Evren Özarslan^1,2
1Department of Biomedical Engineering, Linköping University, Linköping, Sweden, ²Center for Image Science and Visualization, Linköping University, Linköping, Sweden

Diffusion magnetic resonance has been employed for determining the distribution of net displacements (ensemble average propagator), moments and correlations of net displacements, and the steady-state distribution of magnetized particles. All such quantities are accessible via the diffusion propagator, which characterizes the diffusion process fully. Here, a novel diffusion encoding and data analysis framework is introduced with which the diffusion propagator can be recovered.

Robert Jones¹, Chiara Maffei¹, Qiuyun Fan¹, Jean Augustinack¹, Barbara Wichtmann², Aapo Nummenmaa¹, Susie Huang¹, and Anastasia Yendiki^1
1Radiology, Athinoula A. Martinos Center for Biomedical Imaging, Massachusetts General Hospital and Harvard Medical School, Charlestown, MA, United States, ²Department of Radiology, University Hospital Bonn, Bonn, Germany, Bonn, Germany

We take advantage of the high spatial resolution that is feasible ex vivo on a preclinical 9.4T system to investigate between-bundle differences and within-bundle continuity of microstructural indices. In a human brain sample that has also undergone optical imaging to obtain direct measurements of axonal orientations, we identify regions occupied by motor fibers or association fibers. We collect diffusion-weighted images with two diffusion times and eight q-shells and use them to estimate the parallel and perpendicular diffusion coefficient in each fiber bundle. We show that the diffusion coefficients vary smoothly along each bundle, and have different profiles between bundles.

Ante Zhu¹, J. Kevin DeMarco^2,3, Robert Y. Shih^2,3, Radhika Madhavan¹, Tim Sprenger⁴, Chitresh Bhushan¹, Maureen Hood^2,3, Luca Marinelli¹, Vincent B. Ho^2,3, and Thomas K.F. Foo^1
1GE Global Research, Niskayuna, NY, United States, ²Uniformed Services University of the Health Sciences, Bethesda, MD, United States, ³Walter Reed National Military Medical Center, Bethesda, MD, United States, ⁴GE Healthcare, Stockholm, Sweden

Diffusivity measurements of the human brain have been shown to increase at short diffusion time by using oscillating gradient spin echo (OGSE). The time-dependent diffusivity averaged in large white matter parcels has been measured to assess microstructure characteristics. In this study, we assessed time-dependent diffusivity in finer white matter parcels to study the regional microstructure differences. In four healthy volunteers, we consistently observed higher parallel diffusivity values and higher increasing rate over OGSE frequencies in two sub-parcels of the genu and two of the splenium in the corpus callosum, compared to other regions, indicating regional microstructure differences of the brain.

Kevin B Borsos^1,2, Desmond HY Tse², Paul I Dubovan^1,2, and Corey A Baron^1,2,3
1Department of Medical Biophysics, Western University, London, ON, Canada, ²Centre for Functional and Metabolic Mapping, Western University, London, ON, Canada, ³Robarts Research Institute, Western University, London, ON, Canada

Frequency dependent diffusion kurtosis has historically been difficult to measure with oscillating gradient spin echo (OGSE) sequences and even more so without the use of gradient insert coils. Here we present a new OGSE gradient waveform and determine the optimal frequency to observe kurtosis differences between PGSE and OGSE acquisitions using a conventional gradient system. Using this method we present in vivo differential kurtosis maps based on the frequency dependence.  

Samuel Bryce Jones¹, Emre Kopanoglu², Chantal Tax², and Derek Jones^2
1Radyr Comprehensive School, Cardiff, United Kingdom, ²CUBRIC, School of Psychology, Cardiff, United Kingdom

This work asks a very simple but important question:  To what extent can we recover lost SNR by signal averaging in a standard diffusion tensor MRI experiment?  Based on the theory that if SNR of a signal is reduced by a factor (1/m), then m² signal averages will recover that SNR,  the diffusion MRI literature often assumes that averaging is the 'magic bullet'. Here, using Monte Carlo simulations, we show that this only true to a certain extent and, despite averaging, low SNR data results in biased DTI estimates.

