Rachel W Chan¹, Constantin von Deuster^2,3, Christian T Stoeck², Jack Harmer³, Sebastian Kozerke^2,3, and David Atkinson^1
1University College London, London, London, United Kingdom, ²University and ETH Zurich, Zurich, Zurich, Switzerland, ³King's College London, London, London, United Kingdom

Fractional anisotropy (FA) obtained by diffusion tensor imaging (DTI) can be used to non-invasively assess kidney function. A multi-slice targeted FOV technique that uses a non-coplanar application of excitation and refocusing pulse was extended to free-breathing kidney DTI with a navigator-triggered technique. The purpose of this study was to obtain high-resolution FA and diffusion-tensor maps in healthy subjects who were scanned with two (dual- and single-kidney) targeted FOV scans. The FA maps showed consistency between the two targeted FOVs and across subjects. The medulla and cortical FA values, quantified over multiple contiguous slices, were well-differentiated.

Xinwei Shi^1,2, Xiaodong Ma², Wenchuan Wu², Feng Huang³, Chun Yuan^2,4, and Hua Guo^2
1Department of Electrical Engineering, Stanford University, Stanford, CA, United States, ²Center for Biomedical Imaging Research, Tsinghua University, Beijing, Beijing, China,³Philips Healthcare, FL, United States, ⁴Department of Radiology, University of Washington, Seattle, WA, United States

Increasing acquisition efficiency is always a challenge in high-resolution diffusion tensor imaging (DTI), which has low SNR and is sensitive to image artifacts. In this work, a parallel imaging and compressed sensing combined reconstruction framework is proposed, which features multi-shot motion error correction, parallel imaging kernel calibration and anisotropic sparsity model utilizing inter-image correlation tailored for high-resolution DTI. The proposed method, titled as AS-SPIRiT, is implemented based on multi-shot variable density spiral, and evaluated in in-vivo brain DTI experiment. Compared with traditional parallel imaging methods and other sparsity models, AS-SPIRiT provides better preserved image quality and more accurate DTI parameters.

Andriy Fedorov¹, Kemal Tuncali¹, Lawrence P. Panych¹, Janice Fairhurst¹, Clare Tempany¹, and Stephan E. Maier^1
1Department of Radiology, Brigham and Women's Hospital, Boston, MA, United States

We compare image quality and distortion in the prostate diffusion-weighted MRI using single-shot EPI (SS-DWI) and multi-shot readout segmented EPI (MS-DWI) in clinical patients (N=12) undergoing in-bore MRI-guided prostate biopsy. The use of MS-DWI led to marked reduction of distortion under conditions that resulted in non-diagnostic SS-DWI images, with moderate increase in scan time (2 vs 3:30 min). Image quality and distortion severity were strongly correlated with the presence of air for SS-DWI, but not for MS-DWI. Distortion reduction agreed with the theoretically expected values. MS-DWI is thus preferable for improved accuracy of prostate target localization and interventional applications.

Jiazheng Wang¹ and Yongchuan Lai^2
1STO-MR, GE Healthcare, Beijing, Beijing, China, ²STO-MR, GE Healthcare, Beijing, China

Gradient heating and TE are two limiting and mutually restrictive parameters in DWI scan. This work presents a gradient optimization method to reduce gradient heating without sacrificing TE, and quantitatively demonstrates that the method can significantly increase the spatial coverage in slice direction with diffusion imaging. In addition, since this method utilizes diffusion gradients with very short high-amplitude plateau, it theoretically has less sensitivity to motion effects. The proposed method also shows some benefits in eddy currents, leading to sharper subtle structures due to reduced distortion.

Ken Sakaie^1
1Imaging Institute, The Cleveland Clinic, Cleveland, OH, United States

The magnetic resonance fingerprinting (MRF) framework has a number of attractive features for quantitative imaging such as speed, precision and robustness against artifact. We examine, by simulation, the potential of MRF for diffusion MRI.

Jian Cheng¹, Pew-Thian Yap¹, and Dinggang Shen^1
1University of North Carolina at Chapel Hill, chapel Hill, NC, United States

We propose a integer linear programming framework to design single/multi-shell sampling schemes in diffusion MRI.

Eric Gibbons¹, John Pauly², and Adam Kerr^2
1Department of Bioengineering, Stanford University, Stanford, California, United States, ²Department of Electrical Engineering, Stanford University, Stanford, California, United States

We present a method to incorporate isotropic diffusion weighting into a twice-refocused spin-echo (TRSE) preparation with a single image acquisition. The gradient design is optimized to compensate for eddy current effects of a given time constant and to minimize the TE. By doing this, images with isotropic diffusion contrast are obtained in three times faster than the standard acquisition procedures while also preserving edge sharpness and geometric integrity.

Eric Peterson¹, Samantha Holdsworth¹, Rafael O'Halloran¹, Eric Aboussouan¹, and Roland Bammer^1
1Radiology, Stanford University, Stanford, CA, United States

The advent of slice accelerated imaging has allowed the significant speedup of diffusion tensor imaging. Slice accelerated imaging excels at accelerating imaging which has a relatively long repetition time, such as readout segmented imaging, which is inherently a dual-echo imaging technique and typically has a repetition time of over 100ms. This work demonstrates that the image quality of diffusion tensor readout segmented imaging is comparable to non-slice accelerated imaging, despite the acquisition being at least 2 times faster.

Thomas L. Chenevert¹, Dariya I. Malyarenko¹, Michael A. Boss², and Scott D. Swanson^1
1Radiology, University of Michigan, Ann Arbor, Michigan, United States, ²Electromagnetics Division, National Institute of Standards and Technology (NIST), Boulder, Colorado, United States

Phantoms to assess diffusion measurement accuracy require knowledge of media temperature. Proton resonant frequency shift derived by phase difference, MRS, or echo interference pattern have been used to infer temperature. Thulium-based MR thermometry offers greater sensitivity, but requires wide frequency range. In this work, an EPI sequence is used to estimate the temperature-dependent spatial distance between two chemical shift targets. Sensitivity is improved by low bandwidth EPI and long-T2 chemical moieties. Spatial separation of two targets for temperature calibration is automatically determined using cross correlation. The proposed methodology is evaluated for absolute in-situ MR thermometry in a multi-component diffusion phantom.

Mathias Engström^1,2, Magnus Mårtensson^1,3, Ola Norbeck², Enrico Avventi¹, and Stefan Skare^1,2
1Clinical Neuroscience, Karolinska Institutet, Stockholm, Stockholm, Sweden, ²Neuroradiology, Karolinska University Hospital, Stockholm, Stockholm, Sweden, ³EMEA Research and Collaboration, GE Applied Science Laboratory, GE Healthcare, Stockholm, Sweden

A Diffusion-Weighted 3D Multi-Slab EPI protocol is presented as an alternative to 2D Diffusion-Weighted single-shot EPI for clinical imaging of the brain, when high-resolution and high-SNR is required.

Xiaodong Ma¹, Feng Huang², Xinwei Shi¹, Wenchuan Wu¹, Binbin Sui³, and Hua Guo^1
1Center for Biomedical Imaging Research, Department of Biomedical Engineering, School of Medicine, Tsinghua University, Beijing, Beijing, China, ²Philips Research China, Beijing, China, ³Department of Neuroradiology, Beijing Neurosurgical Institute, Beijing Tiantan Hospital, Capital Medical University, Beijing, China

Multi-shot DWI requires long scan time especially in diffusion tensor imaging (DTI). Parallel imaging reconstruction (CG-SENSE) suffers from low SNR when relatively large reduction factor is used. Existence of isotropic diffusion provides common information among different directions, which is utilized in this study through an interleaved acquisition scheme to enhance image reconstruction. Illustrated as an example, Common inhanced SPIRiT (CI-SPIRiT) reconstruction method in VDS DTI is proposed. Preliminary results show that it improves data sampling efficiency and obtains images of higher SNR.

Julia Reinhardt¹, Tilman Schubert¹, Michael Amann^1,2, and Christoph Stippich^1
1Division of Diagnostic and Interventional Neuroradiology, Department of Radiology, University Hospital Basel, Basel, Basel, Switzerland, ²Department of Neurology, University Hospital Basel, Basel, Switzerland

DWI is increasingly applied in the whole field of diagnostic imaging. Here we evaluated applicability of b-values from 1000-10000s/mm2 in a collective of healthy volunteers regarding SNR and overall image quality using a 3T-Scanner with a gradient strength of 80mT/m. The SNR values decrease exponentially. The b-values of grey matter structures drop more pronounced with increasing b-values as compared to white matter or mixed brain tissue, which may be exploited as additional diagnostic information. In this work we were able to demonstrate that DWI with b-values up to 10000s/mm2 with still reasonable SNR can be achieved with high gradient-strength MR-scanners.

Michael A. Jacobs¹, Li Pan², Doris G. Leung³, and John A. Carrino^4
1Russell H. Morgan Department of Radiology, The Johns Hopkins University School of Medicine, Baltimore, Maryland, United States, ²Center for Applied Medical Imaging, Siemens Corporation Corporate Technology, Baltimore, MD, United States, ³Center for Genetic Muscle Disorders, Kennedy Krieger Institute, Baltimore, MD, United States,⁴Russell H. Morgan Department of Radiology, The Johns Hopkins University School of Medicine, Baltimore, MD, United States

The use of multiparametric whole body imaging is rapidly gaining favor in conducting non-ionizing interrogation of the whole body in a short period of time. The use of functional parameters are needed to better define tissue characteristics, in particular, with the use of DWI. We have tested and demonstrated improved DWI with ADC mapping using advanced pulse sequences incorporating mono- and bipolar gradients schemes during continuous table movement WB-MRI in a group of subjects.

Yuichi Suzuki¹, Yoshitaka Masutani¹, Kuniaki Saito², Akira Kunimatsu¹, Akitake Mukasa², Masaki Katsura¹, Katsuya Maruyama³, Thorsten Feiweier⁴, Kenji Ino¹, Yasushi Watanabe¹, Takeo Sarashina¹, Masami Goto¹, Jiro Sato¹, Keiichi Yano¹, Nobuhito Saito², and Kuni Ohtomo^1
1Department of Radiology, The University of Tokyo Hospital, Bunkyo-ku, Tokyo, Japan, ²Department of Neurosurgery, The University of Tokyo Hospital, Tokyo, Japan,³Siemens Japan, Tokyo, Japan, ⁴Health care Neurology Applications Development, Siemens AG, Bavaria, Germany

Our goal is to compare the depiction abilities among our anisotropic MPG set and Jones (isotropic) MPG sets for the brain disease patients.

Jesper Andersson¹ and Stamatios N Sotiropoulos^1
1FMRIB-Centre, Oxford, Oxfordshire, United Kingdom

A very sensitive method for detecting and correcting signal dropout in diffusion images has been developed. It is incorporated into a framework for simultaneous correction of all off-resonance distortions and subject movement. When an outlier is detected it is replaced by its predicted value based on a Gaussian Process.

Mustafa Okan Irfanoglu^1,2, Amritha Nayak^1,2, Andrew Knutsen³, Pooja Modi¹, Joelle Sarlls³, and Carlo Pierpaoli^1
1NICHD, NIH, Bethesda, MD, United States, ²CNRM, USUHS, Bethesda, MD, United States, ³NIH, Bethesda, MD, United States

In this work, we propose a novel EPI distortion correction scheme for diffusion weighted images with reversed phase encoding gradient schemes. In addition to describing the proposed method and comparing to existing methodologies, we highlight several important factors affecting these “blip-up blip-down” corrections and analyze the effects of these factors. Our analysis indicated that using additional anatomical image information helps the registration under imperfect data conditions and that using diffusion weighted image contrast along with b=0 s/mm2 image significantly improves anatomical accuracy, especially in tightly packed regions such as the brain stem.

Markus Nilsson¹, Filip Szczepankiewicz², Danielle van Westen³, and Oskar Hansson^4
1Lund University Bioimaging Center, Lund University, Lund, Sweden, ²Dept. of Medical Radiation Physics, Lund University, Sweden, ³Dept. of Diagnostic Radiology, Lund University, Sweden, ⁴Clinical Memory Research Unit, Dept. of Clinical Sciences Malmö, Lund University, Sweden

Motion correction of high b-value diffusion MRI data can be performed using the non-diffusion weighted image as reference, but here we show that this conventional motion correction leads to substantial registration errors in terms of misalignment between low and high b-value volumes. We suggest a method that can eliminate this registration error by extrapolating reference volumes from low b-value data to high b-value data. The improved motion correction leads to a reduction in the standard deviation in the FA of up to 40%, which is associated to a large increase in the statistical power of group studies.

Jan Hering^1,2, Peter F. Neher¹, Hans-Peter Meinzer¹, and Klaus H. Maier-Hein^1,3
1Div. Medical and Biological Informatics, German Cancer Research Center, Heidelberg, Germany, ²Mannheim University of Applied Sciences, Mannheim, Germany, ³Quantitative Image-based Disease Characterization, German Cancer Research Center, Heidelberg, Germany

Retrospective head motion correction counts to the most important post-processing steps in diffusion-weighted MR. However, due to the missing ground-truth, a quantitative evaluation of correction algorithms is possible only to a limited extent. Usage of reference transforms applied to an already aligned data is restricted to translation, as the diffusion signal is not invariant under rotation. We present a novel method for construction of ground-truth head motion data using the Fiberfox tool for realistic diffusion-weighted signal simulation. We further show that the rotation must not be neglected in motion correction and evaluation.

Benjamin R Morgan¹, Wayne Lee¹, and Margot J Taylor^1,2
1Hospital for Sick Children, Toronto, Ontario, Canada, ²University of Toronto, Toronto, Ontario, Canada

Diffusion imaging acquisitions are sensitive to head motion, making the collection of a large number of directions problematic, especially in motion-prone populations, such as children. In this study, we propose splitting long acquisitions into separate, shorter acquisitions and including interleaved b=0 reference volumes to facilitate improved motion and b-vector correction. A novel procedure for preprocessing motion-corrupted data collected in this manner is introduced and compared to the standard eddy correction algorithm. Preliminary evidence, using a series of motion-free and motion-corrupted datasets acquired on one healthy adult, shows an advantage to the proposed technique in fractional anisotropy and principle eigenvector estimation.

Gwendolyn Van Steenkiste¹, Ben Jeurissen¹, Paul Parizel², Dirk H.J. Poot^3,4, and Jan Sijbers^1
1iMinds Vision Lab, University of Antwerp, Wilrijk, Antwerp, Belgium, ²department of Radiology, University of Antwerp, Antwerp, Belgium, ³Imaging Science and Technology, Delft University of Technology, Delft, Netherlands, ⁴BIGR (dept. of Medical informatics and Radiology), Erasmus Medical Center Rotterdam, Rotterdam, Netherlands

Diffusion weighted (DW) images are acquired with a low spatial resolution to obtain a reasonable signal-to-noise ratio within a clinically feasible scan time. Recently, a method has been proposed that improves this trade-off by acquiring multiple anisotropic DW images with different slice orientations, and recovering the underlying high resolution (HR) DW images via super-resolution reconstruction (SRR). Here, we present an improved method (SRR-DTI) which includes the diffusion tensor model. We show using whole brain tractography that fiber tracking in a SRR-DTI data set is more accurate than in a HR DW data set acquired within the same scan time.

