Hans Weber¹, Martin Haas¹, Denis Kokorin^1,2, Daniel Gallichan¹, Jürgen Hennig¹, and Maxim Zaitsev^1
1Department of Radiology, Medical Physics, University Medical Center Freiburg, Freiburg, Germany, ²International Tomography Center, Novosibirsk, Russian Federation

ExLoc allows excitation and geometrically matched spatial encoding of customized curved slices, based on a combination of linear and nonlinear gradients. However, the nonlinearity of the applied slice-selection field results in curved slices with varying thickness. In this study we investigate the combination of the ExLoc technique with multi-dimensional RF-pulses for the adaptation of the slice-thickness variation. Compared to conventional multi-dimensional excitation with linear encoding fields to excite a curved slice, using this approach allows considerably shorter RF-pulses to be used.

Chao Ma^1,2, Dan Xu³, Kevin F. King³, and Zhi-Pei Liang^1,2
1Department of Electrical and Computer Engineering, University of Illinois at Urbana-Champaign, Urbana, IL, United States, ²Beckman Institute, University of Illinois at Urbana-Champaign, Urbana, IL, United States, ³Global Applied Science Lab, General Electric Healthcare, Milwaukee, WI, United States

We recently proposed a novel method for reduced FOV excitation using spatial-spectral RF pulses and second-order gradients. The method leverages the unique spatial dependence of the second-order gradients to excite a circular region-of-interest in a thin slice using a 2D spatial-spectral RF pulse. This work presents the first experimental results of the method on a 3.0T commercial MRI scanner.

Martin G Busch^1,2, and Jürgen Finsterbusch^1,2
1Department of Systems Neuroscience, University Medical Center Hamburg-Eppendorf, Hamburg, Germany, ²Neuroimage Nord, University Medical Centers Hamburg-Kiel-Lübeck, Hamburg-Kiel-Lübeck, Germany

2D-selective RF excitations are an interesting tool to minimize partial volume effects in MR spectroscopy. The blipped-planar trajectory is a simple and robust basis for such 2DRF excitations but it suffers from reduced signal efficiency if segmentation is applied to shorten the 2DRF pulse durations. This is due to the fact that with conventional segmentation only one segment samples the k-space center and yields a high signal amplitude. Here, a modified segmentation scheme is presented where each segment covers the central k-space line. Thus, the signal efficiency of segmented, blipped-planar 2DRF excitations can be considerably improved.

Eric Wong^1
1Radiology/Psychiatry, UC San Diego, La Jolla, CA, United States

Simultaneous excitation of multiple slices is proving useful for highly accelerated imaging. For N simultaneously excited slices, the peak RF power is increased by a factor of N² for excitation with uniform phase across slices. We demonstrate here that numerical optimization of the excitation phase across slices allows one to approach the theoretical minimum increase in peak power of a factor of N rather than N².

Wei Feng¹, and E Mark Haacke^1,2
1Radiology, Wayne State University, Detroit, MI, United States, ²Biomedical Engineering, Wayne State University, Detroit, MI, United States

Two-dimensional spatially selective RF pulse design evolved from investigation under the small-tip-angle (STA) assumption to general design without such a constraint, using either least squares optimization or an optimal control approach. Most work focused on optimization of the RF waveform given a fixed gradient trajectory. In this work, we propose a RF and gradient joint design algorithm under the framework of optimal control theory using a cubic B-spline waveform model. Preliminary results show that the algorithm can improve the excitation profile over the design under the STA assumption, especially for large tip angles.

Ives R Levesque¹, Jason Su², Mohammad Mehdi Khalighi³, John M Pauly¹, and Brian K Rutt^2
1Electrical Engineering, Stanford University, Stanford, CA, United States, ²Radiology, Stanford University, Stanford, CA, United States, ³Global Applied Science Laboratory, GE Healthcare, Menlo Park, CA, United States

The method of driven-equilibrium single-pulse observation of T₁ relaxation (DESPOT1) is a method which provides fast volumetric T₁ mapping, but is notoriously dependent on flip angle calibration. We propose to use k_T-points pulses to pre-compensate B₁⁺ heterogeneity in the context of brain T₁ mapping with DESPOT1. The use of k_T-points pulses resulted in improved B₁⁺ homogeneity, which translated to improved homogeneity in the volumetric T₁ map. T₁ variability in brain white matter was reduced in the by a factor of 1.2× and range by a factor of 1.4×.

