Kawin Setsompop¹, Berkin Bilgic¹, Aapo Nummenmaa¹, Qiuyun Fan¹, Stephen F Cauley¹, Susie Huang¹, Itthi Chatnuntawech², Yogesh Rathi³, Thomas Witzel¹, and Lawrence L Wald^1
1Martinos Center for Biomedical Imaging, Charlestown, MA, United States, ²Massachusetts Institute of Technology, Cambridge, MA, United States, ³Brigham and Women's Hospital, Boston, MA, United States

Sub-millimeter in vivo diffusion imaging (DI) is extremely challenging. In this work, we propose a new slice encoding strategy that enables 700μm isotropic whole-brain DI at b>1000s/mm2 in a reasonable time. Specifically, we introduce and validate the SLIDER-SMS method, which combines SMS imaging with super resolution via sub-voxel shift in the slice direction. In order to additionally gain high in-plane resolution, we use ZOOPPA to reduce the phase FOV and distortion. We demonstrate that SLIDER-SMS with ZOOPPA can provide high quality 700um whole-brain DI data in 40min, where MB-2 and 3x-SLIDER provides a √6 SNR gain compared to conventional DI.

Bertram Jakob Wilm¹, Signe Johanna Vannesjo¹, and Klaas Paul Pruessmann^1
1University and ETH Zurich, Zurich, Zurich, Switzerland

Single-shot diffusion-weighted MRI of the brainstem is hampered by B0 off-resonance distortions, a problem that is emphasized at high field strength (7T). To address this problem, we implemented a reduced FOV approach by spin-echo selection using second-order shim fields for multi-slice imaging. The method allowed for effective FOV reduction and thereby for robust and SNR efficient diffusion imaging of the brain stem at high field strength.

Myung-Ho In¹, Posnansky Oleg¹, and Oliver Speck^1
1Biomedical Magnetic Resonance, Otto-von-Guericke University, Magdeburg, Germany

For high-resolution diffusion-weighted imaging, which is used for microstructural characterization of human brain in research as well as clinical applications, single-shot echo-planar imaging may be not suitable due to severe T2 blurring and susceptibility- and eddy-current-induced geometric distortions, especially at ultra-high field. Several multi-shot approaches have been proposed to mitigate such effects. In this study, a point-spread function based diffusion-weighted imaging approach is newly proposed. In contrast to multi-shot approaches, this method doesn’t suffer any T2 blurring and geometric distortions and enables a clear and detailed delineation of human brain structures in-vivo.

Jonathan I. Sperl¹, Tim Sprenger^1,2, Ek T. Tan³, Marion I. Menzel¹, Christopher J. Hardy³, and Luca Marinelli^3
1GE Global Research, Munich, BY, Germany, ²IMETUM, Technical University Munich, Munich, BY, Germany, ³GE Global Research, Niskayuna, NY, United States

Spherical Deconvolution (SD) is a model-based approach to retrieve angular fiber information from HARDI data. This work proposes to apply concepts and algorithms known in the context of Compressed Sensing, namely L1-sparsity and minimum Total Variation, to regularize the numerically ill-posed inverse problem addressed by SD. Moreover, the proposed method allows undersampling the data to substantially speed up the data acquisition by a factor of three. Improved fiber peak detection and tractography results are shown for simulated as well as for in vivo human subject data.

Martin Genet¹, Constantin von Deuster^1,2, Christian T Stoeck^1,2, and Sebastian Kozerke^1,2
1Institut for Biomedical Engineering, ETHZ, Zurich, Switzerland, ²Imaging Sciences and Biomedical Engineering, KCL, London, United Kingdom

Recent advances in cardiac diffusion tensor imaging (DTI) have enabled robust imaging of the in-vivo human heart, allowing the non-invasive assessement of myofiber and myosheet orientations. However, in view of the relatively low scan efficiency and scan time constraints in-vivo, cardiac coverage and signal-to-noise ratio are limited. The objective of the present work is to develop an approach to (i) extract the most significant features from noisy myofiber and myosheet angle maps measured with in-vivo DTI, and (ii) extrapolate the data across the entire left ventricle based on a limited number of acquired short-axis views.