Jelle Veraart¹ and Jan Sijbers^1
1Vision Lab, University of Antwerp, Antwerp, Belgium

For clinically relevant SNR values (SNR>2), the weighted linear least squares estimator is theoretically expected to be as accurate as advanced estimators that incorporate prior knowledge of the data distribution in the estimation of DTI/DKI model parameters. However, one must bear in mind that the high accuracy vanishes if magnitude operations are applied prior to model fitting. After magnitude operations, which are generally included in the diffusion MRI processing pipeline, the prior knowledge of the noise parameter becomes a must in order to define an unbiased estimator.

Matthew Liptrot¹ and François Lauze^1
1Department of Computer Science, University of Copenhagen, Copenhagen, Copenhagen, Denmark

We present a simple, novel approach to the voxelwise classification of brain tissue acquired with diffusion-weighted imaging (DWI). By working directly upon the individual DWI volume data, it makes no assumption of an underlying diffusion model. In addition, by summarising statistics across the diffusion gradient directions, we obtain features that are rotationally invariant. We show an example of how well a resulting cluster spatially matches a high FA region, thereby corresponding to probable single-tract voxels. The method could have application during tractography pre-processing, and has potential as a complementary approach for analysis of DWI datasets.

Feng Yang¹, Yue-Min Zhu², Maria A. Zuluaga³, and Pierre Croisille^2
1School of Computer and Information Technology, Beijing JiaoTong University, Beijing, China, ²CREATIS, CNRS UMR 5220; Inserm U 1044; INSA of Lyon; University of Lyon, Lyon, France, ³Centre for Medical Image Computing, University College London, London, United Kingdom

We use diffusion tensor magnetic resonance imaging to investigate the effects on fractional anisotropy of diffusion tensor interpolation methods in the presence of myocardial infarction. Our results suggest that precaution should be taken with diffusion tensor interpolation when using absolute indexes (as the FA value) to distinguish infarction and remote regions, since interpolation methods may affect FA values in a non-negligible manner.

Mie Kee Lam¹, Clemens Bos¹, Chrit T W Moonen¹, Max A Viergever¹, and Lambertus W Bartels^1
1Image Sciences Institute, UMC Utrecht, Utrecht, Utrecht, Netherlands

IVIM MRI allows extraction of perfusion and diffusion information from diffusion weighted MR data and is receiving interest for use in well-perfused organs. Monte Carlo simulations were used to investigate the precision and accuracy of IVIM parameter estimation. To achieve acceptable precisions, unrealistically high SNRs are required for perfusion fractions typical of brain. At perfusion fractions typical of well-perfused organs, acceptable precisions and accuracies can be achieved with realistic SNRs. However, at high perfusion fractions systematic errors may become more prominent.

Daniel Güllmar¹, Ferdinand Schweser¹, and Jürgen R Reichenbach^1
1Medical Physics Group, Institute of Diagnostic and Interventional Radiology I, Jena University Hospital - Friedrich Schiller University Jena, Jena, Germany

Artificial increase in fractional anisotropy occurs due to truncation artifacts in b0 (unweighted) images in DTI data. This artifact can be reduced using total variation constrained data extrapolation. Uncorrected and corrected data were used to obtain FA maps and the differences were investigated in a detailed ROI based analysis. This analysis showed a clear dependence of the artifact with respect to pial surface distance as expected from the theory.

Gabriel Varela¹, Alexandra Tobisch^2,3, Tony Stoecker², and Pablo Irarrazaval^1,4
1Biomedical Imaging Center - Pontificia Universidad Catolica de Chile, Santiago, Metropolitan District, Chile, ²German Center of Neurological Diseases, North Rhine-Westphalia, Germany, ³University of Bonn, North Rhine-Westphalia, Germany, ⁴Department of Electrical Engineering, Pontificia Universidad Catolica de Chile, Metropolitan District, Chile

Compressed Sensing allows accelerating Diffusion Spectrum Imaging (DSI) acquisitions by reconstructing the Ensemble Average Propagator from a significantly reduced number of q-space samples. Nevertheless, the reconstruction performance is highly dependent on the sparse domain, which has not been fully studied for the specific DSI application. In this work we propose a new sparse domain based on Curvelets, a multi-resolution geometric analysis that incorporates explicitly an angular decomposition with parabolic scaling and location to characterize bounded curve-singularities in a sparse matter. We show that this domain allows even higher accelerating factors for DSI and thus significantly shortening the scan time.

Susan Doshi¹, Derek Jones², and Daniel Barazany^2,3
1Computer Science and Informatics, Cardiff University, Cardiff, Glamorgan, United Kingdom, ²CUBRIC, Cardiff University, Cardiff, United Kingdom, ³Department of Neurobiology, Tel Aviv University, Tel Aviv, Israel

We use Bayesian statistical modelling to regularise parameter estimates in advanced diffusion imaging. By incorporating prior knowledge (such as spatial smoothness) during estimation, we exploit the information more fully than applying smoothing as post-processing. We use a Markov random field for the prior probability. This approach allows the possibility of non-isotropic smoothing, and for edges in one part of the data to guide the fitting of other parts. We demonstrate the approach with CHARMED data, using ex-vivo porcine spinal cord as a biological phantom. The parameter estimates in homogeneous areas are smooth (agreeing with our prior belief), with edges preserved.

Neil P Jerome¹, Matthew R Orton¹, James d'Arcy¹, Thorsten Feiweier², Dow-Mu Koh³, David J Collins¹, and Martin O Leach^1
1CR-UK and EPSRC Cancer Imaging Centre, The Institute of Cancer Research, Sutton, Surrey, United Kingdom, ²Imaging & Therapy Division, Siemens AG, Erlangen, Germany,³Department of Radiology, Royal Marsden Hospital, Sutton, Surrey, United Kingdom

Respiratory motion commonly confounds abdominal DWI, and motion minimisation strategies adversely affect scan efficiency and comfort. Blurring is due to post-acquisition combination of images from separate signal averages and diffusion-gradient directions, and is not inherent to the images. In a volunteer cohort where all images were stored separately, taking a (voxel-by-voxel) median image instead of a mean at each b-value yields parameter maps with much improved sharpness while still retaining tissue features. ADCs from ROIs in liver and kidneys were 108±18 vs 120±26 (p=0.007) and 182±17 vs 188±13 (p=0.04) x10^-5 mm²s^-1 for median and mean, respectively.

Suguru Yokosawa¹, Hisaaki Ochi¹, Yoshitaka Bito², Kenji Ito³, and Makoto Sasaki^3
1Central Research Laboratory,Hitachi, Ltd., Kokubunji-shi, Tokyo, Japan, ²MRI System Division, Hitachi Medical Corporation, Chiba, Japan, ³Division of Ultrahigh Field MRI, Institute for Biomedical Science, Iwate Medical University, Iwate, Japan

In this work, we proposed a robust estimation method for DKI that reduces noise while suppressing image blurring. In our method, instead of being applied to all image data, a spatial smoothing filter is applied only to voxel dataset that contain outliers. These outliers are detected by deviation of calculated DKI coefficient from constrained condition. We demonstrated that the proposed method obtains a MK map with less blurring than constrained fitting with spatial smoothing and reduces pepper noises. The results confirm the effectiveness of our method applied in clinical DKI.

Martijn Froeling¹, Chantal M.W. Tax², Sjoerd B. Vos², Peter R. Luijten¹, and Alexander Leemans^2
1Department of Radiology, University Medical Center Utrecht, Utrecht, Netherlands, ²Image Sciences Institute, University Medical Center Utrecht, Utrecht, Netherlands

For this studie we have acquired multiple anatomical scans and a diffusion data set containing 8000 diffusion weighted volumes, consisting of five shells and two Cartesian grids. Our dataset, coined Multiple Acquisitions for Standardization of Structural Imaging Validation and Evaluation (MASSIVE), allows for comparison of various techniques for e.g., registration, segmentation, tractography, clustering, denoising/ regularization and bootstrapping, eliminating the bias of variable data acquisition and sub-optimal sampling schemes.

Chitresh Bhushan¹, Justin P. Haldar¹, Anand A. Joshi¹, David W. Shattuck², and Richard M. Leahy^1
1Ming Hsieh Department of Electrical Engineering, University of Southern California, Los Angeles, California, United States, ²Department of Neurology, University of California, Los Angeles, California, United States

Accurate registration between MPRAGE and diffusion MRI images is essential for many multi-modal neuroimaging studies. We describe a new method, INVERSION (Inverse contrast Normalization for VERy Simple registratION), that robustly aligns MPRAGE and b=0 s/mm² images by leveraging known "inverted" contrast relationships between these two modalities. We transform the contrast of the b=0s/mm² image to match the contrast of the MPRAGE image, and achieve consistently accurate registration using the simple sum of squared differences cost function. Unlike most multi-modal registration approaches, INVERSION uses a locally smooth, and frequently convex, cost function that is relatively easy to numerically optimize.

Asif K Suri^1,2, Roman Fleysher¹, and Michael L Lipton^1,3
1Albert Einstein College of Medicine, Bronx, NY, United States, ²Bronx Psychiatric Center, Bronx, NY, United States, ³Montefiore Medical Center, Bronx, United States

Brain registration of a subject and set of controls to a template is an essential step in determining diffusion abnormalities obtained from MR diffusion imaging. Limitations in the registration process compounded with use of a canonical brain atlas, results in misregistrations that can result in erroneous identification of abnormal diffusion values within subjects. This study attempts to reduce misregistrations using subject based template, which increases the accuracy of registration. This strategy demonstrates a substantial decrease in volume of abnormally low FA attributed to the improved accuracy of registration and can have useful clinical applications.

Jhimli Mitra¹, Jurgen Fripp¹, Kaikai Shen¹, Kerstin Pannek¹, Pierrick Bourgeat¹, Olivier Salvado¹, Bruce Campbell², Susan Palmer³, Leeanne Carey³, and Stephen Rose^1
1The Australian e-Health Research Centre, CSIRO Computational Informatics, CSIRO Preventative-Health Flagship, Herston, QLD, Australia, ²Department of Medicine and Neurology, Melbourne Brain Centre at the Royal Melbourne Hospital, Parkville, VIC, Australia, ³The Florey Institute of Neuroscience and Mental Health, Parkville, VIC, Australia

Our hypothesis is that loss-in-connectivity in brain regions post-stroke is correlated with post-stroke depression (PSD). We propose an automated detection of cortical/sub-cortical regions that are associated with PSD. The method involves pairwise comparison of network connectivity matrices between normal and stroke patients using diffusion tractography and network based statistics to identify the networks affected by ischemic stroke. Then a groupwise linear regression analysis is performed between the loss-in-connectivity in each brain region and the respective patients' depression scores at 3 month post-stroke stage. The results revealed positive correlations between loss-in-connectivity and PSD in some brain regions including the thalamus.

Mustafa Okan Irfanoglu^1,2, Amritha Nayak^1,2, Lindsay Walker³, Carlo Pierpaoli¹, and The Brain Development Cooperative Group^4
1NICHD, NIH, Bethesda, MD, United States, ²CNRM, USUHS, Bethesda, MD, United States, ³Brown University, RI, United States, ⁴www.NIH-PediatricMRI.org, MD, United States

In analyzing quantitative MRI metrics in large population neuroimaging studies, it is convenient to define ROIs on a representative atlas and then transpose the ROIs onto individual subject space and extract values. Here we assess the contribution to overall variance that different ROI transformation methods introduce on the analysis of developmental trajectories of the diffusion MRI pediatric neuroimaging project, a large database of more than 400 subjects age 0-20. We find that the contribution to overall variance varies significantly with various methods and we propose a novel strategy that is more robust than previous approaches.

Alexandru V Avram¹, Alan S Barnett^1,2, and Peter J Basser^1
1Section on Tissue Biophysics and Biomimetics, NICHD, National Institutes of Health, Bethesda, MD, United States, ²The Henry Jackson Foundation, Bethesda, MD, United States

Scalar indices derived from diffusion weighted MR images (DWIs) provide invaluable clinical information for characterizing changes in brain microstructure and cytoarchitecture. In this study we apply the recently developed mean apparent propagator (MAP)-MRI framework to quantify the brain diffusion properties in healthy volunteers and describe and interpret the variation of scalar descriptors of the MAP (such as zero-displacement probability, non-gaussianity and propagator anisotropy) along the major white matter fiber pathways and discuss the clinical significance of our findings.

Paul A Taylor^1,2 and Ziad S Saad^3
1Faculty of Health Sciences, University of Cape Town, Cape Town, Western Cape, South Africa, ²African Institute for Mathematical Sciences, South Africa, ³NIMH, National Institutes of Health, Bethesda, MD, United States

We present developments in the AFNI-FATCAT suit of tools for analyzing MRI functional and structural connectivity. Improvements include: enhanced deterministic tracking to utilize voxelwise uncertainty; increased options for combatting false positives and negatives with including multi-directional tracking and anti-masking ROIs; combined visualization with SUMA and AFNI, allowing interactive manipulation of tracking and regions. These additions to AFNI-FATCAT increase researchers' capabilities for integrating functional and diffusion-based tractographic connectivity.

Yu-Jen Chen¹, Yun-Chin Hsu¹, Yu-Chun Lo¹, and Wen-Yih Isaac Tseng^1
1Center for Optoelectronic Medicine, National Taiwan University College of Medicine, Taipei, Taipei, Taiwan

In this study, we proposed a whole brain tract-specific analysis using a high quality DSI template, predetermined tracts in the template, and LDDMM coregistration method. Compared to manual approach, TBAA showed higher overlap of tract positions and smaller variability of the sampled GFA. In conclusion, TBAA overcomes the problem of variability in manual tractography on individual DSI, and is potentially useful in high-throughput tract-specific analysis of the whole brain.

Caroline Presseau¹, Pierre-Marc Jodoin¹, and Maxime Descoteaux^1
1University of Sherbrooke, Sherbrooke, Quebec, Canada

Fiber tracking dataset are heavy to compute, but also difficult to visualize and hard to store on disk (especially when dealing with a collection of brains). These problems call for a fiber-specific compression. As of today, no fiber compression format has yet been adopted and the need for it is now becoming an issue for future connectomics research. In this work, we propose a new compression format for tractography datasets reconstructed from diffusion magnetic resonance imaging (dMRI)

Bryce Wilkins¹, Namgyun Lee^1,2, Meng Law^1,3, and Natasha Leporé^1,4
1Biomedical Engineering, University of Southern California, Los Angeles, California, United States, ²Center of Magnetic Resonance Research, Korea Basic Science Institute, Ochang, Korea, ³Radiology, Keck School of Medicine of USC, Los Angeles, California, United States, ⁴Radiology, Children's Hospital of Los Angeles, California, United States

This abstract describes the development of simulated DW-MRI brain images and quantitative tools for evaluating the performance of diffusion analysis methods in terms of fiber orientation estimation and false-positive/-negative fiber rates, which are of fundamental importance to tractography based studies. Synthetic data is generated at SNRs of 9, 18 and 36, using sets of 20, 30, 40, 60, 90 and 120 gradient directions, at diffusion-weighting of b=1000 common to clinical acquisitions. We illustrate correspondence between in-vivo and synthetic data analysis, and provide an example of quantitative results obtained from six well-known DW-MRI analysis methods.

