|Improving Data Quality for Diffusion Imaging|
|16:30||133.||An Eddy-Current-Compensated Diffusion-Weighting
Preparation Based on a Single Spin Echo
Jürgen Finsterbusch1, 2
1University Medical Center Hamburg-Eppendorf, Hamburg, Germany; 2Neuroimage Nord, Hamburg-Kiel-Lübeck, Germany
A double-spin-echo preparation is commonly used in diffusion-weighted MR to minimize artifacts caused by eddy currents, e.g. geometric distortions in echo-planar imaging. However, due to the two refocusing RF pulses required, it is at the expense of a higher sensitivity to RF inhomogeneieties that are in particular pronounced at higher static magnetic fields. An extension of the Stejskal-Tanner experiment is presented that introduces a third gradient lobe prior to the RF refocusing pulse. As the double-spin-echo preparation, it is able to compensate eddy currents of a given decay constant for appropriate gradient durations but is less sensitive to RF inhomogeneieties.
Single Step Non-Linear Diffusion Tensor Estimation in
the Presence of Microscopic and Macroscopic Motion
Murat Aksoy1, Chunlei Liu1, Michael Moseley1, Roland Bammer1
1Stanford University, Stanford, California , USA
Due to its long acquisition time, DTI can suffer considerably from artifacts introduced by patient motion. Thus far, the correction of motion artifacts in DTI has been focused mostly to correcting for miniscule motion and avoiding ghosting artifacts. However, for the accuracy of the reconstructed diffusion tensors, gross patient motion should also be accounted for. While a simple registration operation is sufficient for single-shot data, for multi-shot sequences, the correction of gross patient motion requires accounting for the altered diffusion encoding direction between successive k-space interleaves. In this study, we investigated the combined effect of microscopic and macroscopic motion on the k-space data and proposed a single step non-linear optimization algorithm that reconstructs the diffusion tensors directly from k-space data in the presence of microscopic and macroscopic motion.
|16:54||135.||Reducing Distortions in DW-EPI with a Dual-Echo
Daniel Gallichan1, Jesper L R Andersson1, Mark Jenkinson1, Matthew D. Robson1, Karla L. Miller1
1University of Oxford, Oxford, UK
DW-EPI is intrinsically affected by distortions due to inhomogeneities in the B0 field. Parallel acceleration reduces these distortions, but does not remove them entirely. We propose the addition of a second spin-echo in the accelerated pulse sequence with the phase encode blips in the opposing direction to the first echo. This allows a blip-up-blip-down reconstruction of the corrected image with a vastly reduced scan time compared to acquiring the images separately.
Crossing Fibers, Diffractions and Non-Homogeneous
Magnetic Field: Correction of Artifacts by Bipolar Gradient Pulses
Amnon Bar-Shir1, Yoram Cohen1
1Tel-Aviv University, Tel-Aviv, Israel
Recently, several approaches were suggested for increasing the accuracy of determining fiber orientations, especially those of crossing-fibers. Therefore there is a need for complex phantoms to challenge different indices extracted from the different diffusion MR techniques used. In the present study we describe several micron-scale phantoms of increasing complexity and decreased homogeneity. Our results demonstrate the effect of background gradients on the expected diffraction patterns observed in high b-value q-space NMR. We showed that bipolar diffusion gradients can reduced such artifacts suggesting the use of bipolar diffusion gradients in clinical diffusion MRI.
Partial-Volume Modelling in Diffusion MRI
Stephen Smith1, Jennifer McNab1, Karla Miller1
1FMRIB, Dept. Clinical Neurology, UK
In diffusion-based analyses of white-matter integrity (e.g., using fractional anisotropy (FA) as the primary marker), it is of concernthat many tracts of interest have thickness on the same order as thedata resolution; an apparent change in FA could be due to either achange in tract thickness (extent of partial-voluming) or a change inthe underlying “true” FA (and that in itself can vary as a result ofdiffering mixtures of crossing fibres). In this work we use veryhigh-resolution diffusion data in order to test whether disambiguationof FA changes and tract thickness might be possible. We use thehigh-resolution data as the “gold-standard”, from which to generatedata at more normal resolutions for which the underlying ground-truthis known. We demonstrate the nonlinearities in the generation of theapparent FA seen at normal resolutions, and attempt to recover theunderlying true FA from down-sampled data, even in thinner tracts.
