Kevin Moulin^1,2,3, Mike Loecher^1,2, Matthew J Middione^1,2, and Daniel B Ennis^1,2,3
1Department of Radiology, Stanford University, Stanford, CA, United States, ²Department of Radiology, Veterans Administration Health Care System, Palo Alto, CA, United States, ³Cardiovascular Institute, Stanford University, Stanford, CA, United States

Traditional diffusion encoding waveforms are usually composed of two symmetric trapezoidal gradients (TRAP). Shape-free arbitrary (ARB) gradient waveforms offer a higher b-value than ARBs. They can be designed analytically and symmetric or numerically and asymmetric. The objectives of this work were to analyze the performances of ARB and TRAP for asymmetric numerically designs and for symmetric analytically designed diffusion encoding gradient waveforms. 

Hong Jiang¹ and Daniel Topgaard^1
1Physical Chemistry, Lund University, Lund, Sweden

Diffusion encoding with either oscillating gradients or as a function of the "shape" of the b-tensor have both recently found powerful applications for characterization of tissue microstructure in clinical MRI. We here propose a simple scheme based on the "double rotation" technique in solid-state NMR spectroscopy to generate a family of gradient waveforms enabling comprehensive sampling of the multidimensional space defined by the tensor-valued encoding spectrum with special emphasis on the frequency and shape dimensions. The approach is demonstrated by microimaging experiments on phantoms with well-defined anisotropy and restriction properties, thereby paving the way for future clinical implementations.

Samo Lasic^1,2 and Henrik Lundell^1
1Danish Research Centre for Magnetic Resonance, Centre for Functional and Diagnostic Imaging and Research, Copenhagen University Hospital Hvidovre, Copenhagen, Denmark, ²Random Walk Imaging, Lund, Sweden

Ellipsoidal tensor encoding (ETE) with independent control of spectral anisotropy (SA) and tuning provides two distinct encoding frequency windows in a single experiment and yields distinctly different signal signatures for compartments with different anisotropic time-dependent diffusion. ETE can be orientation invariant, depending on SA and restriction geometry. Signal orientation variation minima depends on size relative to tuning but not on orientation dispersion. This popery could be useful for quick size estimation and geometry detection. Such encoding strategy could potentially provide new contrasts sensitive to specific pathological variations.

Anders Garpebring^1
1Radiation Sciences, div. Radiation Physics, Umeå University, Umeå, Sweden

Non-parametric diffusion tensor distribution estimation is very computationally expensive and can require several hours of processing for a single 3D volume. In this work a new formulation of the estimation problem is developed as well as a new more efficient algorithm. The results show that the computational times can be reduced to minutes or even seconds rather than hours. Thus, making these types of analysis suitable also in a clinical setting.  

Kristin Koller¹, Chantal MW Tax^1,2, Dmitri Sastin^1,3,4, and Derek K Jones^1,5
1Cardiff University Brain Research Imaging Centre, Cardiff, United Kingdom, ²Image Sciences Institute, University Medical Center Utrecht, Utrecht, Netherlands, ³Department of Neurosurgery, University Hospital of Wales, Cardiff, United Kingdom, ⁴BRAIN Biomedical Research Unit, Health & Care Research Wales, Cardiff, United Kingdom, ⁵Mary MacKillop Institute for Health Research, Australian Catholic University, Melbourne, Australia

Diffusional variance decomposition (DIVIDE) is a promising state-of-the-art approach to measure microscopic microstructure using ‘free’ gradient waveforms (including spherical and linear tensor encoding).  Here, in a cohort of 6 participants scanned 5 times each with the DIVIDE protocol, we quantify the microscopic FA (μFA) and isotropic and anisotropic diffusional variance (MKi and MKa) across scans, thus demonstrating high test-retest repeatability.