Ariel Rokem¹, Kimberly L Chan¹, Jason D Yeatman¹, Franco Pestilli¹, Aviv Mezer², and Brian A Wandell^2
1Stanford University, Stanford, CA, United States, ²Stanford University, Stanford, California, United States

Models of diffusion MRI (DWI) are used for inferences about the properties of the tissue and fiber orientations. Though stability of DWI model parameters is often evaluated, there are no extensive studies of model prediction accuracy. We evaluated different models using cross-validation in a test-retest data set and data from the Human Connectome Project. In most of the white matter and multiple b-values, we find that the classic diffusion tensor model predicts the data more accurately than test-retest reliability. However, modeling the signal as a combination of contributions from distinct white matter fascicles provides more accurate model predictions.

Bagrat Amirbekian^1,2 and Roland G. Henry^1,2
1Graduate Program in Bioengineering, UC San Francisco and UC Berkeley, San Francisco, CA, United States, ²Neurology, UC San Francisco, San Francisco, CA, United States

Spherical harmonic (SH) functions have been widely used to fit HARDI data, but residual bootstrap methods have mostly focused on other models for fitting this data. Here we use the example of estimating SNR to show that bootstrap estimates, using the SH functions, are accurate for many choices of maximum SH order. Additionally we show that for good precision of those estimates, the maximum SH order must be high enough to fit the data well and low enough to preserve sufficient degrees of freedom.

Xiaodong Ma¹, Feng Huang², and Hua Guo^1
1Center for Biomedical Imaging Research, Department of Biomedical Engineering, School of Medicine, Tsinghua University, Beijing, Beijing, China, ²Philips Research China, Beijing, China

In DWI, common information from other diffusion direction can be shared to increase the image quality of one diffusion direction. However, inappropriate selection of complementary information may provide suboptimal image results. In this study, we systematically compared the performance of common information enhanced reconstruction when different direction numbers are used for the image calculation. Results show that care should be taken when including other directions for one direction reconstruction in multiple direction DWI or high angular diffusion imaging.

Gregory Lemberskiy^1,2, Dmitry S Novikov¹, and Els Fieremans^1
1New York University School of Medicine, Bernard and Irene Schwartz Center for Biomedical Imaging, Department of Radiology, New York, NY, United States, ²Sackler Institute of Graduate Biomedical Sciences, New York University School of Medicine, New York, NY, United States

DWI images in the body are often afflicted with systematic vibration artifacts (SVA), characterized by localized regions of signal dropout. Generalized image processing techniques, such as outlier detection and histogram-based methods, require either large datasets or cannot account for nonlinear phenomena such as RF inhomogeneity. To overcome these limitations, we propose a rejection scheme based on unphysical diffusion coefficient values of the directional ADC, called Artifact Correction based on Diffusion Coefficient (ACDC). ACDC does not require many directions or averages to be efficient. Hence, its implementation in clinical systems would be straightforward.

Zhen Wang¹, Yuan Wang¹, Jianye Zhang¹, Xu Yan², Hui Liu², and Zeping Xiao^1
1Shanghai Mental Health Center, Shanghai, China, ²MR Collaboration NE Asia, Siemens Healthcare, Shanghai, China

The study tested the estimation accuracy of Neurite orientation dispersion and density imaging (NODDI) using a common 2-shell protocol with the same gradient direction at each shell. The experiment used visual and quantitative comparisons to demonstrate that the common 2-shell protocol had very similar performance as the optimized NODDI protocol, but is easy to configure in a commercial scanner, thus can be widely adopted in clinical study.

Kentaro Akazawa¹, Koji Sakai², Rennie, Yung-Chieh Chen³, Jun Tazoe¹, Kaori Nishida¹, Akiko Takahata¹, Mariko Goto¹, Koshi Terayama¹, and Kei Yamada^1
1Department of Radiology, Graduate School of Medical Science, Kyoto Prefectural University of Medicine, Kyoto, Kyoto, Japan, ²Department of Human Health Science, Graduate School of Medicine, Kyoto, Kyoto, Japan, ³Department of Biomedical Imaging and Radiological Sciences, National Yang-Ming University, Taipei, Taipei, Taiwan

Diffusion-weighted image (DWI)-thermometry is one of the most clinically applicable approaches among some magnetic resonance methods. The DWI-based method calculates the temperature of cerebrospinal fluid based on diffusion coefficient of the lateral ventricles. However, the technique is relatively recent and insufficient yet in its analysis. We, therefore, need further evaluation. The results revealed that lateral ventricular (LV) temperature measurement was successfully performed using DWI-thermometry, slice thickness had significant effect on the calculation of LV temperature, and thinner slice thickness was considered to produce better results than the thick slices. Also we found that LV temperature would change with tympanic temperature.

Rong Wang¹, Hao Fu¹, Hui Zhang¹, Chenxia Li¹, and Jian Yang^1
1Department of Radiology, the First Affiliated Hospital of Xi'an Jiaotong University, Xi'an, Shaanxi, China

The prognoses of non-invasive therapies such as magnetic resonance-guided focused ultrasound surgery (MRgFUS) and GnRH analogue of uterine leiomyomas were proved have a relationship with the signal intensity of tumor on T2WI. However the reason is not definite yet. We performed Intravoxel incoherent motion (IVIM) MR study on a cohort of patients to compare the diffusion and perfusion characteristics of uterine fibroid with different T2WI type. The results indicated that the cellularity is tighter in Type 1 and looser in Type 3 among 3 fibroid types, whereas perfusion parameters of Intravoxel incoherent motion (IVIM), seem no differences among them. Thus, the present results implied that IVIM technique facilitated the understanding of the tissue characteristics in uterine fibroid and proposed to guide patient selection for MRgFUS ablation.

Christopher Lee Welsh¹, Edward W Hsu¹, and Edward VR DiBella^2
1Department of Bioengineering, University of Utah, Salt Lake City, UT, United States, ²Radiology, UCAIR, University of Utah, Salt Lake City, UT, United States

Diffusion kurtosis imaging (DKI) is a way to model tissue microstructure that is more realistic than DTI since it measures the degree of non-Gaussian diffusion. However, DKI requires a long scan time. A model-based strategy is presented to estimate diffusion and kurtosis tensors directly from accelerated k-space data. The accuracy of the model-based method with an acceleration factor of 3 was compared to using all acquired data. The findings suggest the proposed strategy can be used to reduce DKI scan time if used in conjunction with SIR, while still characterizing non-Gaussian diffusion and neural fiber crossings without loss of accuracy.

Eizou UMEZAWA¹, Daichi ISHIHARA¹, and Shota SEKO^1
1School of Health Sciences, Fujita Health University, Toyoake, Aichi, Japan

Recently, clinical usefulness of diffusional kurtosis has been reported, e.g., for distinguishing benign from malignant prostate tissues. Measured diffusional kurtosis must contain a systematic error owing to two factors: approximations in method for kurtosis measurement and perfusion. The error is unclear because the true value is still unknown. Clarifying the error is important to improve the kurtosis measurement method and explore its clinical significance. We first investigate the systematic error in measured kurtosis by numerical experiments. We then develop a new method for measuring kurtosis. Finally, we assess clinical significance of the kurtosis obtained by the new method with simulation.

Fumitaka Kumagai^1,2, Akira Nishikori^1,2, Masaaki Hori², Yoshitaka Masutani³, Masaki Katsura³, Koji Kamagata², Kohei Kamiya², Michimasa Suzuki², Issei Fukunaga^1,2, Nozomi Hamasaki², Syuji Sato², Yuriko Suzuki², Mariko Yoshida², Hajime Arai⁴, Hisato Ishii⁴, Shigeki Aoki², Atsushi Senoo¹, and Akihide Kondo^4
1Department of Radiological Sciences, Graduate School of Human Health Sciences, Tokyo Metropolitan University, Arakawa-ku, Tokyo, Japan, ²Department of Radiology, Juntendo University School of Medicine, Bunkyo-ku, Tokyo, Japan, ³Department of Radiological Medicine, Graduate School Medicine, University School of Medicine, Bunkyo-ku, Tokyo, Japan, ⁴Department of Neurosurgery, Juntendo University School of Medicine, Bunkyo-ku, Tokyo, Japan

Diffusional kurtosis imaging (DKI) has been highlighted as a new technique based on non-Gaussian water diffusion analysis. Although recent studies have reported values of diffusional MR imaging in glioma grading, underlining microstructural changes corresponding to changes of diffusion metrics are not fully elucidated. This study investigated differences in DKI metrics among different components within glioblastoma, by means of voxel-by-voxel analysis. This study give us insight and better understanding of changes in diffusion metrics of different tissue components of glioblastoma.

Brian Hansen¹, Torben E. Lund¹, Ryan Sangill¹, and Sune N. Jespersen^1,2
1CFIN/MindLab, Aarhus University, Aarhus, Denmark, ²Dept. of Physics and Astronomy, Aarhus University, Denmark

Diffusion kurtosis imaging (DKI) is a popular extension of diffusion tensor imaging accounting for non-gaussian aspects of diffusion in biological tissue. Several studies have indicated enhanced sensitivity of mean kurtosis (MK) to pathology, including stroke. Recently, we proposed a fast acquisition and postprocessing scheme based on a linear combination of only 13 diffusion weighted images for estimation of the mean tensor kurtosis. Here we extend this scheme by incorporating exact estimation of the mean diffusivity and show that this produces an improved estimate of mean tensor kurtosis across large brain regions. Our extension also permits acquisition b-values to be optimized numerically and the experimental uncertainty and precision to be mapped.

Yuichi Suzuki¹, Akira Kunimatsu¹, Harushi Mori¹, Junichi Hata¹, Hiroki Sasaki¹, Shiori Amemiya¹, Yoshitaka Masutani¹, Katsuya Maruyama², Jian Xu³, Yasushi Watanabe¹, Kenji Ino¹, Masami Goto¹, Jiro Sato¹, Keiichi Yano¹, and Kuni Ohtomo^1
1Department of Radiology, The University of Tokyo Hospital, Bunkyo-ku, Tokyo, Japan, ²Siemens Japan, Tokyo, Japan, ³Siemens Medical Solutions USA Inc, Pennsylvania, United States

to evaluate the neonatal/infants brain development with DKI, and to compare the quantitative values derived from DKI with the conventional diffusion value (apparent diffusion co-efficiency [ADC] and fractional anisotropy [FA]).

Roger M. Bourne¹ and Matt G. Hall^2
1University of Sydney, Sydney, NSW, Australia, ²University College London, London, United Kingdom

This study explores a development of the conventional biexponential model by adding a stretching factor to the lower diffusivity component. The superior Akaike information content and leave-one-out prediction performance of the stretched over the conventional model at short diffusion time is consistent with the presence of a diversity of diffusion environments experienced by the “slow” component of the models.

Alexandru V Avram¹, Alan S Barnett^1,2, Evren Ozarslan^1,3, Joelle E Sarlls⁴, M. Okan Irfanoglu^1,2, Elizabeth Hutchinson^1,5, Carlo Pierpaoli¹, and Peter J Basser^1
1Section on Tissue Biophysics and Biomimetics, NICHD, National Institutes of Health, Bethesda, MD, United States, ²The Henry Jackson Foundation, Bethesda, MD, United States, ³Department of Radiology, Brigham and Women's Hospital, Harvard Medical School, Boston, MA, United States, ⁴NINDS, National Institutes of Health, Bethesda, MD, United States, ⁵Center for Neuroscience and Regenerative Medicine, USUHS, Bethesda, MD, United States

Orientationally invariant measures such as Fractional Anisotropy or mean diffusivity are invaluable for characterizing changes in cytoarchitecture and microanatomical organization of brain tissue during stroke, cancer, neurodegenerative diseases and aging. In this study we apply the recently developed MAP-MRI framework to measure displacement profiles of water molecules in healthy human volunteers. Moreover, we characterize the mean apparent diffusion propagator with novel orientationally invariant scalar measures of zero-displacement probability, non-gaussianity, and propagator anisotropy and evaluate the feasibility of generating such images from clinical acquisitions.

Yogesh Rathi¹, Oleg Michailovich², Frederik Laun³, and Carl-Fredrik Westin^4
1Psychiatry, Harvard Medical School, Boston, MA, United States, ²University of Waterloo, Ontario, Canada, ³German Cancer Research Center, Germany, Germany,⁴Harvard Medical School, MA, United States

We propose a novel method for reconstructing the diffusion signal in the q-space from very sparsely sampled data spread in over multiple b-values. Our preliminary results on a physical phantom shows that at-least 60 measurements are needed for good reconstruction quality of the data.

Kouhei Kamiya^1,2, Masaaki Hori¹, Yuriko Suzuki³, Atsushi Nakanishi¹, Issei Fukunaga¹, Koji Kamagata¹, Mariko Yoshida¹, Masakazu Miyajima⁴, Hajime Arai⁴, and Shigeki Aoki^1
1Department of Radiology, Juntendo University School of Medicine, Bunkyo, Tokyo, Japan, ²Deparment of Radiology, Graduate School of Medicine, the University of Tokyo, Bunkyo, Tokyo, Japan, ³Philips Electronics Japan, Ltd, Tokyo, Japan, ⁴Department of Neurosurgery, Juntendo University Graduate School of Medicine, Bunkyo, Tokyo, Japan

The gait disturbance is a characteristic treatable symptom of idiopathic normal pressure hydrocephalus (iNPH), though the etiology is not clear yet. This study evaluated axon diameter and axon density of the cortico-spinal tract (CST) in iNPH patients by using two-component low-q fit method of q-space imaging. Compared with controls, patients with iNPH had increased axon density of the CST running near the lateral ventricle, whereas the axon diameter was not altered. The present results suggest the idea that the gait disturbance in iNPH is not due to irreversible axonal damage, but results from compression and/or stretching of neural tracts.

Alan Seth Barnett^1,2, Elizabeth Hutchinson^1,2, Ferenc Horkay¹, Michal Komlosh^1,2, Evren Ozarslan³, Carlo Pierpaoli¹, and Peter Basser^1
1NICHD, NIH, Bethesda, MD, United States, ²CNRM, USUHS, Bethesda, MD, United States, ³Department of Radiology, Harvard Medical School, Boston, MA, United States

Mean apparent propagtor MRI (MAP-MRI) is a method of using diffusion MRI to map tissue microstructure. To evaluate the accuracy and precision of MAP-MRI we use a test system consisting of an array of microscopic cylinders. We use both simulated data, computed using the multiple correlation function (mcf) method, and real diffusion imaging data of a special phantom. We report the accuracy and precision of several MAP parameters as a function of the signal-to-noise ratio of the non-diffusion weighted images.

Luis Miguel Lacerda¹, Jonathan I. Sperl², Gareth Barker¹, and Flavio Dell'Acqua^1
1Centre for Neuroimaging Sciences, Institute of Psychiatry, King's College London, London, Denmark Hill, United Kingdom, ²GE Global Research, Munich, BY, Germany

Several techniques have been used to explore the information contained in the diffusion propagator. A standard methodology to evaluate such techniques (and the influence of factors such as acceleration methods) is to compare the reconstructed orientation distribution functions against a predefined "gold standard". Nonetheless additional measures may retrieve more information from the diffusion propagator. In this work we present preliminary results that may allow a more robust analysis and comparison of diffusion data and acceleration methods. Specific indices were extracted and appear to be helpful in the characterization of the diffusion propagator and in the comparison of different reconstruction methods

Aurobrata Ghosh¹ and Rachid Deriche^1
1Project Team Athena, Inria Sophia Antipolis Méditerranée, Sophia Antipolis, PACA, France

The Fiber Orientation Density (FOD) is a robust method for mapping crossing WM fibers. However, in clinical settings with minimalistic (~30) acquisitions, the FOD is restricted to 4th-6th order SHs, which limits its angular resolution. We proposed a non-negatively constrained sparse recovery of the FOD based on Non-Negative Least-Squares (NNLS) to overcome this limitation. We found NNLS solutions to be constrained and sparse. Here we experimentally compare the NNLS to l1-minimization and find NNLS superior in sparsity & robustness. We conclude by highlighting the greedy design of NNLS, which mirrors Orthogonal Matching Pursuit, and is the cause of its sparsity.