Olivier Reynaud^1,2, Kerryanne V Winters^1,2, Dung Minh Hoang^1,2, Youssef Zaim Wadghiri^1,2, Dmitry S Novikov^1,2, and Sungheon Gene Kim^1,2
1Center for Advanced Imaging Innovation and Research (CAI2R), Department of Radiology, New York University School of Medicine, New York, NY, United States, ²Bernard and Irene Schwartz Center for Biomedical Imaging, Department of Radiology, New York University School of Medicine, New York, NY, United States

In this study, we combine OGSE and PGSE measurements to probe the short diffusion time ([1.2-30]ms) dependence of ADC in a GBM. Based on tumor microenvironment modeling, parametric maps of Extracellular Space (ECS), Cell Radius and ECS free diffusivity are derived, showing high ECS heterogeneity inside the tumor and positive correlation of ECS with PGSE-ADC at long diffusion times. Fitting on ex vivo ADC measurements performed with a MR microcoil on 100um thick fixed brain slices are compared to cellular membrane immuno-staining (GLUT1) to validate ECS quantification. Regions of high/low ECS correlate with regions of low/high cell volume fraction.

Jonathan Scharff Nielsen¹, Tim B Dyrby¹, and Henrik Lundell^1
1Danish Research Centre for Magnetic Resonance, Copenhagen University Hospital Hvidovre, Hvidovre, Denmark

In this work we explore the possibility for performing micro-anisotropy measurements on fibers with sharp undulations with double diffusion encoding (DDE) at short length scales using circularly polarized oscillating gradient spin echo (CP-OGSE). We show in simulations that the method is sensitive to anisotropy and that the curvature can be assessed from the signals frequency dependence. MRI experiments on post mortem tissue shows similar behavior. We suggest that the method can provide a novel contrast for gray matter complexity.

Silvia De Santis^1,2, Derek K Jones¹, and Alard Roebroeck^2
1CUBRIC Cardiff University, Cardiff, United Kingdom, ²Maastricht University, Maastricht, Netherlands

The purpose of this work is to analyse the impact of diffusion time in estimating axonal density and axonal diameters using STEAM diffusion at 7T. Using a 2-ways ANOVA, we demonstrate that the estimates of axonal density obtained at different diffusion times are significantly different. In addition, by accounting for the diffusion time dependency of the extra-axonal signal, we show that estimates of axonal diameter in agreement with histology can be obtained.

Ivana Drobnjak¹, Hui Zhang¹, Andrada Ianus¹, Enrico Kaden¹, and Daniel C Alexander^1
1Centre for Medical Image Computing, Department of Computer Science, University College London, London, London, United Kingdom

Imaging axon diameter could provide insight into basic brain operation as well as neuronal diseases that alter axon diameter distribution. This work aims to identify the optimal diffusion MRI sequence that maximizes sensitivity to axon diameter in practical applications. The study shows that although standard PGSE always gives maximum sensitivity for the simple case of gradients perfectly perpendicular to straight parallel fibres, low frequency trapezoidal OGSE provides higher sensitivity in real-world scenarios where fibres have unknown or dispersed orientation. This is a novel fundamental insight into both sequences and their benefits for imaging axon diameter.

Vera Catharina Keil¹, Burkhard Maedler², Hans Heinz Schild¹, and Dariusch Reza Hadizadeh^1
1Radiology, UK Bonn, Bonn, NRW, Germany, ²Radiology MRI Unit, PHILIPS Healthcare, Hamburg, Germany

A restriction to the clinical application of intravoxel-incoherent motion (IVIM) "microperfusion" MRI in the brain is the ill-posed problem to deconvolute the multi-exponential process of water diffusion. We compared mono- and bi-exponential fitting methods with a recently established semi-continuous multi-exponential non-negative least squares function diffusion model (32 b-values, 0 - 2000 s/mm²) and T1-weighted dynamic contrast-enhanced MRI in 30 patients and 9 healthy controls. Perfusion fractions and ADC values varied significantly between all approaches and were not comparable to results of T1-DCE MRI. This study discusses possible effects of fitting choice to be considered when appIying IVIM in the CNS.n. We compared mono- and bi-exponential fitting methods with results of a recently established semi-continuous multi-exponential non-negative least squares function diffusion model (32 b-values, 0 - 2000 s/mm²) and T1-weighted dynamic contrast-enhanced MRI in 30 patients and 9 healthy controls. Perfusion fractions vPF and ADC values varied significantly between all approaches and were not comparable to results of T1-DCE MRI. This study discusses possible effects of fitting choice to be considered using IVIM in the CNS.