Maryam Afzali¹, Hans Knutsson^2,3, Evren Özarslan^2,3,4, and Derek K Jones^1,5
1Cardiff University Brain Research Imaging Centre (CUBRIC), School of Psychology, Cardiff University, Cardiff, United Kingdom, ²Department of Biomedical Engineering, Linköping University, Linköping, Sweden, ³Center for Medical Image Science and Visualization, Linköping University, Linköping, Sweden, ⁴These authors share last authorship, Linköping, Sweden, ⁵These authors share last authorship, Cardiff, United Kingdom

Numerous applications in diffusion MRI involve computing the orientationally-averaged diffusion-weighted signal. Most approaches assume that the gradient vectors are uniformly distributed on a sphere, computing the orientationally-averaged signal through arithmetic averaging. One challenge is that not all acquisition schemes have gradient vectors distributed over perfect spheres. Alternative averaging methods include: weighted signal averaging; spherical harmonic; and Mean Apparent Propagator MRI (MAP-MRI). Here, these methods are compared under different signal-to-noise (SNR) realizations. With dense and isotropically-distributed sampling, all methods give comparable results.  As the SNR and number of data points are reduced, MAP-MRI-based approaches give pronounced improvements over the other methods.

Sean C Epstein¹, Timothy J.P. Bray², Margaret A. Hall-Craggs², and Hui Zhang^1
1Department of Computer Science & Centre for Medical Image Computing, University College London, London, United Kingdom, ²Centre for Medical Imaging, University College London, London, United Kingdom

We present a novel computational method for quantitative assessment of experimental design choices for diffusion-weighted imaging (DWI). This approach is motivated by the observation that real-world tasks (e.g. clinical classification) are assessed by metrics (e.g. AUC) that depend non-trivially on the accuracy and precision of DWI-derived parameters. The proposed method enables, for the first time, the assessment of such metrics in the course of computational experimental design. Evaluation with clinical datasets demonstrates its ability to accurately predict real-world task performance for a range of experimental designs. Illustrative use cases are presented to demonstrate its advantages over existing computational approaches.

Tommaso Ciceri¹, Alberto De Luca^2,3, Filippo Arrigoni^1,4, and Denis Peruzzo^1
1NeuroImaging Unit, Scientific Institute IRCCS Eugenio Medea, Bosisio Parini, Italy, ²Neurology Department, UMC Utrecht, Utrecht, Netherlands, ³PROVIDI Lab, Image Sciences Institute UMC Utrecht, Utrecht, Netherlands, ⁴Radiology Unit, Fatebenefratelli Hospital, Milan, Italy

Complementary aspects of the tissue microstructure can be addressed using different models to quantify diffusion MRI. However, there is no consensus on a common acquisition scheme within a clinical feasible time to support the quantification of multiple models. We acquired a large dataset with multiple b-values and directions, and recursively subsampled it to identify the minimum acquisition scheme (MAS) for each model. Finally, we investigated the impact of the MAS on the parameter estimates in the main fiber bundles of the brain. This work supports the transition of advanced analyses to the clinical practice.

Xin Li¹, Eric M. Baker¹, Brendan Moloney¹, Cory Wyatt¹, Eric Baetscher¹, Erin W. Gilbert², Charles S. Springer¹, Alexander R. Guimaraes^1,3, and William D. Rooney^1
1Advanced Imaging Research Center, Oregon Health & Science University, Portland, OR, United States, ²Surgery, Oregon Health & Science University, Portland, OR, United States, ³Diagnostic Radiology, Oregon Health & Science University, Portland, OR, United States

The optimal data acquisition strategy for accurate metabolic and cytometric parameter extraction using a digital DWI library is investigated using simulations under the constant data-acquisition time (same number of b-values) constraint.  Base on DWI data from pancreatic tail tissue and using a seven b-value approach, the optimal maximum b-value is found to be data signal-to-noise ratio dependent, and generally should be below 3,000 s/mm².  In addition, evenly-spaced b-values are generally within the optimal b-spacing range.  

Noam Omer¹, Neta Stern¹, Tamar Blumenfeld-Katzir¹, Meirav Galun², and Noam Ben-Eliezer^1,3,4,5,6
1Department of Biomedical Engineering, Tel-Aviv University, Tel-Aviv, Israel, ²Department of Computer Science and Applied Mathematics, Weitzman institute of science, Rehovot, Israel, ³Department of Orthopedics, Shamir Medical Center, Zerifin, Israel, ⁴Sackler Faculty of Medicine, Tel-Aviv University, Tel-Aviv, Israel, ⁵Sagol School of Neuroscience, Tel-Aviv University, Tel-Aviv, Israel, ⁶Center for Advanced Imaging Innovation and Research (CAI2R), New-York University Langone Medical Center, New York, NY, United States

Multicomponent T₂ analysis (mcT₂) yields a voxel-wise distribution of T₂ values, which can be used to estimate sub-voxel information such as myelin content. Producing such data, however, remains challenging due to the large ambiguity in the T₂ space. We present a data-driven approach for mcT₂ analysis, which learns the anatomy in question and identifies microscopic tissue-specific features as a preprocessing step. It then utilizes them for analyzing each voxel locally using a designated optimization scheme. Experiments in human brain data show reproducible myelin content estimations at clinical settings without any prior assumptions.