Eleftherios Garyfallidis¹, Mauro Zucchelli², Jean-Christophe Houde³, and Maxime Descoteaux^3
1University of Sherbrooke, Sherbrooke, Quebec, Canada, ²University of Verona, Italy, ³University of Sherbrooke, Quebec, Canada

The purpose of this work is to compare and extend recent state-of-the-art single-shell and multi-shell local reconstruction techniques for the HCP sampling scheme consisting of three-shells. We are interested in the best ODF reconstruction quantified locally with angular error (AE) and correct number of fibre compartments but most importantly, on the ODF reconstruction that leads to the best overall tractography output evaluated by the Tractometer.

Koichi Oshio¹, Hiroshi Shinmoto², and Robert Mulkern^3
1Department of Diagnostic Radiology, Keio University, Shinjuku-ku, Tokyo, Japan, ²Department of Radiology, National Defence Medical College, Tokorozawa, Saitama, Japan, ³Department of Radiology, Children's Hospital, Boston, MA, United States

Although many models have been proposed to interpret non-Gaussian diffusion MRI data in biological tissues, it is often difficult to see the correlation between the MRI data and the histological changes in the tissue. In this work, a statistical model based on the gamma distribution is proposed. Using this model, the diffusion MR data is well fitted, and histological interpretation of the data is possible.

Jeff Kershaw¹, Sayaka Shibata¹, Ichio Aoki¹, Takayuki Obata², and Hiroshi Ito^1
1Molecular Imaging Centre, National Institute of Radiological Sciences, Chiba, Japan, ²Research Centre for Charged Particle Therapy, National Institute of Radiological Sciences, Chiba, Japan

An earlier study of ex vivo mouse brain was the first to demonstrate that increasing the MPG frequency strongly enhanced the mean diffusivity in the cerebellar granule cell layer of the cerebellum. There was also a much weaker increase for the cerebellar molecular layer. However, no data was presented for the cerebellar white matter (CBWM), which is odd because investigating the diffusion characteristics of white matter is one of the main targets of diffusion-weighted MRI. The present study was performed to address this omission and report the first quantitative OGSE DTI measurements for mouse CBWM.

Henrik Lundell¹, Casper Kaae Sønderby¹, and Tim B Dyrby^1
1Danish Research Centre for Magnetic Resonance, Copenhagen University Hospital Hvidovre, Hvidovre, Denmark

Oscillating gradient spin echo (OGSE) experiments offers new insight in tissue microstructure by probing diffusion processes on ultra-short time scales. In this study we examine different possibilities for sampling OGSE on anisotropic tissue. We find that the best results for very limited gradient strength is achieved with square/trapezoidal gradients in 6 gradients, whereas a circularly polarized OGSE (CP-OGSE) perform better at higher gradient strengths. Our results emphasize that the optimal solution is highly dependent on the given limitations and the parameters of interest.

Ned Charles¹, Gary Cowin², Nyoman Kurniawan², and Roger Bourne^3
1University of Sydney, Camperdown, NSW, Australia, ²University of Queensland, Brisbane, QLD, Australia, ³University of Sydney, Sydney, NSW, Australia

Multi b-value data from 80 micrometer voxels was acquired from fixed prostate tissue in a 16T microimaging system. Biexponential modeling demonstrated distinct regional variations in diffusion behavior that correlate with microscopic tissue structure features. Regions dense in glands have a higher proportion of the lower diffusivity component, and this component has lower diffusivity in the glandular regions than in regions of fibromuscular stroma. An increasing partial volume of low diffusivity epithelial cells, rather than “higher cellularity”, may explain the clinical observation of decreasing ADC as prostate cancer Gleason grade increases.

Debbie Anaby¹, Darya Morozov¹, Ian D. Duncan², and Yoram Cohen^1,3
1School of Chemistry, Tel Aviv University, Tel Aviv, Israel, ²Department of Medical Sciences, University of Wisconsin-Madison, Wisconsin, United States, ³Sagol School of Neurosciences, Tel Aviv University, Tel Aviv, Israel

Conventional single-pulsed-field-gradient (s-PFG) methodologies are capable of faithfully depicting diffusion anisotropy in coherently ordered structures. However, microstructure is more difficult to characterize where macroscopic organization is absent. Double-PFG (d-PFG) MR methodologies were recently suggested as an alternative for studying microstructure in such cases. The spinal cords’ white matter (WM) of Long Evans shaker (les) rats, a dysmyelinated model, was recently studied by q-space MRI. Here, we characterize the spinal cords of les rats and their controls by d-PGSE, focusing on the gray matter (GM), in comparison with DTI. d-PFG clearly differentiates between three GM areas of the les spinal cord.

Christian Langkammer¹, Nikolaus Krebs², Christoph Birkl¹, Lukas Pirpamer¹, Florian Borsodi¹, Michaela Haindl¹, Gernot Reishofer³, David Andrew Porter⁴, Eva Scheurer², Franz Fazekas¹, and Stefan Ropele^1
1Department of Neurology, Medical University of Graz, Graz, Austria, ²Ludwig Boltzmann Institute for Clinical-Forensic Imaging, Graz, Austria, ³Department of Radiology, Medical University of Graz, Graz, Austria, ⁴Siemens Healthcare, Erlangen, Germany

The aim of this study was to investigate the feasibility of high-resolution diffusion MRI in the unfixed human brain as basis for validating diffusion-derived measures with histology.

Christian Wieseotte^1,2, Manfred Wagner², and Laura Schreiber^1
1Department of Radiology, Section of Medical Physics, Johannes Gutenberg University Medical Center, Mainz, Germany, ²Max Planck Institute for Polymer Research, Mainz, Germany

The diffusivity of contrast agents in the human body is a fundamental property which is still a factor of uncertainty in many applications. The main reason is that due to the highly paramagnetic Gd³⁺ ion, direct observations by means of nuclear magnetic resonance are not possible. This study pursues the approach of directly measuring the diffusion of the hydrodynamically analogue PET tracer Gd-DOTA in deuterated water. The diffusivity in blood plasma can be calculated by correcting for the higher solvent viscosity. We report a diffusivity of Ga-/Gd-DOTA in blood plasma at 310K of D = 2.92 (0.25)∙10^-10 m²s-1.

Tim B Dyrby¹, Mark Burke², Daniel C Alexander³, and Maurice Ptito^1,4
1Danish Research Centre for Magnetic Resonance, Copenhagen University Hospital Hvidovre, Hvidovre, Denmark, ²Departments of Physiology and Biophysics, Howard University, Washington, United States, ³Centre for Medical Image Computing, University College London, London, United Kingdom, ⁴School of Optometry, University of Montreal, Montreal, Canada

Axon diameter (AD) estimates from diffusion-weighted imaging (DWI) is known being overestimated when compared with electron microscopy (EM). Mean AD in ten regions across the corpus callosum (CC) was obtained from painstaking EM. The AD’s from DWI using ActiveAx method were found overestimated but similar AD contrast as with EM was observed across CC. With classical histology, undulating axons was observed throughout CC, and crossing axons in genu also. Undulation leads to larger cross-sectional area than straight axons hence AD overestimation. This suggests that macroscopic features as undulating axons can explain the overestimation we observed from AD estimates using DWI.

Marco Reisert¹ and Valerij G. Kiselev^1
1Medical Physics, University Medical Center Freiburg, Freiburg, Germany

Diffusion-Weighted MRI can reveal the microstructural organization of human brain white matter. In this work the diffusion model fitting is formulated as a global optimization problem assuming spatial coherence of the underlying diffusion parameters, which enables us to pool the variance from different voxels by their anatomical locations. The decision for single or multi-fiber voxels is directly integrated into the optimization process by using the Bayesian Information criterion. From a computational point of view the problem is quite hard, it is non-convex and combinatorial. At the heart of our approach lies a reversible jump Monte-Carlo Markov-Chain (RJMCMC) sampling strategy.

Enrico Kaden¹, Frithjof Kruggel², and Daniel C. Alexander^1
1Centre for Medical Image Computing, University College London, London, United Kingdom, ²Department of Biomedical Engineering, University of California, Irvine, Irvine, CA, United States

Diffusion MR imaging has enabled us to study the geometry of white matter in the individual human brain noninvasively. The present work aims to quantify the voxel-averaged diffusion coefficients parallel to an axon and perpendicular to it without any prior knowledge about how the axons are oriented within a voxel because the fiber orientation distribution is typically not known in advance. The axonal water diffusivity reflects the fiber microanatomy such as the axon diameter, the myelination, and the interaxonal space.

John G Georgiadis^1,2, Caroline G Tennyson¹, Danchin D Chen¹, Armen A Gharibans¹, and Curtis L Johnson^1,2
1Mechanical Science and Engineering, University of Illinois at Urbana-Champaign, Urbana, IL, United States, ²Beckman Institute, University of Illinois at Urbana-Champaign, Urbana, IL, United States

The successful employment of DTI to assess muscle quality in older adults in vivo hinges on connecting the signal to muscle cytoarchitecture. Seeking to explore the well-established anisotropy of the DTI metrics in muscle, the diffusion-weighted signal decay along the secondary and tertiary eigenvector tracts as a function of b-value was measured in the vastus medialis muscle of a single subject. We introduced a continuum myofiber model and attributed the non-exponential signal decay with b-value to the presence of myocellular diffusion barriers. This behavior cannot be accounted by lumped two-compartment models commonly considered in skeletal muscle DTI.

Aymeric Stamm¹, Benoit Scherrer¹, Stefano Baraldo², Olivier Commowick³, and Simon Warfield^1
1Computational Radiology Laboratory, Boston Children's Hospital, Boston, MA, United States, ²MOX, Politecnico di Milano, Milan, Italy, ³VISAGES, INRIA, Rennes, France, Metropolitan

Noise in diffusion MRI is known to be characterized by a non-central chi distribution. Many denoising methods have accounted for this but, for the estimation of diffusion models, the noise is most of the time still approximated by a Gaussian distribution. In this abstract, we examine the impact of this approximation to determine the optimal number of fascicles required for the estimation of multi-compartment models. We show that performing the models' estimation within a non-central chi framework significantly reduces over-fitting thus yielding a more reliable selection of the optimal number of fascicles.

Eleftheria Panagiotaki¹, Rachel W Chan², Nikolaos Dikaios², Hashim U Ahmed³, David Atkinson², Shonit Punwani², David J Hawkes¹, and Daniel C Alexander^1
1Centre for Medical Image Computing, University College London, London, London, United Kingdom, ²Centre for Medical Imaging, University College London, London, United Kingdom, ³Research Department of Urology, Div of Surgery & Interventional Sci, University College London, London, United Kingdom

This study utilised diffusion-weighted MRI to probe in vivo the microstructure of benign and cancerous prostate tissue non-invasively. For this purpose, a mathematical model was employed to capture the Vascular, Extracellular and Restricted DIffusion for Cytometry in Tumours (VERDICT). We demonstrate the technique in patients with prostate cancer of Gleason score 7 (3+4) in the peripheral zone (all imaged before undergoing biopsy). Experiments using the VERDICT model identified a significant increase in vascular volume in the tumour regions. The standard apparent diffusion coefficient (ADC) and intravoxel incoherent motion (IVIM) models failed to detect any significant differences between benign and tumour areas.

Eleftheria Panagiotaki¹, Simon Walker-Samuel², Bernard Siow^1,2, Peter S Johnson³, Vineeth Rajkumar³, R.Barbara Pedley³, Mark F Lythgoe², and Daniel C Alexander^1
1Centre for Medical Image Computing, University College London, London, London, United Kingdom, ²Centre for Advanced Biomedical Imaging, University College London, London, United Kingdom, ³UCL Cancer Institute, University College London, London, United Kingdom

This work presents a novel technique that quantifies and maps histological features of tumours, in vivo. The technique, named VERDICT (Vascular, Extracellular and Restricted DIffusion for Cytometry in Tumors), couples DW-MRI with a mathematical model of tumour tissue to access features such as cell size, vascular volume fraction, intra- and extracellular volume fractions, and pseudo-diffusivity associated with blood flow. We demonstrate the technique in two tumour xenograft models of colorectal cancer with contrasting cellular and vascular phenotypes. Experiments identified known differences in the microstructure of the tumours and a significant decrease in cell volume following administration of gemcitabine, possibly reflecting apoptotic volume decrease.

Gemma Nedjati-Gilani¹, Matt G Hall¹, Claudia Angela M Wheeler-Kingshott², and Daniel C Alexander^1
1Computer Science & Centre for Medical Image Computing, University College London, London, London, United Kingdom, ²Department of Neuroinflammation, Institute of Neurology, UCL, London, United Kingdom

Deriving analytical models of the diffusion MR signal which account for permeability is inherently difficult and often requires strong assumptions to be made about the compartmentation of water within the tissue. Given these problems, in this study we construct a computational model using Monte Carlo simulations and machine learning. We use random forest regression to learn a mapping between simulations and microstructure parameters and obtain an efficient and accurate model for microstructure imaging that accounts for permeability. We show that unseen microstructure parameters are well-predicted by the random forest regressor for both noise-free and noisy data.

Hamed Y. Mesri^1,2, Kelvin J. Layton^1,3, Iven M. Y. Mareels¹, and Leigh A. Johnston^1,3
1Department of Electrical and Electronic Engineering, The University of Melbourne, Melbourne, Victoria, Australia, ²Victoria Research Laboratory, National ICT Australia, Melbourne, Victoria, Australia, ³Florey Institute of Neuroscience and Mental Health, Melbourne, Victoria, Australia

Cramer Rao Lower Bound analysis is used to evaluate two compartment hindered/restricted diffusion models for estimation of mean axon diameter or axon diameter distributions from diffusion weighted MRI data. Our best-case model analysis demonstrates that the models are prone to high uncertainty levels. In practice, experimental data is acquired in regimes far from best-case model assumptions. Thus estimator performance is necessarily worse than the Cramer Rao error rates, which casts doubt on the ability of these models to robustly estimate microstructural features from diffusion MRI data. The Cramer Rao analysis technique is extensible to all parametric model-based inference methods.

Uran Ferizi^1,2, Torben Schneider², Eleftheria Panagiotaki¹, Maira Tariq¹, Hui Zhang¹, Claudia A. M. Wheeler-Kingshott², and Daniel C. Alexander^1
1Department of Computer Science and Centre for Medical Image Computing, University College London, London, United Kingdom, ²NMR Research Unit, Department of Neuroinflammation, Queen Square MS Centre, UCL Institute of Neurology, London, United Kingdom

In this work we compare parametric diffusion MRI models which explicitly seek to explain fibre dispersion in nervous tissue. These models aim at providing more specific biomarkers of disease by disentangling these structural contributions to the signal. Some models are drawn from recent work in the field; others have been constructed from combinations of existing compartments that aim to capture both intracellular and extracellular diffusion. To test these models we use a rich dataset acquired in vivo on the corpus callosum of a human brain, and then compare the models via the Bayesian Information Criteria. We test this ranking via bootstrapping on the data sets, and cross-validate across unseen parts of the protocol. We find that models that capture fiber dispersion are preferred. The results show the importance of modelling dispersion, even in apparently coherent fibers.

