shell data. The reconstruction error is shown to be less than 3%. The proposed method enables utilization of multi-shot EPI trajectories for diffusion microstructure and connectivity studies requiring multi-shell coverage, without prolonging scan time.

Daan Christiaens^1,2, Maximilian Pietsch¹, Lucilio Cordero-Grande¹, Anthony N Price^1,3, Jana Hutter^1,3, Emer Hughes¹, Serena J Counsell¹, Joseph V Hajnal^1,3, and J-Donald Tournier^1,3
1Centre for the Developing Brain, School of Biomedical Engineering and Imaging Sciences, King's College London, London, United Kingdom, ²Department of Electrical Engineering (ESAT-PSI), KU Leuven, Leuven, Belgium, ³Biomedical Engineering Department, School of Biomedical Engineering and Imaging Sciences, King's College London, London, United Kingdom

Modelling tissue microstructure during fetal and neonatal brain development relies on robustly sampling diverse, low-SNR image contrast in moving subjects. Here, we extend slice-to-volume reconstruction to multi-echo diffusion MRI, in order to correct subject motion across the combined diffusion-T₂*-weighted signal. Our results suggest that the spin echo and gradient echo dMRI signal have different trends across gestational age. Tissue microstructure models could therefore leverage these multidimensional data to reveal new insights in brain development.