Chantal Tax^1,2, Edwin Versteeg¹, Dennis J.W. Klomp¹, Martijn F. Froeling¹, Alberto de Luca¹, and Jeroen C.W. Siero^1,3
1University Medical Center Utrecht, Utrecht, Netherlands, ²CUBRIC, Cardiff University, Cardiff, United Kingdom, ³Spinoza Centre for Neuroimaging Amsterdam, Amsterdam, Netherlands

Diffusion weighting is achieved by the application of external field gradients typically for tens of milliseconds, during which the signal substantially decays due to inherent T₂ relaxation. This work focuses on the benefits of strong gradients - here provided by a gradient head insert - for high SNR and short TE diffusion imaging at 7T. Proof-of-principle images show that a short TE (21 ms) at a b-value of 1000 s/mm² is achievable at 7T using an EPI-readout. 

Gregory Lemberskiy¹, Santiago Coelho¹, Thomas K.F. Foo², Radhika Madhavan², Luca Marinelli², Jaemin Shin³, Els Fieremans¹, and Dmitry S Novikov^1
1Radiology, NYU School of Medicine, New York, NY, United States, ²GE Research, Niskayuna, NY, United States, ³GE Healthcare, New York, NY, United States

A multishell diffusion neuro protocol at 1.5 mm isotropic resolution was acquired in 10 minutes with a high performance MAGNUS head gradient coil (200 mT/m with 500 mT/m/ms slew) at 3.0 T, and reconstructed using random matrix theory (RMT) denoising at the coil level. The SNR enhancement due to RMT reconstruction, and the nearly distortion-free images due to high slew rate, yield precise diffusion and kurtosis maps, fiber-dispersion, as well as microstructure parameters estimated using 3-compartment Standard Model of diffusion. We draw compartment fractions and diffusivities on fiber-tracts reconstructed based on fiber ODFs estimated by deconvolving model-based local fiber response.

Zejun Wang¹, Bao Wang², Yingchao Liu³, and Ruiliang Bai^1,4
1Key Laboratory of Biomedical Engineering of Ministry of Education, College of Biomedical Engineering and Instrument Science, Zhejiang University, Hangzhou, China, ²Department of Radiology, Qilu Hospital of Shandong University, Jinan, China, ³Department of Neurosurgery, Shandong Provincial Hospital Affiliated to Shandong First Medical University, Jinan, China, ⁴Department of Physical Medicine and Rehabilitation, Interdisciplinary Institute of Neuroscience and Technology, The Affiliated Sir Run Run Shaw Hospital, School of Medicine, Zhejiang University, Hangzhou, China

Vascular water exchange is a highly sensitive marker of BBB dysfunction and a potential biomarker of metabolic activity. In this study, we compared two different MRI methods for vascular water exchange measurement, including shutter speed (SS) DCE-MRI and filter-exchange imaging (FEXI) in high-grade glioma patents. Our results demonstrated consistent vascular water exchange assessments by SS DCE-MRI and FEXI in both normal-appearing white matter and tumor.

Jaume Coll-Font^1,2,3, Shi Chen¹, Robert A Eder¹, and Christopher T. Nguyen^1,2,3
1Cardiovascular Research Center, Cardiology Division, Massachusetts General Hospital, Charlestown, MA, United States, ²Athinoula A. Martinos Center for Biomedical Imaging, Charlestown, MA, United States, ³Harvard Medical School, Boston, MA, United States

Recent advances in in vivo cardiac DTI have enabled rapid acquisition of the required diffusion weighted (DW) images. With the possibility of acquiring more DW images, it is not clear what is the best sampling strategy. Here we evaluate whether there are any differences between acquiring twice as many directions versus duplicating the number of repetitions. Our results indicate that it is marginally better to acquire multiple directions and that the loss of accuracy is produced by the reduction in the total number of DW images.

Samo Lasic^1,2, Henrik Lundell¹, Beata Wereszczyńska³, Matthew Budde⁴, Nadira Yuldasheva³, Filip Szczepankiewicz⁵, Erica Dall’Armellina³, Jürgen E. Schneider³, and Irvin Teh^3
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, ³Leeds Institute of Cardiovascular and Metabolic Medicine, University of Leeds, Leeds, United Kingdom, ⁴Department of Neurosurgery, Medical College of Wisconsin, Milwaukee, WI, United States, ⁵Clinical Sciences, Lund University, Lund, Sweden

Tensor-valued diffusion encoding with simultaneous nulling of velocity, acceleration and concomitant gradients can be applied with high b-values on a preclinical 7T scanner. Results for ex-vivo mouse hearts confirm that time-dependent diffusion can significantly affect estimation of mean diffusivity. The estimated restriction sizes are consistent with results from pig hearts. Signal attenuations at high b-values suggest relatively low microscopic anisotropy and a strong influence of time-dependent diffusion on microstructure characterization.