Aymeric Stamm¹, Benoit Scherrer¹, Olivier Commowick², Christian Barillot³, and Simon Warfield^1
1Computational Radiology Laboratory, Boston Children's Hospital, Boston, MA, United States, ²VISAGES, INRIA, Rennes, France, Metropolitan, ³VISAGES, CNRS, Rennes, France, Metropolitan

It is well known that the white matter has a complex architecture composed mainly of axon bundles or fascicles and glial cells. Fascicles cross in most parts of the white matter and multi-compartment models have been devised to study this complex microstructure. These models require that the number of compartments is known a priori, which is not the case in practice. In particular, determining the number of fascicles is difficult. It can however be reliably estimated from the generalization error at the cost of huge computational time. We propose a novel approach that relies on model averaging theory and generates the same results as the generalization error in a dramatically reduced computational time.

Damien J McHugh^1,2, Penny L Hubbard^1,2, Josephine H Naish^1,2, and Geoffrey JM Parker^1,2
1Centre for Imaging Sciences, University of Manchester, Manchester, United Kingdom, ²Biomedical Imaging Institute, University of Manchester, Manchester, United Kingdom

In this work we compare the diffusion-weighted MRI signals obtained from diffusion experiments in ordered and disordered microstructural environments. We used a simple tissue model of impermeable spherical cells and performed separate Monte Carlo diffusion simulations for ordered and disordered cell packing geometries. Synthetic signals were generated for pulse sequence parameters available on clinical scanners. Differences between the signals from ordered and disordered microstructures were small when extra-cellular diffusion alone was considered, and these differences were even smaller when considering combined intra- and extra-cellular diffusion. This suggests that the diffusion signal is insensitive to the packing geometry’s order/disorder for the sequence parameters and tissue properties covered here.

Benoit Scherrer¹, Maxime Taquet¹, Mustafa Sahin², Sanjay P. Prabhu¹, and Simon K. Warfield^1
1Radiology, Harvard Medical School, Boston Children's Hospital, Boston, MA, United States, ²Neurology, Harvard Medical School, Boston Children's Hospital, Boston, MA, United States

We propose a novel biophysical model of the diffusion that characterizes the distribution of three-dimensional (3D) microstructural environments in each voxel. This enables characterization of each fascicle in each voxel and of isotropic diffusion and may lead to novel biomarkers and novel investigations of the white-matter microstructure.

Sheryl L Herrera¹, Trevor J Vincent^2,3, Morgan E Mercredi¹, Richard Buist⁴, Christopher P Bidinosti^1,2, and Melanie Martin^2,5
1Physics & Astronomy, University of Manitoba, Winnipeg, Manitoba, Canada, ²Physics, University of Winnipeg, Manitoba, Canada, ³Astronomy & Astrophysics, University of Toronto, Ontario, Canada, ⁴Radiology, University of Manitoba, Manitoba, Canada, ⁵Physics & Astronomy, Radiology, University of Manitoba, Manitoba, Canada

We simulate cylindrical geometries using OGSE and AxCaliber to determine diameter distributions. The fit results agree reasonably with the geometry. Numbers of small axons are underestimated and larger axons are overestimated. The calculated mean radius (1.570.01µm) is comparable with the actual mean radius (1.67µm). The CHARMED model simulates the extracellular space as hindered, whereas ours was restricted. Because the CHARMED model signal is proportional to r², the larger axons are overestimated. With noise the smaller axons are harder to identify. These reasons could account for some discrepancy between the fitted and actual data. Modifications are necessary to the CHARMED model.

Andre Bongers¹, Aritrick Chatterjee², Geoffrey Watson³, and Roger Bourne^4
1Mark Wainwright Analytical Centre, University of New South Wales, Sydney, NSW, Australia, ²University of Sydney, Sydeny, NSW, Australia, ³Royal Prince Alfred Hospital, Sydney, NSW, Australia, ⁴University of Sydney, Sydney, NSW, Australia

As DWI is able to sensitively sample cellular changes in tissues the method is a promising biomarker for prostate cancer detection. Standard PGSE methods usually sample ADCs in the long diffusion time regime, meaning that ADC quantification is affected by cellular restrictions from multiple scales. This study explores the ability of oscillating gradient DWI (OGSE) methods to detect subcellular changes in fixed and unfixed ex-vivo prostate tissue at high magnetic field. Our findings show that -in the short diffusion time regime- ADC-diffusion-time behavior depends significantly on prostate tissue type and suggest that that prostate cancer induces intracellular structure changes.

Dan Wu¹, Frances J Northington², and Jiangyang Zhang^3
1Biomedical Engineering, Johns Hopkins University School of Medicine, Baltimore, Maryland, United States, ²Pediatrics, Johns Hopkins University School of Medicine, Baltimore, Maryland, United States, ³Radiology, Johns Hopkins University School of Medicine, Baltimore, Maryland, United States

In this study, we showed that in vivo oscillating gradient spin echo (OGSE) diffusion MRI provided unique tissue contrasts to delineate structures in the adult mouse cerebellum. Comparisons between in vivo and ex vivo OGSE data showed that death and fixation significantly reduced OGSE ADC measurements and altered their frequency dependency. In neonatal mouse brains with hypoxic-ischemic injury, in vivo OGSE showed larger increase in ADC with frequency in regions with cytotoxic edema than normal tissues, and thereby reduced edema contrast at high oscillating frequencies, which suggested that the microstructural changes under cytotoxic edema were within certain spatial scales.

Marco Lawrenz^1,2 and Juergen Finsterbusch^1,2
1Department of Systems Neuroscience, University Medical Center Hamburg-Eppendorf, Hamburg, Hamburg, Germany, ²University Medical Centers Hamburg-Kiel-Luebeck, Neuroimage Nord, Hamburg-Kiel-Luebeck, Hamburg-Kiel-Luebeck, Germany

Double-wave-vector diffusion-weighting (DWV) experiments involving two successively applied diffusion weightings are able to detect diffusion anisotropy on a microscopic scale, even if the diffusion macroscopically appears isotropic. This has been demonstrated recently in a human brain white matter region that yields a fractional anisotropy of 0. In this study, human gray matter is targeted in vivo using a DWV sequence with inversion recovery to suppress WM signals and avoid partial volumen effects. While no signal modulation is observed in a fluid phantom, the modulation curve characteristic for diffusion anisotropy is observed although the modulation amplitude is reduced compared to WM.

Casper Kaae Sønderby¹, Henrik Lundell¹, and Tim B. Dyrby^1
1Danish Research Centre for Magnetic Resonance, Copenhagen University Hospital, Hvidovre, Denmark

Double Pulsed Field Gradient sequences have recently been used to detect novel tissue contrasts such as the Apparent Exchange Rate and Compartment Eccentricity. We compare the effects of compartmental specific T2-relaxation in both single and double pulsed field gradient sequences. We measure a decrease in the Apparent Diffusion Coefficient as either the mixing time or the echo time is increased indicating filtering of compartment specific signal components with fast diffusivities. Our results indicate that both single and double pulsed field gradient sequences should adopt for the same TE in all experiments to maintain the relative T2-relaxation weighting of the individual compartments.

Gregory T Baxter¹, Vitaly Galinsky¹, Evren Ozarslan², Peter J Basser³, and Lawrence R Frank^1,4
1Center for Scientific Computation in Imaging, Dept. of Radiology, University of California, San Diego, CA, United States, ²Department of Radiology, Brigham and Women's Hospital, Harvard Medical School, Boston, MA, United States, ³Section on Tissue Biophysics and Biomimetics, PPITS, NICHD, National Institutes of Health, Bethesda, MD, United States, ⁴Center for Functional MRI, Dept of Radiology, University of California, San Diego, CA, United States

We present Monte-Carlo simulations of diffusion within hexagonally packed arrays of cylinders mimicking the white-matter, and report the signal behavior for double-PFG experiments with zero mixing times. We investigate the accuracy of the axonal diameter estimates when a bi-compartmental model that assumes Gaussian diffusion outside the cells is employed. Despite the substantially underestimated bulk diffusivity values, cell diameter and volume fraction estimates are both adequately captured by such a bi-compartmental model. Some underestimation of cell diameters is expected only for tightly packed configurations.

Marco Lawrenz^1,2 and Juergen Finsterbusch^1,2
1Department of Systems Neuroscience, University Medical Center Hamburg-Eppendorf, Hamburg, Hamburg, Germany, ²Neuroimage Nord, University Medical Centers Hamburg-Kiel-Luebeck, Hamburg-Kiel-Luebeck, Hamburg-Kiel-Luebeck, Germany

Double-wave-vector diffusion-weighting (DWV) experiments with two weighting periods applied successively offer access to microscopic tissue properties, e.g. the diffusion anisotropy present on a microscopic scale. This anisotropy is independent of the axon orientation distribution and, thus, may better reflect axon integrity than measures derived from DTI. To investigate whether measures of th microscopic diffusion anisotropy, e.g. the so-called MA index, are feasible for application in clinical or neuroscientific research, normal values in white matter and their variation within a group of healthy volunteers as well as their reproducability within and between sessions are determined.

Björn Lampinen¹, Filip Szczepankiewicz¹, Danielle van Westen², Pia Sundgren², Freddy Ståhlberg^1,2, Jimmy Lätt³, and Markus Nilsson^4
1Dpt. of Medical Radiation Physics, Lund University, Lund, Sweden, ²Dpt. of Diagnostic Radiology, Lund University, Lund, Sweden, ³Center for Medical Imaging and Physiology, Lund University Hospital, Lund, Sweden, ⁴Lund University Bioimaging Center, Lund University, Lund, Sweden

Water exchange can be mapped noninvasively in diffusion MRI using filter-exchange imaging (FEXI) to obtain the apparent exchange rate (AXR). We here aim to prepare FEXI for future clinical research that investigates differences in mean AXR across populations. A test-retest study is conducted on 18 healthy volunteers to assess how inter-subject differences and measurement error contribute to variance in AXR in various brain regions. A power analysis is then performed to interpret AXR variance in terms of study group sizes required to detect a range of differences while ensuring a statistical power of 0.8. We show that with study groups under 10 subjects, mean differences between populations of 40-80% in AXR, depending on the region, can be detected.

Debbie Anaby¹, Darya Morozov¹, and Yoram Cohen^1,2
1School of Chemistry, The Raymond and Beverly Sackler Faculty of Exact Science, Tel-Aviv University, Tel-Aviv, Ramat-Aviv, Israel, ²Sagol School of Neurosciences, Tel-Aviv University, Ramat-Aviv, Israel

Angular double-pulsed-field gradient (d-PFG) MRI has been suggested as an alternative for studying microstructure in the brain. It has been shown that angular d-PFG MR experiments performed with long tms are able to distinguish between compartments with different eccentricities. Here, we study the effects of different experimental parameters on the obtained apparent eccentricity (aE) and residual phase maps of an ex-vivo rat brain obtained from angular d-PFG MRI collected with finite tms.

Valerij G. Kiselev¹ and Bibek Dhital^1,2
1Department of Radiology, Medical Physics, University Medical Center Freiburg, Freiburg, Germany, ²German Cancer Consortium (DKTK), Heidelberg, Germany

We discuss two major challenges of diffusion measurements with oscillating gradients: (i) achieving the highest frequency on preclinical scanners and (ii) performing measurements on human scanners. In order to respond to (i), we propose a ramp that does not compromise the maximal achievable frequency below the hardware limits and show explicitly the parameter range for which this extension is crucial. Other important ingredients of diffusion measurements with oscillating gradient are the relative phase of two gradient trains in the spin echo measurements, fast ramps for low-frequency measurements in human scanners and the optimal total gradient duration.

Andreas Wetscherek¹, Bram Stieltjes², and Frederik Bernd Laun^1,2
1Medical Physics in Radiology (E020), DKFZ, Heidelberg, Germany, ²Quantitative Imaging Based Disease Characterization (E011), DKFZ, Heidelberg, Germany

Diffusion experiments using oscillating gradients are performed. In the pancreas of healthy volunteers, sine gradients were found to cause larger signal attenuation than cosine gradients of the same b-value. In addition, the signal attenuation was found to decrease for the cosine profile when the number of oscillations was increased. The experimental findings can be understood in the picture of temporal diffusion spectroscopy and in the intravoxel incoherent motion model. The method might be of particular interest for probing the microvasculature in organs in which the characteristic timescale ô of the incoherent motion is shorter than the echo time.

Matt G Hall¹, Gemma Nedjati-Gilani¹, and Daniel C Alexander^1
1Centre for Medical Image Computing, University College London, London, United Kingdom

We present a new method that enables Monte Carlo simulation of diffusing spins for diffusion MR data synthesis to be performed using tissue regions with sizes comparable to a typical scan voxel. We show how the technique may be optimised such that run time is minimised and also show that the new technique leads to improved reproducibility in synthetic diffusion-weighted measurements.

Wen-Chau Wu^1,2, Kuan-Lin Chen³, and Shu-Hua Lien^3
1Graduate Institute of Oncology, National Taiwan University, Taipei, Taiwan, ²Graduate Institute of Clinical Medicine, National Taiwan University, Taipei, Taiwan, ³Medcial Imaging, National Taiwan University Hospital, Taipei, Taiwan

Intravoxel incoherent motion (IVIM) MRI is a technique proposed for concurrent measurement of water diffusion coefficient D and perfusion-related pseudo-diffusion coefficient D*. In this study, we investigated the sensitivity of IVIM MRI in the brain by comparing with arterial spin-labeling (ASL) MRI. Monte Carlo simulation was performed to assess the reliability of the perfusion indexes derived from nonlinear fitting. Results show that IVIM perfusion indexes are less reliable when D*/D < 10 and vascular volume fraction f < 0.05. Overall, f is more robust an index than D*. ASL-derived blood flow values were found mildly correlated with fD* and f.

Petra Pouwels¹, Frederik Barkhof², Rudolf Verdaasdonk¹, and Joost Kuijer^1
1Physics & Medical Technology, VU University Medical Center, Amsterdam, Netherlands, ²Radiology and Nuclear Medicine, VU University Medical Center, Amsterdam, Netherlands

Regional changes in DTI parameters were observed in healthy volunteers at 3T after a hardware upgrade. Pairwise comparison (mimicking longitudinal studies) showed significant FA differences in 26.7% of the WM skeleton, and in all diffusivities, especially in the body of the corpus callosum. Unpaired comparison (mimicking cross-sectional studies) still showed FA differences in 5.5% of the skeleton. A change in gradient system and related eddy-current behaviour might be the cause of the observations. No pairwise differences were observed in repeated measurements on the same scanner. These results emphasize the need to take scanner upgrade into account in the study design.

Jung-Hoon Lee^1,2, Sang-Young Kim¹, Do-Wan Lee¹, Jin-Young Jung¹, Kyu-Ho Song¹, and Bo-Young Choe^1
1Department of Biomedical Engineering, Research Institute of Biomedical Engineering, College of Medicine, The Catholic University of Korea, Seoul, Seoul, Korea, ²Department of Radiology, Kyunghee Medical Center, Seoul, Seoul, Korea

We evaluated geometric accuracy, slice position accuracy, image intensity uniformity, percent-signal ghosting and, low-contrast object detectability provided by ACR Guidance as well as image distortion, ADC and FA values measured with 6, 15, and 32 directions at 1.5T and 3.0T MR systems.