Kentaro Akazawa¹, Koji Sakai¹, Tomoaki Kitaguchi¹, Tomonori Toyotsuji¹, Thorsten Feiweier ², Hiroshi Imai³, and Kei Yamada^1
1Radiology, Kyoto Prefectural University of Medicine, Kyoto, Japan, ²Siemens Healthcare GmbH, Erlangen, Germany, ³Siemens Healthcare K.K., Shinagawa, Japan

The mixing time for double diffusion encoding (DDE) should be set low to reduce the relaxation effects but also high enough for estimating microscopic fractional anisotropy. We tested the adequacy of the mixing time of 30 msec by comparing acquisitions with parallel and anti-parallel directions as well as with orthogonal and collinear directions. Relatively short mixing time for our cohort was adequate to evaluate the microscopic fractional anisotropy not only in the normal brain area of the white matter and the central gray matter, but also in pathologically abnormal areas such as brain tumors.

Rafael Neto Henriques¹, Sune Nørhøj Jespersen^2,3, and Noam Shemesh^1
1Champalimaud Research, Champalimaud Centre for the Unknown, Lisbon, Portugal, ²Center of Functionally Integrative Neuroscience (CFIN) and MINDLab, Clinical Institute, Aarhus University, Aarhus, Denmark, ³Department of Physics and Astronomy, Aarhus University, Aarhus, Denmark

Multidimensional diffusion encoding (MDE) has been gaining attention for its potential to describe tissue microstructure with enhanced specificity by resolving kurtosis sources, albeit with significant assumptions (no time dependence, no intra-compartmental kurtosis). Correlation Tensor Imaging (CTI) has been introduced as a novel methodology capable of resolving kurtosis sources without relying on a-priori assumptions. Here, we harnessed CTI to validate the accuracy of tensor-valued MDE metrics and assess the importance of intra-compartmental kurtosis (K_intra) in tissues. Our results reveal that K_intra is non-negligible and skews the estimates of tensor-valued MDE approaches, even in the absence of detectable diffusion time dependence.

Rafael N Henriques¹, Sune N. Jespersen^2,3, Derek K. Jones^4,5, and Jelle Veraart^6
1Champalimaud Research, Champalimaud Centre for the Unknown, Lisbon, Portugal, ²Department of Clinical Medicine, Aarhus University, Aarhus, Denmark, ³Department of Physics and Astronomy, Aarhus University, Aarhus, Denmark, ⁴School of Psychology, Cardiff University, Cardiff, United Kingdom, ⁵Mary MacKillop Institute for Health Research, Australian Catholic University, Melbourne, Australia, ⁶Center for Biomedical Imaging, NYU Grossman School of Medicine, New York, NY, United States

The general utility of Diffusion Kurtosis Imaging (DKI) is challenged by its  poor robustness to imaging artifacts and thermal noise that often lead to implausible kurtosis values. A robust scalar kurtosis index can be estimated from powder-averaged diffusion-weighted data. We introduce a novel DKI estimator that uses this scalar kurtosis index as a proxy for the mean kurtosis to regularize the fit. The regularized DKI estimator improves the robustness and reproducibility of the kurtosis metrics and results in parameter maps with enhanced quality and contrast; thereby promoting the wider use of DKI in clinical research and potentially diagnostics. 

Alexandru V Avram^1,2, Qiyuan Tian³, Qiuyun Fan³, Susie Y Huang^3,4, and Peter J Basser^1
1Eunice Kennedy Shriver National Institute of Child Health and Human Development, National Institutes of Health, Bethesda, MD, United States, ²Center for Neuroscience and Regenerative Medicine, The Henry Jackson Foundation, Bethesda, MD, United States, ³Athinoula A. Martinos Center for Biomedical Imaging, Department of Radiology, Massachusetts General Hospital, Harvard Medical School, Charlestown, MA, United States, ⁴Harvard-MIT Division of Health Sciences and Technology, Massachusetts Institute of Technology, Cambridge, MA, United States

We acquire mean apparent propagator (MAP) MRI data in the human brain using different diffusion times. We characterize the diffusion-time dependence of propagator metrics in vivo and compute the statistical distance between co-registered propagators measured with short and long diffusion times. We derive time-scaling parameters that can assess anomalous diffusion in disordered, fractal-like tissue environments. Preliminary results suggest that the diffusion-time dependence of in vivo MRI signals is strongly modulated by restrictions and hindrances that occur over a range of length scales and could provide new contrasts to quantify structural and architectural differences in healthy and diseased tissues.