Zungho Zun¹, Brian A. Hargreaves¹, and Greg Zaharchuk^1
1Radiology, Stanford University, Stanford, CA, United States

A matched-phase 90°-180° pair can achieve sharper slice profile, lower peak B1, or shorter echo time than conventional 90° and 180° RF pulses in spin echo imaging. We demonstrate that a matched-phase pair can be used to reduce the perturbation to the adjacent slice, enabling smaller slice spacing in multislice perfusion imaging using velocity-selective arterial spin labeling (ASL). The matched-phase pairs allowed the slice spacing to be reduced to one third of the imaging slice thickness, without compromising SNR or ASL signal stability.

Zhe Wang^1,2, Abbas N. Moghaddam^1,3, Meral L. Reyhan^1,4, Subashini Srinivasan^1,2, Yutaka Natsuaki⁵, J.Paul Finn^1,4, and Daniel B. Ennis^1,2
1Department of Radiological Sciences, University of California, Los Angeles, California, United States, ²Biomedical Engineering Interdepartmental Program, University of California, Los Angeles, California, United States, ³Department of Biomedical Engineering, Amirkabir University of Technology(Tehran Polytechnic), Tehran, Iran, ⁴Biomedical Physics Interdepartmental Program, University of California, Los Angeles, California, United States, ⁵Siemens Medical Solutions, Malvern, Pennsylvania, United States

Due to the gross annular geometry of the left ventricle, a radial tagging sequence may be advantageous for measuring LV contraction and myocardial twist. Radial tagging, in general, requires shifting the patient table away from the iso-center of the main magnetic field to generate a tagging profile that is both sharp and centered at the middle of the LV cavity. In this study we show that >91% of patients can be acceptably imaged with the radial tagging sequence by retrospectively analyzing the short-axis slice position and orientation information from 75 patients in the Cardiac Atlas Project database.

Jun Shen^1
1NIMH, Bethesda, MD, United States

A fast transform is proposed that converts, under the linear response approximation, existing RF pulses into ones that preserve frequency selectivity in the presence of rapid T₂^* relaxation.

Bastien Guerin¹, Matthias Gebhardt², Elfar Adalsteinsson^3,4, and Lawrence L. Wald^1,4
1Martinos Center for Biomedical Imaging, Dept. of Radiology, Massachusetts General Hospital, Charlestown, MA, United States, ²Siemens Healthcare, Erlangen, Germany, ³Dept of Electrical Engineering and Computer Science, Massachusetts Institute of Technology, Cambridge, MA, United States, ⁴Harvard-MIT Division of Health Sciences Technology, Cambridge, MA, United States

Electromagnetic simulations have shown that pTx pulse design with average forward power regularization does not constrain local SAR. We propose two pulse design interior point algorithms that explicitly constrain spatial fidelity, local and global SAR, average and maximum forward power. The first (second) minimizes excitation error (local SAR) while constraining local SAR (excitation error), global SAR and maximum/average power. We show the equivalence of these two approaches and their capability in handling large numbers of constraints without the need of defining a tradeoff parameter, giving more flexibility to the operator to explore the pulse design parameter space in real time.

Tomasz Dawid Lindel^1,2, Andre Kuehne^1,2, Patrick Waxmann^1,2, Frank Seifert^1,2, Thoralf Niendorf², and Bernd Ittermann^1,2
1Medical Metrology, Physikalisch-Technische Bundesanstalt (PTB), Abbestr. 2-12, 10587 Berlin, Germany, ²Max Delbrück Center for Molecular Medicine (MDC), Berlin Ultra-High Field Facility (BUFF), Berlin, Berlin, Germany

For brain imaging two-dimensional spatially selective excitation using parallel transmission (pTx) is a well-established technique these days. The missing selectivity along the third spatial dimension is a severe limitation, however. To overcome this we combined 2D-SSE with a conventional slice selective refocusing pulse for phantom and in vivo imaging at 7T. Initially, this resulted in severe signal inhomogeneities due to the spatially non-uniform refocusing at the ultrahigh field. Using RF shimmed refocusing pulses combined with adapted pTx pulses the problem could be resolved.