Ben Jeurissen¹, Jacques-Donald Tournier^2,3, and Jan Sijbers^1
1iMinds-Vision Lab, Dept. of Physics, University of Antwerp, Antwerp, Belgium, ²Centre for the Developing Brain, King's College London, London, United Kingdom, ³Dept. of Biomedical Engineering, King's College London, London, United Kingdom

Advanced processing of diffusion-weighted (DW) MRI often relies on properly aligned anatomical scans and their segmentations to identify specific tissue types, which can prove challenging due to EPI distortions. We introduce a fast, data-driven method for tissue-type segmentation of multi-shell DW MRI images based on non-negative matrix factorization. Experiments show that our method provides good quality segmentation of CSF, GM and WM, straight from the raw DW and without any spatial priors. We show that the proposed technique can be used to estimate response functions for multi-shell, multi-tissue constrained spherical deconvolution, removing the dependency of this technique on anatomical scans.

David Romascano¹, Alessandro Dal Palú², Jean-Philippe Thiran^1,3, and Alessandro Daducci^1,4
1Signal Processing Laboratory (LTS5), École Polytechnique Fédérale de Lausanne, Lausanne, Vaud, Switzerland, ²Department of Mathematics and Computer Science, University of Parma, Parma, Italy, ³Department of Radiology, University Hospital Center and University of Lausanne, Lausanne, Vaud, Switzerland,⁴Center for Biomedical Imaging, Signal Processing Core, Lausanne, Vaud, Switzerland

In diffusion MRI, traditional tractography algorithms do not recover truly quantitative tractograms and the structural connectivity has to be estimated indirectly by counting the number of fiber tracts or averaging scalar maps along them. Recently, global and efficient methods have emerged to estimate more quantitative tractograms by combining tractography with local models for the diffusion signal, like the Convex Optimization Modeling for Microstructure Informed Tractography (COMMIT) framework. In this abstract, we show the importance of using both (i) proper multi-compartment diffusion models and (ii) adequate multi-shell acquisitions, in order to evaluate the accuracy and the biological plausibility of the tractograms.

Michiel Cottaar¹, Saad Jbabdi¹, Matthew F Glasser², Krikor Dikranian², David C van Essen², Timothy E Behrens¹, and Stamatios N Sotiropoulos^1
1FMRIB Centre, University of Oxford, Oxford, United Kingdom, ²Washington University School of Medicine, Saint Louis, Missouri, United States

Close to the cortical surface axons often bend towards the surface on a spatial scale unresolved in diffusion MRI, which causes these bends to be missed by classical tractography algorithms. Here we present a simple model, which with a single free parameter can reproduce the fibre orientation measured in a high-resolution myelin-stained macaque gyral slice within 4 degrees. This low-parameter model can be fitted to the dispersion orbital distribution function measured in diffusion MRI to reproduce the sub-voxel fibre bending to the cortical surface and hence the cortical connectivity.

Robert Elton Smith¹, J-Donald Tournier^2,3, Fernando Calamante^1,4, and Alan Connelly^1,4
1Imaging division, The Florey Institute of Neuroscience and Mental Health, Heidelberg, Victoria, Australia, ²Centre for the Developing Brain, King's College London, London, United Kingdom, ³Department of Biomedical Engineering, King's College London, London, United Kingdom, ⁴Department of Medicine, The University of Melbourne, Heidelberg, Victoria, Australia

When performing whole-brain fibre-tracking using streamlines tractography, streamline seeding is typically done in an uninformed, homogeneous fashion. Here we demonstrate a novel feedback mechanism, that uses the comparison between the estimated fibre densities (provided by the diffusion model) and reconstructed streamlines density to influence the placement of subsequent streamline seeds. This ensures that a greater number of streamlines are seeded in regions of the image that are otherwise poorly reconstructed, and improves the correspondence between the streamlines reconstruction and the underlying image data.

Peter F. Neher¹, Michael Götz¹, Tobias Norajitra¹, Christian Weber¹, and Klaus H. Maier-Hein^1
1Medical Image Computing Group, German Cancer Research Center (DKFZ), Heidelberg, Germany

Current tractography pipelines incorporate several modelling assumptions about the nature of the diffusion-weighted signal. We present a purely data-driven and thus fundamentally new approach that tracks fiber pathways by directly processing raw signal intensities. The presented method is based on a random forest classification and voting process that guides each step of the streamline progression. We evaluated our approach quantitatively and qualitatively using phantom and in vivo data. The presented machine learning based approach to fiber tractography is the first of its kind and our experiments showed promising performance compared to 12 established state of the art tractography pipelines.