Francesca Pizzorni-Ferrarese¹ and Franco Pestilli^2
1Department of Psychology, The Royal Holloway, University of London, Egham, Surrey, United Kingdom, ²Psychology, Stanford University, Stanford, CA, United States

We propose a method to segments the human thalamus automatically using diffusion MR signal and to validate the accuracy of a alternative thalamic segmentations in individual brains. The method treats the clusters in the segmentation as models of the white matter tissue. Thalamic nuclei are assumed to have homogeneous tissue properties. Under this assumption the mean diffusion signal within each nucleus is used as prediction of the segmentation model. We use the model to validate a family of candidate thalamic segmentations by looking at how the reliability of the model prediction compares to the reliability of the measured diffusion data.

Benoit Scherrer¹, Maxime Taquet¹, Mustafa Sahin², Sanjay P Prabhu¹, and Simon K. Warfield^1
1Radiology, Harvard Medical School, Boston Children's Hospital, Boston, MA, United States, ²Neurology, Harvard Medical School, Boston Children's Hospital, Boston, MA, United States

We propose a novel framework to achieve quantitative evaluation of biophysical models of the diffusion with in-vivo data. It enables identification of the model that best predicts the diffusion signal, and therefore identification of the model that best captures the underlying biophysical mechanisms. We investigate the performance of DTI, NODDI and DIAMOND.

Pim Pullens¹, Wim van Hecke², and Paul M Parizel^1
1Radiology, Antwerp University Hospital, Edegem, Antwerp, Belgium, ²Icometrix, Leuven, Belgium

There is more and more interest in testing reproducibility of DTI measures, which could lead to use of DTI measures as clinical biomarkers. Coefficient of variation (std/mean) is often used to assess reproducibility. However, it is not trivial to compare CVs from different studies, each with a different number of subjects. In this meta-analysis across DTI reproducibility studies, we test if CVs are equal among published FA reproducibility studies with an appropriate statistical test. Results show that although confidence intervals of FA show deviations because of analysis or scanner type, CVs are comparable.

David M. Morris¹, Hossein Ragheb¹, Neil Thacker¹, Naomi H.M. Douglas², and Alan Jackson^3
1Centre for Imaging Sciences, University of Manchester, Manchester, United Kingdom, ²Institute of Cancer Research, Sutton, United Kingdom, ³Wolfson Molecular Imaging Centre, University of Manchester, Manchester, United Kingdom

Apparent diffusion Coefficient (ADC) is a biomarker shown to have efficacy in monitoring the response of tumors to interventions. To be used successfully in multi-center trials a quality assurance protocol must be developed using an ice water phantom. The results of a multi-vendor, multi-site trial using standardized protocols are presented where different measurements of the accuracy of the systems using the phantom are compared. Then the relationship of these with the variability observed within the normal liver of volunteers at each site. While the protocols can differentiate between scanners there is no predictive power for performance in-vivo.

Neil P Jerome¹, Keiko Miyazaki¹, David J Collins¹, Matthew R Orton¹, James d'Arcy¹, Lucas Moreno², Andrew D J Pearson², Lynley V Marshall², Fernando Carcellar², Martin O Leach¹, Stergios Zacharoulis², and Dow-Mu Koh^3
1CR-UK and EPSRC Cancer Imaging Centre, The Institute of Cancer Research, Sutton, Surrey, United Kingdom, ²Paediatric Drug Development Team, Cancer Therapeutics and Clinical Studies, The Institute of Cancer Research, Sutton, Surrey, United Kingdom, ³Department of Radiology, Royal Marsden Hospital, Sutton, Surrey, United Kingdom

Diffusion-weighted imaging provides a sensitive way of detecting functional changes that may precede changes in tumour size and in response to cytostatic effects; though widely used for functional imaging in adult clinical trials, there is a need to establish the repeatability of DWI-derived parameters in children. For a paediatric cohort (median age 11, range 6 – 15 years) of seven extra- and eight intra-cranial solid tumours, DWI was performed on two successive days, showing lowest CV values for ADC with varying b-value inclusion (CV 2.4 – 4.1%), and for D derived from IVIM (CV 2.5%. IVIM perfusion-related parameters were less repeatable (CV>25%).

Marc Buzzelli¹, XiaoTian Li², William Randazzo¹, and Nathan Yanasak^1
1Department of Radiology and Imaging, Medical College of Georgia, Georgia Regents University, Augusta, Georgia, United States, ²Medical College of Georgia, Georgia Regents University, Augusta, Georgia, United States

Problem To characterize and apply thermal and signal-to-noise corrections to DTI images of a specialized quality assurance phantom via regression, to facilitate future investigation of systematic biases in clinical protocols. Methods A phantom was constructed for diffusion imaging. Scans were obtained utilizing a 3T magnet, monitoring the temperature of the phantom during the scans. FA and ADC maps were calculated. Dependencies in temperature and SNR were determined, to examine baseline systematic variability after regression. Results/Conclusion FA and ADC show linear dependence on temperature and an inverse dependence on SNR. Intrinsic dependences of FA on SNR are significant after regression.

Penny L Hubbard¹, Feng-lei Zhou², Stephen Eichhorn³, and Geoff JM Parker^2
1The University of Manchester, Manchester, Greater Manchester, United Kingdom, ²The University of Manchester, Manchester, United Kingdom, ³University of Exeter, Exeter, United Kingdom

We present a proof-of-principle study of comparing an aligned and simple crossing fibre phantoms composed on co-electrospun hollow fibres, with an inner diameter within the biological range.

Yi-Hsiu Hsiao¹, Cheng-Yu Chen^1,2, Ping-Huei Tsai¹, Hsiao-Wen Chung³, Ming-Chung Chou⁴, Shih-Wei Chiang^3,5, Yung-Chieh Chen⁶, and Hung-Wen Kao^5
1Department of Medical Imaging and Imaging Research Center, Taipei Medical University Hospital, Taipei Medical University, Taipei, Taiwan, ²Graduate Institue of Clinical Medicine, Taipei Medical University, Taipei, Taiwan, ³Graduate Institute of Biomedical Electrics and Bioinformatics, National Taiwan University, Taipei, Taiwan, ⁴Department of Medical Imaging and Radiological Sciences, Kaohsiung Medical University, Kaohsiung, Taiwan, ⁵Department of Radiology, Tri-Service General Hospital, Taipei, Taiwan,⁶Department of Biomedical Imaging and Radiological Sciences, National Yang-Ming University, Taipei, Taiwan

The anterior thalamic nucleus (AN) plays an important role in communication and the actual mechanism has not been well-known. A reliable non-invasive imaging method which is capable of delineating the nucleus is desired. Several studies have demonstrated that MR diffusion tensor imaging (DTI) could help identify the individual regions of the thalamic nuclei by fiber tracking method, but the results may not actually reflect the true anatomic location. A reasonable solution has recently been proposed using constrained spherical deconvolution (CSD) tractography technique. In this study, we compared CSD-based method with conventional diffusion tensor (DT) model in localization of the AN and the fractional anisotropy.

Joelle E Sarlls¹, Carlo Pierpaoli², Qi Duan³, Wolfgang Devine⁴, and Hellmut Merkle^3
1NINDS/NMRF, National Institutes of Health, Bethesda, MD, United States, ²NICHD/STBB, National Institutes of Health, MD, United States, ³NINDS/LFMI, National Institutes of Health, MD, United States, ⁴NINDS-NIMH/RSB, National Institutes of Health, Bethesda, MD, United States

Although 7T provides increased MR signal, attractive to signal-starved diffusion imaging, there are also technical challenges like increase B1 inhomogeneity and susceptibility effects. Accordingly, there is much activity in sequence and hardware development. We introduce a calibrated diffusion phantom that provides the means to accurately test developed sequences and hardware at 7T. We utilized a 35% PVP solution providing a known diffusion coefficient, close to in-vivo brain, and mitigating flow effects. We show that utilizing a 17cm cylindrical shaped container fitted with a grid of 1x1cm plastic columns dramatically improves B1 homogeneity, removing bias in the calculated diffusion coefficient.

Stephanie L. Barnes^1,2, Natenael B. Semmineh^1,3, C. Chad Quarles^1,2, and Thomas E. Yankeelov^1,2
1Vanderbilt University Institute of Imaging Sciences, Vanderbilt University, Nashville, TN, United States, ²Radiology and Radiological Sciences, Vanderbilt University, Nashville, TN, United States, ³Physics and Astronomy, Vanderbilt University, Nashville, TN, United States

Standard evaluation of dynamic contrast enhanced MRI (DCE-MRI) data assumes instantaneous mixing of the contrast agent (CA) within a given voxel. This work aims to evaluate the effect of non-instantaneous mixing due to passive diffusion of the CA on quantitative DCE-MRI. We utilized a finite element model of a (250 µm²) voxel that divides the voxel into vascular, extravascular extracellular, and extravascular intracellular spaces. The results show that diffusion, plasma fraction, and temporal resolution have a substantial effect on the distribution of CA concentration and this can significantly affect the estimation of pharmacokinetic parameters.

Xiaopeng Zhou¹, Ken Sakaie¹, Erik Beall¹, Mark Lowe¹, and Robert Fox^2
1Imaging Institute, Cleveland Clinic, Cleveland, OH, United States, ²Neurological Institute, Cleveland Clinic, Cleveland, OH, United States

We describe the development of a practical quality assurance (QA) protocol for a multi-center diffusion tensor imaging trial. fBIRN phantom was scanned on two 3T scanners to evaluate temporal stability of SNR and DTI parameters. Robustness of QA metrics against positioning provides confidence that the protocol will be effective across sites. Although variation of tensor parameters is small, it does correlate with SNR, suggesting that QA measurements may serve as a valuable covariate for statistical analysis. Tensor parameters appear consistent across platforms. This DTI QA procedure is effective to track scanner performance and can provide reference for multi-center trial.

Irvin Teh¹, Maelene Lohezic¹, Dunja Aksentijevic^1,2, and Jurgen E Schneider^1
1Division of Cardiovascular Medicine, Radcliffe Department of Medicine, University of Oxford, Oxford, United Kingdom, ²Cardiovascular Division, The Rayne Institute, King's College London, London, United Kingdom

Gradient calibration is critical to accurate sampling of q-space in diffusion MRI, particularly for applications such as diffusion spectrum imaging (DSI) that rely on a precise relationship of q-space sampling. Cyclooctane possesses a number of properties that make it suitable for gradient calibration at high b-values, including isotropic diffusion, relatively low diffusivity and high viscosity, and a single proton resonance. We present the novel use of a phantom intended for gradient calibration in DSI and other high b-value q-space imaging applications.

Tiziana M Florio¹, Giuseppina Confalone¹, Loredana Cristiano¹, Alessia Fidoamore¹, Angelo Galante^1,2, Laura Brandolini³, Marcello Allegretti³, Annamaria Cimini¹, and Marcello Alecci^1,2
1Dipartimento Medicina Clinica, Sanita’ Pubblica, Scienze della Vita e dell'Ambiente, University of L'Aquila, L'Aquila, Italy, ²Istituto Nazionale di Fisica Nucleare, Laboratori Nazionali del Gran Sasso, Italy, ³Research Centre, Dompé pha.r.ma s.p.a., L'Aquila, Italy

We have investigated the time course from sacrifice of ADC and T2 maps of whole mouse brain during fixation with formaldehyde, comparing the immersion and perfusion methods. A non-fixed whole brain was studied for comparison.

Usha Sinha¹, Robert Csapo², Vadim Malis¹, Yanjie Xue¹, and Shantanu Sinha^2
1Physics, SDSU, San Diego, CA, United States, ²Radiology, UCSD, San Diego, CA, United States

Age related changes in muscle alter both fiber architecture (flengths and pennation angles) as well the microstructure (fiber type, diameter, fibrosis). Diffusion tensor imaging allows the mapping of fiber architecture as well as the microstructure. We report age related changes in diffusion tensor indices (eigenvalues and FA) of the medial and lateral gastrocnemius in a cohort of 5 young and 5 old women. All three eigenvalues for both muscles increase with age, significant difference in young and old was seen for nearly all the indices. A decreased fiber diameter and muscle fiber fraction may qualitatively explain the current findings.

Carson Ingo¹, Itamar Ronen¹, and Andrew G. Webb^1
1C. J. Gorter Center for High Field MRI, Department of Radiology, Leiden University Medical Center, Leiden, Netherlands

Recently, there was the first report of anomalous diffusion in DWS measurements of intracellular metabolites in the rat brain at 7T utilizing ultra-short diffusion times afforded by oscillating gradients. Here, we consider in vivo measurements of anomalous diffusion for tNAA, tCho, and tCr in the human brain at 7T with conventional diffusion times. Our study suggests for metabolites: 1) white matter anisotropy does not contribute to Gaussian or anomalous diffusion patterns, 2) compartmental distribution does not necessarily contribute to anomalous diffusion, and 3) the unique sub-diffusive behavior of tCr may be influenced by other factors like chemical exchange.

Chun-Hao Huang¹, Woan-Chyi Wang², Yi-Ru Lin¹, and Shang-Yueh Tsai^2
1Electronic and Computer Engineering, National Taiwan University of Science and Technology, Taipei, Taiwan, ²Graduate Institute of Applied Physics, National Chengchi University, Taipei, Taiwan

In this study we investigate the reproducibility of DTI and probabilistic tractography on the brain connectivity measures among 78 grey matter parcellations. Regular DTI protocols (30 directions, b=1000) were performed twice on each 14 healthy subject (7 male and 7 female in 20 to 25 years old). Results show that probabilistic tractography can be successfully applied to calculate the connectivity between GM regions. The inter-scan reproducibility is at level of 10% for GM regions with connectivity strength over 0.2. Therefore, structural connectivity assessed by diffusion tensor tractogrphy can be used to study the brain network.

Xi Wu^1,2, Xu Ran², Adam Anderson², and Zhaohua Ding^2
1Huaxi MRI Research center, Sichuan University, chengdu, sichuan, China, ²Vanderbilt University Institute of Imaging Science, Vanderbilt University, nashville, TENNESSEE, United States

We proposed an improved diffusion MRI-based fiber tractography using functional information. Bold signals are modeled as spatio-temporal correlation tensor and than used to constrain the tractography. This Constrained tracking method can ensure more accurate tracking direction estimation thus benefit more rigid tracking results.

Hong-Hsi Lee¹, Jeng-Wei Chen¹, and Wen-Yih Isaac Tseng^2
1Physics, National Taiwan University, Taipei, Taiwan, Taiwan, ²Center for Optoelectronic Medicine, National Taiwan University, Taipei, Taiwan, Taiwan

Utilizing the correspondence between diffusion equation and Schrodinger equation, we derived a Lagrangian and acquired an equation of motion. This equation is the core of our tractography algorithm which can be applied in the Diffusion Tensor Imaging (DTI) data. Since it contains gradient terms related to diffusion tensors, acute turnings of tracts are possible in several voxels. To demonstrate the feasibility, we simulated two-dimensional diffusion tensors from a picture of magnetic lines, and applied this algorithm to delineate tracts in the original picture. The simulation results show that this algorithm is viable.