Tokunori Kimura¹, Kousuke Yamashita¹, and Kouta Fukatsu^1
1Department of Radiological Science, Shizuoka College of Medicalcare Science, Hamamatsu, Japan

 We evaluated our proposed T2wsup-DWI method of reducing CSF-partial volume effects (PVE) on DTI. We assessed the errors in ADC and FA and the comparison of the ADC SNRs with several methods with simulation, and brain study. We clarified that our proposed T2wsup-DWI technique was superior to already proposed water suppression DWI methods of FLAIR and non-b-zero (NZE) methods in both of the ADC-SNR and the reduction effects of CSF-PVE in DTI parameters of ADC, FA, and tractograpy, typically at the portion of the fornix crus or genu of the corpus callosum which are close to the ventricle.

Sajjad Feizollah^1,2 and Christine L. Tardif^1,2,3
1Department of Neurology and Neurosurgery, McGill University, Montreal, QC, Canada, ²McConnell Brain Imaging Center, Montreal Neurological Institute, McGill University, Montreal, QC, Canada, ³Department of Biomedical Engineering, McGill University, Montreal, QC, Canada

High-resolution diffusion-weighted imaging has been used to investigate the microstructure of the cortex in vivo. However, conventional linear diffusion encoding provides limited insight into the underlying microstructural differences between cortical areas. b-Tensor encoding disentangles macroscopic from microscopic anisotropy in voxels with complex fiber geometries. The SNR and thus resolution of these scans are limited by the longer diffusion encoding times. A DWI sequence with b-tensor encoding was implemented at 7T with a spiral readout trajectory and dynamic field monitoring to image cortical microstructure. Microscopic anisotropy maps of the brain are presented at 1.4 mm isotropic, within minimal distortions and blurring.

Jan Malte Oeschger¹, Karsten Tabelow², and Siawoosh Mohammadi^1,3
1Institute of Systems Neuroscience, University Medical Center Hamburg-Eppendorf, Hamburg, Germany, ²Weierstrass Institute for Applied Analysis and Stochastics, Berlin, Germany, ³Department of Neurophysics, Max Planck Institute for Human Cognitive and Brain Sciences, Leipzig, Germany

Five out of eight axial-symmetric Diffusion Kurtosis Imaging (AxDKI) parameters are directly related to biophysical microstructure parameters including intra- and extra-axonal diffusivities, fiber dispersion and axonal-water fraction. Their estimation, however, is biased at small signal-to-noise ratios (SNR). Here, based on simulations, we investigated the Rician noise bias’s effect and its correction (RBC) on estimated AxDKI and biophysical parameters at varying SNRs for the standard and AxDKI model. Our study suggests AxDKI to be better than standard DKI, here least biased AxDKI estimators were produced with RBC while for biophysical parameters results were branch-dependent (SNR≥23 with RBC and SNR≥33 without RBC).

Nico J. J. Arezza^1,2 and Corey A. Baron^1,2
1Medical Biophysics, Western University, London, ON, Canada, ²Centre for Functional and Metabolic Mapping, Robarts Research Institute, London, ON, Canada

Microscopic fractional anisotropy (μFA) quantifies diffusion anisotropy independent of fiber orientation, giving it high specificity to microstructure. However, it is underestimated in brain regions containing cerebrospinal fluid (CSF) partial volumes. Here, we investigated two methods to reduce CSF partial volume effects: the free-water elimination method and a shifted kurtosis method. Compared to diffusion tensor imaging and diffusion kurtosis imaging, these techniques produced more accurate diffusivity estimates in simulations and higher μFA estimates in brain tissue. This preliminary work demonstrates the potential for CSF-suppressing techniques to improve μFA estimation in regions where CSF partial volume effects are prevalent.

Shujun Lin¹, Bradley Sutton², Richard Magin¹, Aaron Anderson², and Dieter Klatt^1
1Richard and Loan Hill Department of Bioengineering, University of Illinois at Chicago, Chicago, IL, United States, ²Beckman Institute, University of Illinois at Urbana-Champaign, Urbana-Champaign, IL, United States

Simultaneous acquisition of diffusion tensor imaging (DTI) and magnetic resonance elastography (MRE) has been proven feasible in a preliminary study of in vivo human brain. However, the experimental parameters have to be optimized in order to prevent mutual interferences of DTI and MRE acquisitions. We identified in simulations three experimental parameter sets for in vivo human brain DTI-MRE that we classify as good, moderate and poor with regard to optimization and present a pilot study using these parameter sets. The experimental results verify the simulations as we found the best performance of DTI-MRE for the good parameters set.