Non-Gaussian Diffusion

Friday 16 May 2014

Marco Palombo^1,2, Valentina Di Marco^1,2, Giulia Di Pietro^1,2, and Silvia Capuani^1,2
1Physics Department, Sapienza University, Rome, Rome, Italy, ²CNR IPCF UOS Roma Sapienza, Physics Department Sapienza University, Rome, Rome, Italy

We studied the potential of Gaussian and non-Gaussian diffusion methods to obtain microstructural information and water compartmentalization in bone-marrow filling pores in cancellous bone (TBM) and not forced in pores (FBM), by investigating water and fat ADC, and  stretched parameter of the anomalous diffusion model, as a function of diffusion time. We used four TBM and seven FBM samples extracted from calves, to show the ability of Gaussian and non-Gaussian diffusion methods to characterize and discriminate different bone-marrow samples and to highlight that  is highly correlated to intrinsic local microstructural features at the interface between water and bone.

Ben Jeurissen¹, Jacques-Donald Tournier^2,3, Thijs Dhollander⁴, Alan Connelly^2,5, and Jan Sijbers^1
1iMinds-Vision Lab, University of Antwerp, Antwerp, Belgium, ²The Florey Institute of Neuroscience and Mental Health, Melbourne, Victoria, Australia,³Division of Imaging Sciences & Biomedical Engineering, King's College London, London, United Kingdom, ⁴Medical Imaging Research Center, KU Leuven, Belgium, ⁵The Florey Department of Neuroscience, University of Melbourne, Victoria, Australia

Constrained spherical deconvolution (CSD) has become one of the most widely used methods to estimate the white matter fibre orientation distribution function (fODF). However, CSD typically only supports data acquired on a single shell in q-space. In addition, CSD provides unreliable fODF estimates in voxels containing other tissue types than pure white matter. We propose a multi-shell CSD approach that exploits the multi-shell behaviour of different tissue types to estimate a multi-tissue ODF. Our method produces tissue volume fraction maps straight from the diffusion weighted data and provides significantly improved white matter fODF estimates at the tissue interfaces.

Austin Ouyang¹, Xinzeng Wang¹, Mihovil Pletikos², Nenad Sestan², and Hao Huang^1
1Advanced Imaging Research Center, University of Texas Southwestern Medical Center, Dallas, Texas, United States, ²Department of Neurobiology, Yale University, New Haven, CT, United States

Diffusion kurtosis imaging (DKI) provides greater sensitivity to microstructural differences due to its capability of quantifying non-Gaussian diffusion properties; however, the biological interpretation of these DKI metrics has yet to be fully characterized, especially in the cerebral cortex. In this study, we acquired high resolution DKI data of a macaque brain and compared the DKI metrics with histological image of neurofilament staining. Slices with large cortical variations in mean kurtosis (MK) corresponded well with the density of neurofilament staining.

Chantal M.W. Tax¹, Tom C.J. Dela Haije², Andrea Fuster², Remco Duits², Max A. Viergever¹, Luc M.J. Florack², and Alexander Leemans^1
1Image Sciences Institute, University Medical Center Utrecht, Utrecht, Utrecht, Netherlands, ²Imaging Science & Technology, Eindhoven University of Technology, Eindhoven, Noord-Brabant, Netherlands

Recently, a debate in literature discussed the possibility of the cerebral pathways being organized in a grid structure of interwoven sheets. To prove or disprove this statement, a more quantitative evaluation is needed. In this work, we evaluate the capability of the discrete Lie bracket proposed in Wedeen (2012a) to quantify sheet structure as a function of noise and voxel size. On simulated vector fields, we show that this Lie bracket is indeed capable of distinguishing sheet structured from non-sheet structured vector fields within certain limits for noise and voxel size.

Sjoerd B. Vos¹, Chantal M.W. Tax¹, Martijn Froeling², and Alexander Leemans^1
1Image Sciences Institute, University Medical Center Utrecht, Utrecht, Utrecht, Netherlands, ²Department of Radiology, University Medical Center Utrecht, Utrecht, Utrecht, Netherlands

Recently, acquisition and processing parameters have been proposed to optimally describe the angular profile of diffusion-weighted signal profiles in single-fiber voxels. We present an overview of how these factors affect fODF estimation in crossing-fiber voxels in vivo, investigating both the accuracy and precision of reconstructed fODF peaks. Higher b-values provide high levels of precision and accuracy, even for low number of DWIs (50-100). On the whole, more directions at lower b-values performed to similar levels of precision. However, the longer diffusion-weighting pulses for higher b-values likely cost less scan time than the increased number of DWIs required at lower b-values.