Yong Pang¹, Daniel Vigneron^1,2, and Xiaoliang Zhang^1,2
1Radiology and Biomedical Imaging, University of California San Francisco, San Francisco, CA, United States, ²UCSF/UC Berkeley Joint Graduate Group in Bioengineering, San Francisco & Berkeley, CA, United States

In this work, we applied the precompensation method to correct the mutual coupling effect between array elements for the spatial domain method parallel transmission by taking mutual coupling coefficient matrix into pulse design procedure. To investigate its feasibility and efficiency, a 4-element coil array was used and excitation pulses were designed using both the spatial domain method and the precompensation method for comparison. The results show that the proposed precompensation method can effectively reduce the artifacts caused by mutual coupling, yielding enhanced tolerance to mutual coupling of RF arrays in parallel transmission, and providing improved excitation profile.

Wei-Tang Chang^1,2, Jyrki Ahveninen¹, and Fa-Hsuan Lin^1,2
1Martinos center for biomedical imaging, Massachusetts general hospital, Charlestown, Massachusetts, United States, ²Biomedical engineering, National Taiwan University, Taipei, Taiwan

Sparse source cluster reconstruction by compressed magnetic resonance inverse imaging

Yuchou Chang¹, Kevin F. King², Dong Liang³, and Leslie Ying^1
1Electrical Engineering, University of Wisconsin - Milwaukee, Milwaukee, Wisconsin, United States, ²Global Applied Science Laboratory, GE Healthcare, Waukesha, Wisconsin, United States, ³Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen, China

CS-GRAPPA has the benefit of decoupling CS and GRAPPA without the need for coil sensitivities. However, noise and errors from the CS step can propagate and be amplified in GRAPPA. We recently developed a nonlinear GRAPPA (NLGRAPPA) approach that can suppress the GRAPPA noise significantly. In this work, we propose to integrate CS and NLGRAPPA to improve CS-GRAPPA reconstruction. The NLGRAPPA step can reduce the amplification of noise and errors in CS reconstruction. Experimental results using phantom and in vivo data demonstrate that the proposed method can significantly improve the reconstruction quality over CS-GRAPPA at high net reduction factors.

Tom Depew¹, and Qing-San Xiang^1,2
1Physics & Astronomy, University of British Columbia, Vancouver, BC, Canada, ²Radiology, University of British Columbia, Vancouver, BC, Canada

New GRAPPA-like Partially Parallel Imaging (PPI) algorithms consider coil sensitivity information in dimensions other than the phase encode direction that is implicit in PPI data recovery. Sensitivity variation in the frequency encode dimension must be accounted for to allow an optimal reconstruction. The assessment of PPI reconstruction quality often requires a fully acquired k-space reference scan for comparison, which is not possible for partial data. We propose a GRAPPA variant that accounts for FE sensitivity variation and is optimized using gradient energy, an image quality measure that does not require a fully acquired reference scan.

Marcel Gutberlet^1
1Institute of Radiology, Medical School Hannover, Hannover, Lower Saxony, Germany

Conventional SENSE imaging uses the assumption that the voxels are Dirac distributions and that the voxel functions fulfil the orthonormality relation, which is called the weak voxel approach. These two assumptions of the voxel functions allow a simple and fast reconstruction but may also result in residual aliasing. In this work, an analytic expression for the encoding matrix and the correlation matrix of the encoding functions is presented allowing an efficient and accurate SENSE reconstruction using the weak voxel and the strong voxel approach without the assumption of voxels being Dirac distributions.

Jonathan R. Polimeni¹, Kawin Setsompop¹, Borjan A. Gagoski², Jennifer A. McNab¹, Christina Triantafyllou^1,3, and Lawrence L. Wald^1,4
1Athinoula A. Martinos Center for Biomedical Imaging, Department of Radiology, Harvard Medical School, Massachusetts General Hospital, Charlestown, MA, United States, ²Department of Electrical Engineering and Computer Science, Massachusetts Institute of Technology, Cambridge, MA, United States, ³A. A. Martinos Imaging Center, McGovern Institute for Brain Research, Massachusetts Institute of Technology, Cambridge, MA, United States, ⁴Harvard-MIT Division of Health Sciences and Technology, Massachusetts Institute of Technology, Cambridge, MA, United States Minor Outlying Islands

Highly-accelerated imaging allows for rapid single-shot EPI acquisitions to mitigate distortion and blurring effects. However there is an intrinsic SNR penalty that scales with the square-root of the acceleration factor and with the g-factor, ultimately limiting its usability. Multi-shot segmented EPI acquisitions can similarly mitigate these deleterious effects, yet the longer temporal sampling interval amplifies physiological noise and system instabilities. Here, we merge a segmented multi-shot EPI acquisition with the Simultaneous Multi-Shot technique. This combination of techniques allows each EPI segment to employ a distinct multi-slice excitation pulse, enabling advantageous slice-aliasing patterns to reduce the g-factor of the image reconstruction.