Francesco Grussu¹, Kinga Bernatowicz¹, Ignasi Barba², and Raquel Perez-Lopez^1,3
1Radiomics Group, Vall d'Hebron Institute of Oncology, Vall d'Hebron Barcelona Hospital Campus, Barcelona, Spain, ²NMR Lab, Vall d'Hebron Institute of Oncology, Vall d'Hebron Barcelona Hospital Campus, Barcelona, Spain, ³Department of Radiology, Hospital Universitari Vall d'Hebron, Barcelona, Spain

To date, limited attention has been paid to diffusion-weighted (DW) MRI signal modelling of the liver, where new imaging methods are needed to tackle diseases such as cancer. We report on Monte Carlo (MC) simulations run in synthetic hepatic cells to inform the developing of new model-based methods for liver application. We specifically investigate the question: “can cell size and diffusivity be estimated from signal cumulants at fixed diffusion time and realistic SNR?”, and find that the task is feasible for clinical diffusion times and b=0 SNR as low as 20, provided that both apparent diffusivity and kurtosis are considered.

YANLI JIANG¹, Jie Zou¹, FengXian Fan¹, Yuping Bai¹, Jing Zhang¹, and Shaoyu Wang^2
1Department of Magnetic Resonance, LanZhou University Second Hospital, LanZhou, China, ²MR Scientific Marketing, Siemens Healthineers, Shanghai, China

Intravoxel Incoherent Motion (IVIM) is a MRI method that enables simultaneous assessment of diffusion and perfusion. Diffusion Kurtosis imaging (DKI) is based on a mathematical approach that uses a polynomial model with a dimensionless factor called the kurtosis (K). This study evaluated the relationship between IVIM-DKI diffusion models and clinical-pathologic to assess their diagnostic accuracy for staging of hepatic fibrosis. We found that a high level of hepatic fibrosis usually have a higher MK values. A significant negative correlation was observed between FibroScan and MD. In conclusion, compared with IVIM, DKI might be more useful in staging of hepatic fibrosis.

Teddy Xuke Cai^1,2, Nathan Hu Williamson^1,3, Velencia Witherspoon¹, Rea Ravin^1,4, and Peter Basser^1
1Section on Quantitative Imaging and Tissue Sciences, Eunice Kennedy Shriver National Institute of Child Health and Human Development, Bethesda, MD, United States, ²Wellcome Centre for Integrative Neuroimaging, University of Oxford, Oxford, United Kingdom, ³National Institute of General Medical Sciences, Bethesda, MD, United States, ⁴Celoptics, Inc., Rockville, MD, United States

Time-dependent diffusion contains rich information about the tissue microstructure. Conventional methods to measure the time-varying diffusivity probe a single timescale per acquisition, limiting time resolution. Furthermore, access to sub-millisecond timescales is limited by the pulsed gradient hardware. An alternative method is presented here. We extend the static field gradient, Carr-Purcell-Meiboom-Gill cycle by incrementing the $$$\pi$$$-pulse spacings to isolate the on-resonance signal. The resulting spin echo train probes a range of short timescales (50 – 500 microseconds) in one shot and enables a 1-minute time-dependent diffusivity measurement. Proof-of-principle simulations and experimental results on pure liquids and yeast are presented.

Graham Little¹ and Christian Beaulieu^1
1Biomedical Engineering, University of Alberta, Edmonton, AB, Canada

A surface-based deep grey matter segmentation algorithm is proposed that works directly on diffusion images and maps of the brain acquired at a 1.5 mm isotropic resolution.  The method was applied to twenty participants spanning a large age range (6-90 years) resulting in accurate segmentations of the thalamus, caudate, putamen and globus pallidus.  Fractional anisotropy and mean diffusivity showed unique non-linear trajectories across the lifespan. The proposed method avoids the need of problematic coregistration to other scans (anatomical T1) and will accelerate the analysis of microstructural changes of deep grey matter regions with age (or disease) in large populations.

