Diffusion Acquisition

Monday 22 April 2013

Stamatios N. Sotiropoulos¹, Saad Jbabdi¹, Junqian Xu², Jesper L. Andersson¹, Steen Moeller², Edward J. Auerbach², Matthew F. Glasser³, David Feinberg⁴, Christophe Lenglet², David C. Van Essen³, Kamil Ugurbil², Timothy E.J. Behrens¹, and Essa S. Yacoub^2
1FMRIB Centre, University of Oxford, Oxford, United Kingdom, ²Center for Magnetic Resonance Research, University of Minnesota, Minneapolis, MN, United States, ³Department of Anatomy & Neurobiology, Washington University, St Louis, MO, United States, ⁴Advanced MRI Technologies, Sebastopol, CA, United States

The Human Connectome Project (HCP) is an ambitious 5-year effort to map human brain connections in healthy adults. A consortium of HCP investigators will study a population of 1200 subjects using multiple imaging modalities along with extensive behavioral and genetic data. In this overview, we focus on diffusion-weighted MRI and the structural connectivity aspect of the project. We present recent advances in acquisition and preprocessing that allow us to obtain the best possible MR data quality in-vivo, while confronting with the aim of scanning many subjects. The data quality described is representative of the datasets to be released within 2013.

Cornelius Eichner^1,2, Kawin Setsompop^2,3, Peter J. Koopmans⁴, Alfred Anwander¹, Ralf Lützkendorf⁵, Steven F. Cauley⁶, Himanshu Bhat⁷, David G. Norris⁴, Robert Turner¹, Lawrence L. Wald^2,3, and Robin Martin Heidemann^1,8
1Max Planck Institute for Human Cognitive and Brain Sciences, Leipzig, Germany, ²Athinoula A. Martinos Center for Biomedical Imaging, Charlestown, MA, United States, ³Harvard Medical School, Boston, MA, United States, ⁴Donders Institute for Brain, Cognition and Behaviour, Nijmegen, Netherlands, ⁵Otto v. Guericke University, Magdeburg, Germany,⁶Massachusetts General Hospital, Charlestown, MA, United States, ⁷Siemens Medical Solutions, Malvern, PA, United States, ⁸Siemens Healthcare Sector, Erlangen, Germany

The ZOOPPA technique provides highly resolved dMRI data at 7T, but whole-brain data sets take a long time to acquire. EPI volume acquisition time can be greatly reduced using blipped CAIPIRINHA. We have implemented combined ZOOPPA and CAIPIRINHA, enabling highly resolved and accelerated acquisition of dMRI data at 7T.

Robert Frost¹, Karla L. Miller¹, David A. Porter², Rob H. N. Tijssen³, and Peter Jezzard^1
1FMRIB Centre, Nuffield Department of Clinical Neurosciences, University of Oxford, Oxford, United Kingdom, ²Healthcare Sector, Siemens AG, Erlangen, Germany, ³Department of Radiotherapy, UMC Utrecht, Utrecht, Netherlands

We present a 3D multi-slab readout-segmented echo-planar imaging (rs-EPI) sequence for diffusion imaging that minimizes motion-induced phase artefacts with 2D navigator correction and real-time reordering of k-space with respect to the cardiac cycle. Options for acquisition schemes and motion-induced phase corrections were simulated and the chosen strategy was implemented by modifying a standard 2D rs-EPI sequence. Real-time cardiac reordering is validated in simulations and experiment, demonstrating 40-50% reduced variability in a time series of diffusion images compared to a sequential k-space acquisition. The 3D multi-slab rs-EPI method is demonstrated in-vivo and is shown to provide excellent image fidelity.

Chitresh Bhushan¹, Anand A. Joshi², Richard M. Leahy², and Justin P. Haldar^2
1University of Southern California, Los Angeles, CA, United States, ²Signal and Image Processing Institute, University of Southern California, Los Angeles, CA, United States

EPI-based diffusion MRI suffers from localized distortion artifacts in the presence of B0 inhomogeneity, which can cause problems in multi-modal image analysis and when estimating quantitative diffusion parameters. These distortions can be partially corrected with measured field maps, though performance improves substantially if each image is acquired twice with reversed phase encoding gradients (at the expense of doubling the scan time). In this work, we propose a novel acquisition and reconstruction strategy that leverages a constrained reconstruction formulation to enable accurate distortion correction with similar performance to the reversed gradient method, but without increasing the scan time.

Nan-kuei Chen¹, Arnaud Guidon¹, Hing-Chiu Chang¹, and Allen W. Song^1
1Brain Imaging and Analysis Center, Duke University, Durham, NC, United States

DWI data have been mostly acquired with single-shot EPI with limited spatial resolution. Multi-shot EPI could potentially achieve higher spatial resolution and fidelity, but is susceptible to aliasing artifacts due to phase inconsistencies among excitations. Although the shot-to-shot phase variations may be corrected with navigator echoes, the residual artifacts may be pronounced when there exist local and nonlinear motions. To address these challenges, a novel multi-shot DWI technique is developed here to reliably and inherently correct nonlinear shot-to-shot phase variations without navigator echoes. This technique enables very high-resolution DWI mapping of human white matter architecture.