Wei Lin¹, Feng Huang¹, Charles Saylor¹, Enrico Simonotto¹, Arne Reykowski¹, and Randy Duensing^1
1Invivo Corporation, Philips Healthcare, Gainesville, Florida, United States

The SNR performance of radial GROWL is compared with Cartesian SENSE. Monte Carlo simulation results show that GROWL without regularization gives a lower RMSE and g-factor map than Cartesian SENSE, due to the ability to use coils aligned along two orthogonal directions. With a k-space adaptive Tikhonov regularization, the g-factor of GROWL reconstruction can be further lowered than much less than the unity. In vivo brain study demonstrates that radial GROWL can be used to accelerate 3D MP-RAGE acquisition by a factor of eight using an 8-channel head coil with an acceptable image SNR.

Paul T. Weavers¹, Eric A. Borisch¹, Casey P. Johnson¹, and Stephen J. Riederer^1
1MR Research Laboratory, Mayo Clinic, Rochester, MN, United States

The performance of 3D SENSE accelerated CE-MRA can be improved by intelligently apportioning the acceleration between the two phase encode directions. The field of view, anatomy to be imaged, and receive coil array positioning may support an optimal set of accelerations that provides maximum acceleration with minimum penalty. We provide framework for finding this optimal set of accelerations, and demonstrate in-vivo a 20% increase in acceleration with negligible increase in g-factor based noise amplification.

Haifeng Wang¹, Yuchou Chang¹, Dong Liang², King F Kevin³, and Leslie Ying^1
1Department of Electrical Engineering and Computer Science, University of Wisconsin-Milwaukee, Milwaukee, WI, United States, ²Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen, Guangdong, China, ³Global Applied Science Laboratory, GE Healthcare, Waukesha, WI, United States

A novel data acquisition method using cross sampling and image reconstruction method nonlinear GRAPPA are integrated to improve the image quality of GRAPPA at high accelerations. Cross sampling is used to acquire the ACS lines, and a nonlinear model is used in reconstruction of the missing k-space data. The integrated method brings together the benefit of cross-sampled GRAPPA in ACS reduction and the benefit of nonlinear GRAPPA in noise suppression. Results from in vivo experiments demonstrate the proposed method is able to reduce the aliasing artifacts in GRAPPA without compromising SNR when a high net reduction factor is used.

D. Nickel¹, R. Grimm², and K. T. Block^3
1MR Applications Development, Healthcare Sector, Siemens AG, Erlangen, Germany, ²Pattern Recognition Lab, University of Erlangen-Nuremberg, Erlangen, Germany, ³Department of Radiology, NYU Langone Medical Center, New York, United States

Incoherent undersampling of k-space is often desirable for iterative reconstruction techniques like Compressed Sensing. On the other hand, noise propagation from parallel imaging is best explored for regular undersampling patterns. Instead of focusing on the local noise enhancement in the image, typically expressed through g-factor maps, we derive an expression for the averaged noise amplification, which can be used as a measure to evaluate given undersampling patterns. Because the expression can be evaluated efficiently, it can be used during the generation of low-noise sampling patterns, mainly aiming at 3D acquisitions with incoherent undersampling in 2D.

Suchandrima Banerjee¹, Yuval Zur², Atsushi Takahashi¹, Ajit Shankaranarayanan¹, and Douglas A.C. Kelley^3
1Global Applied Science Lab, GE Healthcare, Menlo Park, California, United States, ²GE Healthcare, Tirat Carmel, Israel, ³Global Applied Science Lab, GE Healthcare, San Francisco, California, United States

Simultaneous multi-slice excitation can reduce the repetition time needed to acquire a fixed number of slices. Its potential has recently been demonstrated for high resolution full-brain diffusion imaging and functional MRI. In previous works that exploited coil sensitivity differences to separate slices that were simultaneously excited by multi-band excitation, calibration data was acquired from a separate scan where each slice was excited independently. In this work we propose a time-interleaved parallel imaging approach to slice separation which avoids additional prescan, provides co-localization between calibration and the actual acquisition and allows periodic re-computation of the kernel in a time-series acquisition.