Steven Baete^1,2 and Fernando Emilio Boada^1,2
1Center for Biomedical Imaging, Dept. of Radiology, NYU Langone Medical Center, New York, New York, United States, ²CAI2R, Center for Advanced Imaging Innovation and Research, NYU Langone Medical Center, New York, New York, United States

Diffusion Spectrum Imaging (DSI) would benefit from longer diffusion times as these lead to increased anisotropy due to the restricted/hindered diffusion orthogonal to the fiber bundles. Since longer diffusion times are prohibitive in conventional spin echo sequences due to the concomitant increase in SNR, we investigated the use of a stimulated echo sequence which counters the SNR loss. When comparing DSI datasets acquired with spin echo and stimulated echo diffusion sequences in vivo in a healthy volunteer on a clinical 3T scanner, an increase in anisotropy is noticeable at longer diffusion times. This higher anisotropy might improve fiber tracking results.

Aapo Nummenmaa¹, Thomas Witzel¹, Ville Renvall^1,2, Bruce R Rosen¹, Van J Wedeen¹, and Lawrence L Wald^1
1Athinoula A. Martinos Center for Biomedical Imaging, Department of Radiology, Massachusetts General Hospital, Harvard Medical School, Boston, Massachusetts, United States, ²Brain Research Unit, O.V. Lounasmaa Laboratory, Aalto University, Espoo, Finland

We study the effect of using high diffusion weighting (b-value) and high angular resolution diffusion imaging in resolving multiple intravoxel crossing fiber populations. Three q-ball diffusion-sampling scenarios are considered: b-value of 5000 s/mm² with 128 diffusion sampling gradient directions and b-value of 10000 s/mm² with 128 and 256 directions. The spherical harmonics expansion method is used to reconstruct the diffusion orientation distribution functions (ODFs). The results show that increasing b-value and angular resolution increases the sharpness of the ODF as well as the amplitude of the subdominant fiber population, indicating a more accurate delineation of the intravoxel fiber crossing.

Nathaniel D Kelm^1,2, Kathryn L West^1,2, Daniel F Gochberg^2,3, Robert P Carson^4,5, Kevin C Ess^4,5, and Mark D Does^1,2
1Biomedical Engineering, Vanderbilt University, Nashville, TN, United States, ²Institute of Imaging Science, Vanderbilt University, Nashville, TN, United States, ³Radiology and Radiological Sciences, Vanderbilt University, Nashville, TN, United States, ⁴Neurology, Vanderbilt University, Nashville, TN, United States, ⁵Pediatrics, Vanderbilt University, Nashville, TN, United States

Diffusion kurtosis imaging (DKI) is an extension of DTI with the potential of providing additional information about white matter microstructure and its state of myelination. DKI measures are compared with myelin-related measures, myelin water fraction (MWF) and macromolecular pool size ratio (PSR), in order to assess the utility of DKI in the characterization of myelin. DKI parameters mean kurtosis (MK) and radial kurtosis (RK) showed stronger correlations with MWF and PSR compared to conventional DTI parameters, indicating the potential of DKI for assessment of myelination.

Phillip Zhe Sun¹, Yu Wang¹, Emiri Mandeville², Mark Vangel¹, Eng Lo², and Xunming Ji^3
1Radiology, Martinos Center for Biomedical Imaging, Charlestown, MA, United States, ²Neuroprotection Research Laboratory, Department of Radiology and Neurology, Massachusetts General Hospital, Harvard Medical School, MA, United States, ³Cerebrovascular Diseases Research Institute, Xuanwu Hospital of Capital Medical University, Beijing, China

Recent studies have shown that mean kurtosis (MK) can stratify irreversibly damaged ischemic DWI lesion. However, the kurtosis map is heterogeneous due to complex cerebral structure and composition, reducing its specificity to ischemia. We evaluated the correlation between MK, diffusivity (MD), fractional anisotropy (FA) and relaxation, and found significant correlation between MK and R1 (P<0.001). Using an animal stroke model, we demonstrated that relaxation-normalized kurtosis MRI significantly enhanced tissue segmentation of ischemic kurtosis lesion over the standard MK map. We also found significant diffusion/kurtosis lesion mismatch, with MK and MD lesion volumes measuring 172±78 and 206±93 mm3, respectively.