Free Water Extraction from Diffusion Images
Ofer Pasternak1, Nir Sochen, Nathan Intrator1, Yaniv Assaf, 2
1Tel Aviv University, Tel Aviv, Israel; 2The Whol Institute for Advanced Imaging of the Tel Aviv Sourasky Medical Center, Israel
We propose a method to map and then eliminate free water from diffusion images. While free water contamination is often neglected for healthy brain imaging, it is of high importance to be eliminated for the case of cerebral edema. Our method separates the free water compartment from the remaining, contamination free, brain tissue compartment. The latter can then be analyzed by DTI methods such as FA and tractography. In addition, a free water mapping of the brain is obtained. We present the enhancement of FA and tractography on the free water eliminated data, collected by a regular DTI acquisition.
Regional Distribution of Outliers of Diffusion MRI in
the Human Brain
Lindsay Walker1, Jinzhong Yang2, Xiaoying Wu2, Kristina Simonyan1, Ragini Verma2, Carlo Pierpaoli1
1NIH, Bethesda, Maryland, USA; 2University of Pennsylvania, Philadelphia, Pennsylvania, USA
The presence of outliers in the diffusion weighted images (DWI) used for DTI of the human brain affects tensor derived quantities such as anisotropy and trace(D). We present a preliminary analysis of the regional distribution of outliers in the brain of a population of 20 healthy volunteers using the RESTORE robust tensor fitting algorithm. An outlier rejection probability map is produced showing that the occurrence of outliers is regionally consistent within the population. This implies a regionally varying statistical power across the brain, which should be considered when performing both ROI-based and voxelwise analysis of DTI data.
To Rotate B or Not to Rotate B? the Importance of
Reorienting the B-Matrix During Motion Correction of DT-MRI Data
Alexander Leemans1, Christopher John Evans1, 2, Derek K. Jones1
1School of Psychology, Cardiff University, Cardiff, UK; 2GE Healthcare, Chalfont St. Giles, UK
To estimate DTI measures, such as FA, it has been shown that the acquisition of a higher number of DW gradient directions is needed. Consequently, acquisition times are longer, increasing the adverse effect of subject motion. Although previous research indicates that correcting for such motion artifacts improves the accuracy of the FA, it remains debatable whether rotating the b-matrix (during subject motion correction) can further improve the estimation of the diffusion tensor. In this context, we investigated the effect of such a b-matrix rotation on the FA and the first eigenvector, and looked at the consequences for fiber tractography.
Improved SNR in Diffusion Spectrum Imaging with
Justin P. Haldar1, Van J. Wedeen2, Marzieh Nezamzadeh3, Guangping Dai2, Norbert Schuff3, Zhi-Pei Liang1
1University of Illinois at Urbana-Champaign, Urbana, Illinois, USA; 2Harvard Medical School, Charlestown, Massachusetts, USA; 3University of California San Francisco, San Francisco, California , USA
Diffusion spectrum imaging (DSI) is a powerful technique for the characterization of complex tissue microarchitecture. However, the potential of this technique has not been fully utilized for high-resolution biological studies because of long acquisition times and limited signal-to-noise ratio. This paper presents a new approach for reconstructing DSI images, using a statistical model that takes advantage of the high level of spatial-spectral correlation in DSI images. This method can provide significant improvements in signal-to-noise ratio relative to conventional techniques, revealing additional structures in DSI data which have previously been hidden by noise.
Quality Assessment of DTI-Based Muscle Fiber Tracking
Anneriet M. Heemskerk1, Tuhin K. Sinha1, Kevin J. Wilson1, Zhaohua Ding1, Bruce M. Damon1
1Vanderbilt University, Nashville, Tennessee, USA
DTI-based muscle fiber tracking enables reconstruction of muscle architecture but no methods currently exist to access the quality of fiber tracking. We propose a method to determine the quality of individual tracts based on the location of the fibers points of termination, paths, lengths, and similarity to neighboring fibers. Inaccurate fibers tended to be grouped, which can be expected if there are similar, spatially dependent underlying causes. We found that on average 77% of the fiber tracts generated in the tibialis anterior muscle met our quality assessment criteria.