Iman Aganj¹, Gautam Prasad², Priti Srinivasan¹, Anastasia Yendiki¹, Paul M. Thompson^2,3, and Bruce Fischl^1,4
1Martinos Center for Biomedical Imaging, Radiology Department, Massachusetts General Hospital, Harvard Medical School, Boston, MA, United States, ²Imaging Genetics Center, Institute for Neuroimaging and Informatics, University of Southern California, Los Angeles, CA, United States, ³Depts. of Neurology, Psychiatry, Engineering, Radiology and Ophthalmology, University of Southern California, Los Angeles, CA, United States, ⁴Computer Science and Artificial Intelligence Laboratory, Massachusetts Institute of Technology, Cambridge, MA, United States

-Structural brain connectivity – computed from diffusion-weighted MRI tractography – is useful in studying brain structure in health and disease. Current approaches for computing the structural brain network consider fiber bundles directly connecting brain regions, often disregarding indirect pathways relayed through other regions. Here we take multi-synaptic connections into account using mathematical tools developed for the analysis of resistive electrical circuits. Our results show that such an augmented network can improve the classification of Alzheimer’s disease patients from healthy controls.

Emmanuel Caruyer¹, Alessandro Daducci², Maxime Descoteaux³, Jean-Christophe Houde³, Jean-Philippe Thiran², and Ragini Verma^1
1Section of Biomedical Image Analysis, Department of Radiology, University of Pennsylvania, Philadelphia, PA, United States, ²Signal Processing Laboratory 5, École Polytechnique Fédérale de Lausanne, Lausanne, Vaud, Switzerland, ³Sherbrooke Connectivity Imaging Lab, Université de Sherbrooke, Sherbrooke, Québec, Canada

We introduce Phantomas, an open-source software library to create realistic phantoms in structural and diffusion MRI. The software allows evaluation of each step of a tractography processing pipeline: acquisition design, preprocessing, local reconstruction and tractography algorithm. By starting from a ground truth configuration of fiber bundle geometries, Phantomas provides a basis to quantitatively evaluate connectivity analyses, using metrics based on successful reconstruction of the true connectivity profile. Complementarily, local metrics can be computed to measure the accuracy of local reconstruction methods. We demonstrate that user can easily create crossing, kissing and bending fiber configurations, leading to complex intravoxel diffusion profiles.

Mohammad-Reza Nazem-Zadeh¹, Jason Schwalb², Hassan Bagher-Ebadian³, Fariborz Mahmoudi¹, Mohammad-Parsa Hosseini⁴, Hajar Hamidian¹, Abdelrahman Hassane¹, Oltion Meci¹, Harrini Vijay¹, Mohammad Emari¹, and Hamid Soltanian-Zadeh^1
1Radiology and Research Administration Departments, Henry Ford Hospital, Detroit, MI, United States, ²Neurosurgery Department, Henry Ford Hospital, Detroit, MI, United States, ³Neurology Department, Henry Ford Hospital, Detroit, MI, United States, ⁴Radiology and Research Administration Departments, Henry Ford Hospital, MI, United States

We hypothesizes that the hemispheric asymmetry of FA within the crura of fornix could be used to confirm the laterality of mesial temporal epileptogenicity. The TLE was associated with a reduced FA value in the fornix crus ipsilateral to the seizure onset. Outperforming anatomy-based lateralization methods, the proposed diffusion based method hold promise for improving decision-making for surgical resection and may reduce the need for implantation of intracranial monitoring electrodes.

Helen Schomburg¹, Thorsten Hohage¹, and Christoph Rügge^1
1Institute for Numerical and Applied Mathematics, University of Göttingen, Göttingen, Germany

We present an approach for semi-local fiber tractography on HARDI data extending local tractography by including diffusion information of neighboring regions to reduce accumulation of noise. Considering the neighborhood located behind the current path position we obtain a guiding direction from the previously tracked path. Then, the region in front is explored by determining a set of candidate directions. In addition, we take into account right and left neighbors to observe the position on the fiber bundle perpendicular to the stepping direction and shift the fiber path point toward the center if necessary. Tests are performed on diffusion phantom data.

Henrietta Howells¹, Flavio Dell'Acqua^2,3, Anoushka Leslie², Michel Thiebaut de Schotten^1,4, Mitul Mehta², Declan G. Murphy¹, Andrew Simmons², and Marco Catani^1
1Forensic and Neurodevelopmental Sciences, Institute of Psychiatry, London, United Kingdom, ²Department of Neuroimaging, Institute of Psychiatry, London, United Kingdom,³NIHR Biomedical Research Centre for Mental Health at SLAM NHS Foundation Trust, Institute of Psychiatry, London, United Kingdom, ⁴Université Pierre et Marie Curie-Paris 6, Centre de Recherche de l'Institut du Cerveau et de la Moelle épinière, Paris, France

The frontal aslant tract (FAT) is a newly described pathway connecting posterior Broca's territory with the pre-SMA and cingulate cortex. This tract is significantly left-lateralised in right-handed subjects and plays a role in speech initiation. It has not yet been studied in left-handers. We used spherical deconvolution diffusion tractography to compare this tract in right and left-handed subjects, finding left-lateralisation was not significant in the left-handed group. This may be reflective of more heterogeneous patterns of functional activation for language in left-handed individuals.

Dan Wu¹, Jiadi Xu², Susumu Mori^2,3, and Jiangyang Zhang^3
1Biomedical Engineering, Johns Hopkins University School of Medicine, Baltimore, Maryland, United States, ²F.M. Kirby Research Center for Functional Brain Imaging, Kennedy Krieger Institute, Baltimore, Maryland, United States, ³Radiology, Johns Hopkins University School of Medicine, Baltimore, Maryland, United States

In order to resolve the complex microstructures and neural connectivity in the gray matter structures, high spatial and angular resolution is critical but remains challenging in vivo. We have achieved high-resolution HARDI (0.1 mm isotropic, 60 directions) of live the mouse brain using localized imaging technique and a 3D fast imaging sequence. Several potential tracts within the mouse hippocampus were reconstructed, and single subject and group average tractography data showed the tracts originated from the CA1 and dentate gyrus agreed with existing tracer-based studies. The technique could potentially be used to non-invasively examine the neuronal pathways in gray matter structures.

Henrietta Howells¹, Marco Catani¹, Flavio Dell'Acqua^2,3, Anoushka Leslie², Andrew Simmons², Declan G. Murphy¹, and Michel Thiebaut de Schotten^1
1Forensic and Neurodevelopmental Sciences, Institute of Psychiatry, London, United Kingdom, ²Department of Neuroimaging, Institute of Psychiatry, London, United Kingdom,³NIHR Biomedical Research Centre for Mental Health at SLAM NHS Foundation Trust, Institute of Psychiatry, London, United Kingdom

When comparing two identical objects oriented differently, subjects perform a mental rotation of one object until it is congruent with the other. A recent diffusion tractography study revealed a correlation between right-lateralisation of a fronto-parietal pathway, the second branch of the superior longitudinal fasciculus (SLF II), and performance in certain visuo-spatial tasks. It is unknown whether this pathway is also involved in mental rotation. Using spherical deconvolution tractography, we assessed this correlation in a group of 25 healthy subjects. Our results confirmed lateralisation of this tract is associated with mental rotation performance.

Alia Lemkaddem¹, Alessandro Daducci¹, François Lazeyras², Margitta Seeck³, Jean-Philippe Thiran^1,4, and Serge Vulliemoz^3
1Signal Processing Laboratory (LTS5), Ecole Polytechnique Fédérale de Lausanne (EPFL), Lausanne, Switzerland, ²Dpt of Radiology, University Hospital of Geneva, Switzerland,³EEG and Epilepsy Unit, Neurology clinic, University Hospitals and Faculty of Medicine of Geneva, Switzerland, ⁴Dpt of Radiology, University Hospital Center (CHUV) and University of Lausanne (UNIL), Switzerland

In this study we investigated the effect of right-sided and left-sided temporal lobe epilepsy (TLE) on the global characteristics of brain connectivity estimated by topological measures. We used DSI to construct a connectivity matrix where the nodes represents the anatomical ROIs and the edges are the connections between any pair of ROIs. A significant difference was found between the patient group vs control group in characteristic path length( RTLE: p=0.007, LTLE: p=0.000493), clustering coefficient (p=0.0079, p=0.00025), strength (p=0.0088, p=0.00081) and efficiency (p=0.0099, p=0.00042). This suggests that the TLEs the network is less efficient compared to the network of the control group.

Peter J Lally¹, Hui Zhang², Shreela S Pauliah¹, David L Price³, Alan Bainbridge³, Ernest B Cady³, Seetha Shankaran⁴, and Sudhin Thayyil^1
1Perinatal Neurology and Neonatology, Imperial College London, London, United Kingdom, ²Centre for Medical Image Computing, University College London, London, United Kingdom, ³Medical Physics and Biongineering, University College London Hospitals NHS Trust, London, United Kingdom, ⁴School of Medicine, Wayne State University, Detroit, Michigan, United States

Descriptors of white matter (WM) diffusivity correlate with adverse neurological outcome in neonatal encephalopathy (NE). WM integrity measures derived from clinical diffusion tensor imaging are difficult to interpret in terms of microstructural morphology. We aimed to examine changes in WM microstructure associated with NE, and relate these to tangible biophysical models by fitting single-shell DTI data to the neurite orientation dispersion and density imaging (NODDI) model. Single subject maps of NODDI indices were noisy, but cohort-averaged maps (n=31) enabled characteristic changes in WM fractional anisotropy and radial diffusivity to be related to a possible reduction in neurite density.

K.A. Meijer¹, M. Cercignani², N. Muhlert¹, V. Sethi¹, D. Chard¹, M. Ron¹, A.J. Thompson¹, D.H. Miller¹, J.J.G. Geurts³, and O. Ciccarelli^1
1NMR Unit, Queen Square MS Centre, UCL Institute of Neurology, London, United Kingdom, ²Clinical Imaging Sciences Centre, Brighton and Sussex Medical School, Brighton, United Kingdom, ³Department of Anatomy and Neuroscience, VU University Medical Centre, Amsterdam, Netherlands

We applied independent component analysis (ICA) to patients with multiple sclerosis (MS) in order to examine the specific patterns of correlation in fractional anisotropy (FA) across white matter (WM) tracts (obtained with tract-based spatial statistics). Fifteen out of eighteen components were associated with specific WM tracts and grouped in 5 classes, as previously demonstrated in healthy controls. A cluster of WM tracts, including the supratentorial commissural tracts, showed significant correlations with cognitive performance. These results suggest that microstructurally correlated WM regions are seen in MS patients and they may be important to understand the mechanisms of cognitive dysfunction.

Yasuhiko Tachibana^1,2, Takayuki Obata², Mariko Yoshida¹, Masaaki Hori¹, Michimasa Suzuki¹, Koji Kamagata¹, Issei Fukunaga¹, Kouhei Kamiya¹, Kazumasa Yokoyama³, Nobutaka Hattori³, Tomio Inoue⁴, and Shigeki Aoki^1
1Department of Radiology, Juntendo University School of Medicine, Tokyo, Tokyo, Japan, ²National Institute of Radiological Science, Chiba, Chiba, Japan, ³Department of Neurology, Juntendo University School of Medicine, Tokyo, Japan, ⁴Yokohama City University Graduate School of Medicine, Kanagawa, Japan

Alterations in water diffusion are found in normal appearing white matter (NAWM) of multiple sclerosis (MS) patients. To detect more minor changes in this condition, a new simple method was designed to assess water diffusion using the two-compartment model combined with diffusion tensor imaging. The method allowed greater detectability of minimal changes in NAWM of MS (apparent diffusion coefficient of slow diffusion directed perpendicular to the fibers was higher with significance in MS) compared to conventional mono-exponential fit ADC.

Yuchao Li¹, Guangbin Wang², and Qiang Liu^2
1Shandong Medical Imaging Research Institute, Jinan, Shandong, China, ²Shandong Medical Imaging Research Institute, Jinan, China

the diagnostic performance of monoexponential model and the biexponential model using IVIM MR imaging for in brain.

Alexander D Cohen¹, Kimberly R Pechman², Mona Al-Gizawiy³, Christopher R Chitamber⁴, and Kathleen M Schmainda^1,3
1Biophysics, Medical College of Wisconsin, Milwaukee, WI, United States, ²Neurosurgery, Medical College of Wisconsin, Milwaukee, WI, United States, ³Radiology, Medical College of Wisconsin, Milwaukee, WI, United States, ⁴Medicine, Medical College of Wisconsin, Milwaukee, WI, United States

The apparent diffusion coefficient (ADC) depends on which b-values are used in its calculation. In this study, ADC is calculated using different combinations of b-values and compared between tumor and gray matter (GM) in a rat brain tumor model. Higher ADC was found in tumor vs. GM when lower b-values were used in its calculation, despite increased cellularity in the tumor seen on histology. Lower ADC was found in tumor vs. GM when higher b-values were used in its calculation. ADC calculated with high b-values may be more sensitive to cellularity as the faster diffusing extracellular components have been suppressed.

Samuel Groeschel^1,2, Thomas Schultz^3,4, Gisela Hagberg^2,5, Uwe Klose⁶, Till-Karsten Hauser⁶, Oliver Bieri⁷, Thomas Prasloski⁸, Alex MacKay⁸, Ingeborg Krägeloh-Mann¹, and Klaus Scheffler^2,5
1University Children's Hospital, Tuebingen, Germany, ²MPI for Biological Cybernetics, Tuebingen, Germany, ³Institute of Computer Science, University of Bonn, Germany,⁴MPI for Instelligent Systems, Tuebingen, Germany, ⁵Biomedical Magnetic Resonance, University Hospital, Tuebingen, Germany, ⁶Department of Diagnostic and Interventional Neuroradiology, University Hospital, Tuebingen, Germany, ⁷Radiological Physics, University of Basel, Switzerland, ⁸University of British Columbia, Vancouver, BC, Canada

In our study we assessed several white matter (WM) imaging protocols in their ability to detect the known differences in myelin architecture between the cortico-spinal tract and frontal WM regions in 18 healthy adolescents. We found that diffusion-weighted imaging parameters and myelin water fraction can give quantitative and most sensitive information about differences in myelin microstructure. From the diffusion parameters, neurite density (NODDI) was found to be more sensitive than fractional anisotropy (FA), underlining the limitation of FA in WM crossing fibre regions. Furthermore, these parameters successfully quantified loss of myelin in 5 patients with Metachromatic Leukodystrophy.

Darya Morozov¹, Debbie Anaby¹, and Yoram Cohen^1,2
1School of Chemistry, The Raymond and Beverly Sackler Faculty of Exact Science, Tel-Aviv University, Tel-Aviv, Ramat-Aviv, Israel, ²Sagol School of Neurosciences, Tel-Aviv University, Tel-Aviv, Ramat-Aviv, Israel

Diffusion weighted MRI and DTI, have been extensively used to study pathologies and to perform non-invasive in vivo tractography of the CNS. However, the ADC, FA and MD extracted from DWI and DTI experiments have limited sensitivity and specificity to structural changes. Therefore there is a constant need to develop new potential methods to provide more specific microstructural indices. In the present work we study the effect of global hypoxia on the indices extracted from angular d-PFG MRI at a long mixing time and compare them with the changes observed in the FA and MD obtained from DTI.