Henrik Lundell¹, Samo Lasič^1,2, Filip Szczepankiewicz^3,4,5, Beata Wereszczyńska⁶, Matthew Budde⁷, Erica Dall'Armellina⁶, Nadira Yuldasheva⁶, Jürgen E. Schneider⁶, and Irvin Teh^6
1Danish Research Centre for Magnetic Resonance, Centre for Functional and Diagnostic Imaging and Research, Copenhagen University Hospital Hvidovre, Hvidovre, Denmark, ²Random Walk Imaging, Lund, Sweden, ³Clinical Sciences, Lund University, Lund, Sweden, ⁴Harvard Medical School, Boston, MA, United States, ⁵Brigham and Women's Hospital, Boston, MA, United States, ⁶Leeds Institute of Cardiovascular and Metabolic Medicine, University of Leeds, Leeds, United Kingdom, ⁷Department of Neurosurgery, Medical College of Wisconsin, Milwaukee, WI, United States

Non-invasive characterization of cardiac microstructure by diffusion MRI has provided insights into the healthy and diseased heart. Multidimensional diffusion encoding (MDE) aims for measurements with independent contrasts for specific effects. We suggest a battery of MDE measurements that probe diffusivity and microscopic anisotropy at different diffusion and echo times. We show a clear effect of time-dependent diffusion but a smaller effect from transversal relaxation.

Sofie Rahbek¹, Kristoffer H. Madsen^2,3, Henrik Lundell², Faisal Mahmood^4,5, and Lars G. Hanson^1,2
1Department of Health Technology, Technical University of Denmark, Kgs. Lyngby, Denmark, ²Danish Research Centre for Magnetic Resonance, Centre for Functional and Diagnostic Imaging and Research, Copenhagen University Hospital Hvidovre, Hvidovre, Denmark, ³Department of Applied Mathematics and Computer Science, Technical University of Denmark, Kgs. Lyngby, Denmark, ⁴Laboratory of Radiation Physics, Odense University Hospital, Odense, Denmark, ⁵Department of Clinical Research, University of Southern Denmark, Odense, Denmark

We propose a novel data-driven method for extraction of tissue-related signal components from high-dimensional MRI data. In this method, the standard non-negative matrix factorization (NMF) has been extended with signal monotonicity constraints suitable for several MR signal types, and is termed the monotonous slope NMF (msNMF). Its applications are here demonstrated using both diffusion-weighted and relaxometry data. The msNMF successfully distinguish areas with different cell densities and levels of white matter intra-myelinic edema, respectively, and is potentially useful for diagnosis and therapy evaluation.

Thijs Dhollander^1,2, Rami Tabbara², Jonas Rosnarho-Tornstrand^3,4, J-Donald Tournier^3,4, David Raffelt², and Alan Connelly^2
1Developmental Imaging, Murdoch Children's Research Institute, Melbourne, Australia, ²Florey Institute of Neuroscience and Mental Health, University of Melbourne, Melbourne, Australia, ³Centre for the Developing Brain, School of Biomedical Engineering and Imaging Sciences, King's College London, London, United Kingdom, ⁴Department of Biomedical Engineering, School of Biomedical Engineering and Imaging Sciences, King's College London, London, United Kingdom

Multi-tissue constrained spherical deconvolution of diffusion MRI data yields white matter fibre orientation distributions, from which a quantitative metric of apparent fibre density can be obtained. Unlike most other diffusion MRI models, this fibre density metric is directly proportional to the diffusion-weighted signal magnitude, and thus intensity normalisation and bias field correction are needed to compare it between subjects in a study. Here we propose an intensity and inhomogeneity correction algorithm for multi-tissue constrained spherical deconvolution results, estimating a bias field and global normalisation in the log-domain. It outperforms a previously proposed approach that did not operate in the log-domain.