Suchandrima Banerjee¹, Emine U. Saritas², Gerd Melkus³, and Ajit Shankaranarayanan^1
1Applied Science Lab, GE Healthcare, Menlo Park, CA, United States, ²Bioengineering, University of California Berkeley, Berkeley, CA, United States, ³Radiology and Biomedical Imaging, Univerisity of California San Francisco, San Francisco, CA, United States

2D spatially selective RF excitation obviates the need for a large field-of-view in the phase-encoding direction to avoid aliasing artifacts and can zoom into a region of interest. This reduces distortion in single-shot echo-planar imaging (ssEPI) by reducing readout time, as recently demonstrated by publications on reduced field-of-view (rFOV) diffusion imaging. However the number of slices can be limited due to presence of side lobes with the 2D RF excitation. In this work we demonstrate a tilted 2D RF design which removes the restriction on number of maximum slice locations, while providing robust fat suppression in diffusion-weighted ssEPI.

Berkin Bilgic¹, Itthi Chatnuntawech¹, Kawin Setsompop^2,3, Stephen F. Cauley⁴, Lawrence L. Wald^2,5, and Elfar Adalsteinsson^5,6
1EECS, Massachusetts Institute of Technology, Cambridge, MA, United States, ²A. A. Martinos Center for Biomedical Imaging, Dept. of Radiology, MGH, Charlestown, MA, United States, ³Harvard Medical School, Boston, MA, United States, ⁴A. A. Martinos Center for Biomedical Imaging, Massachusetts General Hospital, Charlestown, MA, United States,⁵Harvard-MIT Division of Health Sciences and Technology, MIT, Cambridge, MA, ⁶EECS, MIT, Cambridge, MA, United States

Significant benefit in Compressed Sensing (CS) reconstruction of Diffusion Spectrum Imaging (DSI) data from undersampled q-space was demonstrated when a dictionary trained for sparse representation was utilized rather than wavelet and Total Variation (TV). However, computation times of both dictionary-based and Wavelet+TV methods are on the order of days for full-brain processing. We present two algorithms that are 3 orders of magnitude faster than these CS methods with reconstruction quality comparable to the previous dictionary-CS approach.

Amith J. Kamath¹, Iman Aganj^2,3, Junqian Xu⁴, Essa S. Yacoub⁴, Kamil Ugurbil⁴, Guillermo Sapiro⁵, and Christophe Lenglet^4
1Electrical and Computer Engineering, University of Minnesota, Minneapolis, MN, United States, ²Martinos Center for Biomedical Imaging, Massachusetts General Hospital, Harvard Medical School, Cambridge, MA, United States, ³Electrical and Computer Engineering, MIT, Cambridge, MA, United States, ⁴Center for Magnetic Resonance Research, University of Minnesota, Minneapolis, MN, United States, ⁵Electrical and Computer Engineering, Duke University, Durham, NC, United States

This work elaborates guidelines for sampling schemes to be used with a multi-shell diffusion MRI acquisition. The reconstruction is based on the CSA-ODF model and we use the Camino toolkit to generate synthetic data for simulations. By varying the b-value, Spherical Harmonics order, SNR and number of gradient directions, we conclude with the observation that b-values including 1000, 2000, and a third shell in the range [3000, 6000] s/mm2 shows better reconstructions, and that 200 total gradient directions appear sufficient.

Tuukka Miettinen¹, Alejandra Sierra¹, Teemu Laitinen¹, Juhana Sorvari¹, and Olli Gröhn^2
1Department of Neurobiology, A. I. Virtanen Institute, University of Eastern Finland, Kuopio, Kuopio, Finland, ²Department of Neurobiology, University of Eastern Finland, Kuopio, Kuopio, Finland

Parametric maps obtained with double-PFG MR imaging and DTI were compared in five brain regions with different degree of microscopic and microscopic anisotropy as verified by histology. Our data show clear difference between FA and apparent eccentricity (|aE|) map in white matter while the residual phase map from d-PFG data could differentiate between grey matter areas with different degrees of partially oriented anisotropic structures. We conclude that double-PFG can provide novel information about microstructure of tissue and it has high potential to detect structural modifications caused by pathological conditions.

Corey Allan Baron¹ and Christian Beaulieu^1
1Biomedical Engineering, University of Alberta, Edmonton, Alberta, Canada

Apparent diffusion coefficients measured using DTI depend on the time allowed for the molecules to probe the local environment because of the presence of cellular structures. Oscillating gradient spin-echo (OGSE) DTI enables greatly reduced diffusion times, which grants a greater sensitivity to diffusion restriction/hindrance over smaller length scales. In this work, DTI with a b-value of 300 s/mm² and maximum OGSE frequency of 50 Hz is demonstrated in 5 healthy subjects, where significant increases of parallel and perpendicular DTI eigenvalues as well as decreases in FA are observed as the diffusion time is decreased from 40 ms to 5 ms.