Inge Timmers^1,2, Alard Roebroeck¹, Matteo Bastiani^1,3, Bernadette Jansma¹, Estela Rubio-Gozalbo², and Hui Zhang^4
1Department of Cognitive Neuroscience, Maastricht University, Maastricht, Netherlands, ²Department of Pediatrics, Maastricht University Medical Center, Maastricht, Netherlands, ³Institute of Neuroscience and Medicine, Forschungszentrum Juelich GmbH, Juelich, Germany, ⁴Department of Computer Science & Centre for Medical Image Computing, University College London, London, United Kingdom

This work demonstrates neurite orientation dispersion and density imaging (NODDI) as a tool for investigating microstructural substrates of white matter abnormalities in a clinical population. NODDI estimates indices of neurite density (NDI) and orientation dispersion (ODI), two key contributors of fractional anisotropy (FA) from diffusion tensor imaging (DTI). We show that NODDI analysis of multi-shell diffusion-weighted data adds microstructural specificity to the findings from DTI. Findings from the NODDI indices overlap substantially with FA, revealing the specific microstructural changes corresponding to the latter. We additionally demonstrate that NODDI analysis of single-shell data can potentially be used for detecting ODI changes.

Ofer Pasternak¹, Robert A. Stern², Michelle Y. Giwerc¹, Charles Yergatian¹, Sai Merugumala¹, Huijun Liao¹, Christine M. Baugh², Carl-Fredrik Westin¹, Martha E. Shenton^1,3, and Alexander P. Lin^1
1Harvard Medical School, Boston, Massachusetts, United States, ²Boston University, Boston, MA, United States, ³VA Boston Healthcare System, Brockton, MA, United States

Recent diffusion models include a free-water compartment accounting for extracellular water molecules not bounded by tissue membranes. In order to reveal the relation between free-water volume and underlying pathologies we correlate the measure with metabolite counts that were obtained in MR spectroscopy. We tested 49 retired National Football League (NFL) players at high risk for having chronic traumatic encephalopathy, and 14 controls. We find free-water differences between the groups, and that free-water mainly correlates with NAA, suggesting sensitivity to atrophy. By covarying for NAA we find that free-water correlates with other metabolites that suggest sensitivity to neuroinflammation and gliosis.

Yohan van de Looij^1,2, Nicolas Kunz², Rajika Maddage², Rolf Gruetter^2,3, Petra S Hüppi¹, and Stéphane V Sizonenko^1
1Division of Child Growth and Development, University of Geneva, Geneva, GE, Switzerland, ²Laboratory for Functional and Metabolic Imaging, Ecole Polytechnique Fédérale de Lausanne, Lausanne, VD, Switzerland, ³Department of Radiology, University of Geneva and Lausanne, Geneva and Lausanne, GE and VD, Switzerland

Diffusion Tensor Imaging (DTI) and Quantitative Susceptibility Mapping (QSM) are two MR techniques given a superb contrast between white and grey matter in the brain. Diffusivity and anisotropy of water in the tissues are derived from DTI whereas QSM is based on the magnetic susceptibility distribution. The aim of this work was to investigate the potential cross correlations between DTI and QSM in the rat brain white matter at ultra-high magnetic field. In addition to the effect of magnetic field orientation and myelin, we suggest an effect of axonal diameter/compaction to the QSM contrast.

Ping-Huei Tsai^1,2, Ming-Chung Chou³, Cheng-Yu Chen^1,2, Shih-Wei Chiang^4,5, Chao-Ying Wang⁵, Hsiao-Wen Chung⁴, Hung-Wen Kao⁵, and Yi-Hsiu Hsiao^1
1Imaging Research Center, Taipei Medical University, Taipei, Taiwan, ²Department of Medical Imaging, Taipei Medical University Hospital, Taipei Medical University, Taipei, Taiwan, ³Department of Medical Imaging and Radiological Sciences, Kaohsiung Medical University, Kaohsiung, Taiwan, ⁴Graduate Institute of Biomedical Electronics and Bioinformatics, National Taiwan University, Taipei, Taiwan, ⁵Department of Radiology, Tri-Service General Hospital, National Defense Medical Center, Taipei, Taiwan

Acute CO intoxication is one of the most common lethal poisons worldwide, and could lead to brain injury resulting from leukoencephalopathy. Although diffusion tenor (DT) related parametric maps could help assess the brain structural changes resulting from demyelination, it may not be highly sensitive to observe the subtle alternations in patients with acute CO intoxication. The purpose of this study is to validate evaluate the ability of diffusion kurtosis imaging (DKI) for early detection of the brain WM microstructural changes. The present finding indicated that the use of DKI method may provide some valuable information in prediction of leukoencephalopathy in these patients.

Chia-Wen Chiang¹, Tsen-Hsuan Lin², Joong Hee Kim¹, and Sheng-Kwei Song^1,3
1Radiology, Washington University School of Medicine, St. Louis, MO, United States, ²Physics, Washington University, St. Louis, MO, United States, ³The Hope Center for Neurological Disorders, Washington University School of Medicine, St. Louis, MO, United States

In this study, in vivo diffusion basis spectrum imaging (DBSI) and manganese-enhanced MRI (MEMRI) were performed to assess axonal pathologies (inflammation, demyelination and axonal injury) and to determine the degree of axonal transport deficit in 12-month old DBA/2J mice, a rodent model of glaucoma which develops progressive degeneration of visual function mimicking human glaucoma. Our results suggest that DBA/2J mice developed inflammation, axonal injury, demyelination with significant axonal transport disruption in optic nerves compared with the age-matched controls.

Magdalena Zurek¹, Kerryanne Winters¹, Jin Zhang¹, Joe Rodriguez¹, Dmitry S. Novikov¹, Sungheon Kim¹, and Els Fieremans^1
1Department of Radiology, New York University School of Medicine, Bernard and Irene Schwartz Center for Biomedical Imaging, New York, New York, United States

We assessed diffusion white matter tract integrity (WMTI) metrics, in particular tortuosity and axonal water fraction (AWF), in an experimental model of cuprizone-induced WM degeneration, against conventional MRI myelin indicators and histopathology. DTI, WMTI metrics, and traditional MR myelin estimators such as T2 and MTR were sensitive to the effect of the cuprizone treatment, however the overall comparison between new and conventional parameters showed that tortuosity, and AWF were the most sensitive measures for cuprizone-induced changes. Our primarily results indicate that WMTI imaging markers can be valuable in assessing WM degeneration.

Lawrence Kenning¹, Martin Lowry², and Lindsay W Turnbull^2,3
1Centre for Magnetic Resonance Investigations, University of Hull, Hull, East Riding of Yorkshire, United Kingdom, ²Centre for Magnetic Resonance Investigations, Hull York Medical School at the University of Hull, Hull, East Riding of Yorkshire, United Kingdom, ³Hull and East Yorkshire Hospitals NHS Trust, East Riding of Yorkshire, United Kingdom

DTI can visualise fiber tracts for surgery, however, oedema and tumour infiltration cause FA to decrease. We investigated the anisotropic component of diffusion to visualise fiber tracts in the presence of oedema. Glioma patients were scanned twice with DTI. ADC, FA and q maps were generated. Percentagewise changes of q following altered levels of oedema were reduced when compared to those of FA. Increased values of ADC are destructive to FA values given the normalisation process which q does not include. Anisotropic component of diffusion maps continue to show fiber tract directionality in the presence of large amounts of oedema.

Catarina Freitas¹, Varun Sethi¹, Nils Muhlert¹, Olga Ciccarelli², Mara Cercignani³, Declan Chard², Hui Zhang⁴, and Claudia Wheeler-Kingshott^1
1Department of Neuroinflammation, UCL Institute of Neurology, London, United Kingdom, ²Department of Brain Repair and Rehabilitation, UCL Institute of Neurology, London, United Kingdom, ³Department of Neuroscience, University of Sussex, Brighton, United Kingdom, ⁴Department of Computer Science & Centre for Medical Image Computing, University College London, London, United Kingdom

Misinterpretation of differences in the diffusion tensor (DT) indices between patients and healthy controls (HC) may occur when the geometrical properties of each dataset are not taken into account. Here, we tested a new analysis method that has been suggested to solve this problem, in a group comparison of HCs and patients with multiple sclerosis (MS). In addition, we investigated the effect of registration by using DT-based and fractional anisotropy (FA)-based methods. The analysis showed the new approach may help to reveal WM subtle changes and the importance of determining which registration method is more appropriate to study WM pathology.

Sila Genc¹, Christopher Steward^1,2, Terence O'Brien^3,4, and Patricia Desmond^1,2
1Department of Radiology, University of Melbourne, Parkville, Victoria, Australia, ²Department of Radiology, Royal Melbourne Hospital, Parkville, Victoria, Australia,³Department of Medicine, University of Melbourne, Parkville, Victoria, Australia, ⁴Department of Neurology, Royal Melbourne Hospital, Parkville, Victoria, Australia

Longitudinal changes in fractional anisotropy in Alzheimer’s disease patients were investigated using three different registration methods. Recent studies have suggested alterations to the TBSS pipeline in order to improve the registration process for subsequent image analysis. Replacing the standard registrations in the TBSS pipeline in FSL, with ANTs registrations, revealed minimal differences between the spatial distributions of FA change in the core white matter tracts within six months. The consistencies between the standard TBSS, ANTS-modified TBSS, and study-specific template results suggests that running the simple TBSS pipeline may be sufficient for longitudinal analyses in complex pathologies such as Alzheimer’s disease.

Qiaowen Yu^1,2, Tina Jeon¹, Austin Ouyang¹, Virendra Mishra¹, Steven Hsiao³, Shuwei Liu², and Hao Huang^1
1Advanced Imaging Research Center, University of Texas Southwestern Medical Center, Dallas, Texas, United States, ²Research Center for Sectional and Imaging Anatomy, Shandong University School of Medicine, Jinan, Shandong, China, ³Mind and Brain Institute, Johns Hopkins University, Baltimore, Maryland, United States

A comprehensive characterization of white matter (WM) anatomy of macaque brain with DTI can not only provide neuroanatomical atlases for neuroscientific studies using macaque model, but also aid the study of the WM evolution. WM tracts underlie the connectivity of brain regions which play essential roles for understanding brain functions and evolution of brain functions. In this study, we aimed to generate a probabilistic and comprehensive digital macaque brain atlas including labeling of WM tracts with ultra-high resolution DTI data (0.35x0.4x0.45mm3) of a population of macaque brains.

Geert De Groof¹, Sofie Van Massenhoven¹, Elisabeth Jonckers¹, and Annemie Van der Linden^1
1Bio-Imaging Lab, University of Antwerp, Antwerp, Antwerp, Belgium

Giovanni Morana¹, Federica De Leo¹, Valentina Tavano², Andrea Mazzaro², Mirco Ros³, Francesca Lucca³, Pierluigi Ciet⁴, and Silvia Bertolo^1
1Radiology, Ca' Foncello Hospital, Treviso, Treviso, Italy, ²Padova's Hospital, Padova, Padova, Italy, ³Pediatrics, Ca' Foncello Hospital, Treviso, Treviso, Italy, ⁴Radiology and Pediatrics Pulmonology, Erasmus MC, Rotterdam, Rotterdam, Netherlands

Cystic Fibrosis is the most common lethal hereditary disease in the caucasian population. Currently, no sensitive, radiation-free methods are available to localize and quantify lung inflammation. Developments in MRI have made possible the clinical application of lung-MRI to obtain not only morphological but also functional information. Our propose is giving an overview of these new MR tecniques and their potential application in CF population. Lung-MRI has the potential to supply new relevant functional information in thoracic imaging. Its impact in CF follow-up has still to be defined, but it might open new therapeutic scenarios in CF and in other lung disease.

Jos Oudeman¹, Gustav J Strijkers², Mario Maas¹, Aart J Nederveen¹, P Luijten³, and Martijn Froeling^1,3
1Radiology, Academic Medical Center, Amsterdam, Noord Holland, Netherlands, ²Biomedical NMR, Department of biomedical engineering, Eindhoven University of Technology, Eindhoven, Brabant, Netherlands, ³Radiology, University Medical Center, Utrecht, Utrecht, Netherlands

In Diffusion Tensor Imaging (DTI) studies of skeletal muscles, segmentation is performed manually, which is tedious and difficult in complex structures. Furthermore, muscle fiber tractography often results in fibers that continue along tendon or aponeurosis due to partial volume effects. This effect leads to overestimation of muscle fiber lengths and making it difficult to identify tendon insertion points and calculate pennation angles.Due to the partial volume effects reconstructed fibers tend to curve along the tendons. Therefore we hypothesized tract density differences can be used for automatic muscle segmentation to distinguish muscle from tendon.

Christina Schraml¹, Susanne Will², Nina F Schwenzer¹, Guenter Steidle², Claus D Claussen¹, Fritz Schick², and Petros Martirosian^2
1Department of Radiology, Diagnostic and Interventional Radiology, Tuebingen, BW, Germany, ²Department of Radiology, Section on Experimental Radiology, Tuebingen, BW, Germany

Aim of the study was to assess application of inner-volume-imaging (IVI) technique for diffusion-weighted imaging (DWI) with DTI and IVIM of the kidneys in healthy volunteers. Qualitative and quantitative comparison between IVI and full-field-of-view (FF) imaging was performed which revealed reduced distortion, less signal voids and better image quality of calculated parametric maps in IVI technique. FA values were comparable between IVI and FF technique while slight differences regarding IVIM parameters were observed. IVI technique is a promising approach for improving DWI of the kidneys. The application in patients with renal diseases is necessary to assess its clinical performance.

Melissa Hooijmans¹, Ece Ercan¹, Robert Brandt¹, Andrew Webb¹, Hermien Kan¹, and Itamar Ronen^1
1Radiology, Leiden University Medical Center, Leiden, Zuid-holland, Netherlands

Conventional diffusion weighted imaging is limited by the lack of compartment specificity of water. In this work, we show the application of 31P diffusion weighted spectroscopy of individual muscles in the lower leg at 7T. SNR was sufficient to determine apparent diffusion coefficients of phosphocreatine (PCr) in two different muscles of the leg. In conclusion we are able to assess the diffusion of PCr in the individual lower leg muscles and these values seem to be affected more by cell size than fiber type distribution.

Kerryanne Winters¹, Dmitry S. Novikov¹, Els Fieremans¹, and Sungheon Kim^1
1Radiology, NYU School of Medicine, New York, NY, United States

We quantified the microstructural changes in normal skeletal muscle growth using time-dependent DTI combined with random permeable barrier model (RPBM). A DTI study with multiple diffusion times was conducted with 4-5 weeks and 10-12 weeks old mice, followed by immunohistochemistry staining of collagen-IV and aquaporin-4. Overall, young mice had lower unrestricted diffusivity, higher membrane surface-to-volume (S/V) ratio, and lower membrane permeability than the old mice. Our preliminary results demonstrate that the DTI-RPBM measures can be used to monitor muscle growth in wild type mice, and substantiates the potential of using the DTI-RPBM measures as quantitative biomarkers of myopathy.

Wenchao Cai¹, Feiyu Li¹, Jintang Ye¹, Queenie Chan², Xiaoying Wang¹, and Xuexiang Jiang^1
1Peking University First Hospital, Beijing, Beijing, China, ²Philips Healthcare, Hong Kong, China

Intravoxel incoherent motion (IVIM) MR imaging is a non-invasive method with the ability of separation of ¡°pure¡± molecular diffusion and perfusion effects. Radiologists perform diffusion-weighted and dynamic contrast-enhanced (DCE) MRI to provide additional diffusion and perfusion information in the routine clinical set, but it needs intravenous contrast agent administration and requires cumbersome procedures. Therefore, in this study we applied the IVIM technique to detect the prostate cancer and to compare the diagnostic performance between IVIM and conventional DWI combined with DCE. Our results demonstrated combination of D and f seemed to be more efficient than the clinical conventional DWI combined with DCE method in the diagnosis of prostate cancer