Michael J Joseph¹, Derek Pisner², Adam Richie-Halford³, Garikoitz Lerma-Usabiaga⁴, Salim Mansour¹, James D Kent⁵, Anisha Keshavan³, Matthew Cieslak⁶, Erin W Dickie¹, Sebastian Tourbier⁷, Aristotle N Voineskos¹, Theodore D Satterthwaite⁶, Russell A Poldrack⁸, Jelle Veraart⁹, Ariel Rokem¹⁰, and Oscar Esteban^7
1The Centre for Addiction and Mental Health, Toronto, ON, Canada, ²Department of Psychology, University of Texas at Austin, Austin, TX, United States, ³eScience Institute, The University of Washington, Seattle, WA, United States, ⁴Basque Center on Cognition, Brain and Language, Donostia - San Sebastian, Spain, ⁵Neuroscience Program, University of Iowa, Iowa City, IA, United States, ⁶Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA, United States, ⁷Lausanne University Hospital and University of Lausanne, Lausanne, Switzerland, ⁸Department of Psychology, Stanford University, Stanford, CA, United States, ⁹NYU Grossman School of Medicine, New York City, NY, United States, ¹⁰Department of Psychology, The University of Washington, Seattle, WA, United States

We present dMRIPrep, a preprocessing pipeline for diffusion MRI (dMRI) inspired by the approach and wide uptake of fMRIPrep. dMRIPrep reliably and consistently performs on diverse data acquired by different studies. dMRIPrep equips researchers with a reliable and transparent tool developed with the best available engineering practices and neuroimaging standards, maintained as part of NiPreps, a community software framework that ensures long-lasting support and public-interest steering.

Evgenios N. Kornaropoulos^1,2, Stefan Winzeck^2,3, Theodor Rumetshofer¹, Anna Wikstrom¹, Linda Knutsson^4,5, Marta Correia⁶, Pia Sundgren^1,7,8, and Markus Nilsson^1
1Diagnostic Radiology, Lund University, Lund, Sweden, ²Division of Anaesthesia, University of Cambridge, Cambridge, United Kingdom, ³Department of Computing, Imperial College London, London, United Kingdom, ⁴Department of Medical Radiation Physics, Lund University, Lund, Sweden, ⁵Russell H. Morgan Department of Radiology and Radiological Science, Johns Hopkins University, Baltimore, MD, United States, ⁶MRC Cognition and Brain Sciences Unit, University of Cambridge, Cambridge, United Kingdom, ⁷Lund University BioImaging Center, Lund University, Lund, Sweden, ⁸Department of Medical Imaging and Physiology, Skane University Hospital, Lund, Sweden

There are many ways to acquire and process diffusion MRI data. However, it is not known which yields the largest effect sizes in group studies. We evaluated the impact of different acquisitions (3T versus 7T and DTI versus DKI) and post-processing strategies (motion correction, Gibbs-ringing and denoising) on effect size estimates of white matter (WM) injury in patients with Systemic Lupus Erythematosus (SLE). Results showed that fractional anisotropy (FA) from a 3T DTI acquisition yielded the largest effect sizes.

Khoi Minh Huynh^1,2, Wei-Tang Chang², and Pew-Thian Yap^1,2
1Biomedical Engineering, UNC Chapel Hill, Chapel Hill, NC, United States, ²Department of Radiology and Biomedical Research Imaging Center (BRIC), UNC Chapel Hill, Chapel Hill, NC, United States

We evaluated the effectiveness of optimal-shrinkage singular value decomposition (OS-SVD) in denoising multi-channel complex-valued diffusion MRI data prior to image reconstruction and show that it outperforms other state-of-the-art denoising methods.

Nastaren Abad¹, Luca Marinelli¹, Radhika Madhavan¹, and Tom K.F Foo^1
1General Electric Global Research, Niskayuna, NY, United States

Higher spatial and angular resolution is essential in diffusion MRI to resolve fiber and structural ambiguities. However, quantitative measures are confounded by low SNR, particularly at high b-values, compensation of which leads to longer acquisition times. In this study, the fundamental question asked is: Can denoising aid the stability of the measurement in the presence of increasing noise? Model based denoising was used to explore accelerated sampling by evaluating bias developed in qualitative and quantitative end points. Experimental results highlight superior performance, compared to ground truth, in noise and bias reduction in metrics along with structure preservation.