Morwan Choli¹, Felix A. Breuer¹, Peter M. Jakob^1,2, and Martin Blaimer^1
1Research Center Magnetic Resonance Bavaria e.V (MRB), Wuerzburg, Germany, ²Department of Experimental Physics 5, University of Würzburg, Wuerzburg, Germany

In parallel imaging, reliable simultaneous multi-slice imaging needs sufficient coil sensitivities in slice direction to separate closely spaced slices. In this work we present an alternative approach for accelerated simultaneous multi-slice imaging. During excitation a phase difference is imprinted on the slices. Simultaneous recorded slices where then separated by using this phase information in the parallel imaging reconstruction by using the virtual coil concept. High quality reconstructions are presented for two slice simultaneous excitation for human head images at 3T.

Daniel S Shefchik¹, Andrzej Jesmanowicz¹, Matthew Budde², and Andrew S. Nencka^1
1Department of Biophysics, Medical College of Wisconsin, Milwaukee, WI, United States, ²Department of Neurosurgery, Medical College of Wisconsin, Milwaukee, WI, United States

The phase of reconstructed spin echo images is generally zero when the echo is centered and is usually discarded upon reconstruction. However, phase can be manipulated through shifting Fourier encoding by one half pixel in the phase encoding direction to introduce a linear phase gradient that can be used for spatial encoding. This spatial encoding, similar to coil sensitivities in SENSE, enables proper unaliasing of an accelerated image with even a single coil. The phase-SENSE method is demonstrated with a spin echo pulse sequence and a single-channel receiver at 9.4T on a rodent brain.

Suhyung Park¹, and Jaeseok Park^1
1Department of Brain and Cognitive Engineering, Korea University, Seoul, Seoul, Korea

Parallel imaging techniques have been widely used to reduce total acquisition time and subject motion in clinical application by using the spatial information inherent in a multiple receiver coils. However, with increasing acceleration factors, they lead to residual artifacts and amplified noises over the whole image due to corrupted data with noise. To overcome these problems, several regularization approaches have been proposed using the framework of Tikhonov regularization, such as prior-regularized GRAPPA, but a direct tradeoff between image blurring and noise amplification still remain substantially. From a different viewpoint, high pass GRAPPA (HP-GRAPPA) tried to address this problem controlling low frequency energy with high pass filter (HPF), but was still challenging to find optimal high pass band in k-space. In this work, we propose generalized HP-GRAPPA (GHP-GRAPPA) resolving a high pass band problem as a new regularization approach with high accuracy and quality.

Anuj Sharma^1,2, Sasidhar Tadanki^1,3, Marcin Jankiewicz^1,3, and William A Grissom^1,2
1Vanderbilt University Institute of Imaging Science, Vanderbilt University, Nashville, TN, United States, ²Department of Biomedical Engineering, Vanderbilt University, Nashville, TN, United States, ³Department of Radiology, Vanderbilt University, Nashville, TN, United States

A method to accelerate Bloch-Siegert B1+ mapping beyond the number of receive channels is presented, in which the Bloch-Siegert phase shift is incorporated into the receive sensitivities for the purposes of reconstruction, leading to an augmented set of virtual receive coils. Results show that the method achieves low B1+ map reconstruction error for a 4-coil 7T TEM transmit array with a factor of 25 acceleration using only 8 physical receive coils

Kang Wang¹, Scott K. Nagle^2,3, Philip J. Beatty⁴, James H. Holmes¹, Dan W. Rettmann⁵, and Jean H. Brittain^1
1Global Applied Science Laboratory, GE Healthcare, Madison, WI, United States, ²Radiology, University of Wisconsin-Madison, Madison, WI, United States, ³Medical Physics, University of Wisconsin-Madison, Madison, WI, United States, ⁴Global Applied Science Laboratory, GE Healthcare, Toronto, ON, Canada, ⁵Global Applied Science Laboratory, GE Healthcare, Rochester, MN, United States

In this work, a new implementation of the HighlY constrained Cartesian Reconstruction (HYCR) is proposed to address the memory and computation time challenges of the original HYCR implementation. This was achieved by performing the channel combination in k-space before Fourier Transform using the DirectVirtual Coil (DVC) technology. Comparison of overall image quality, line profiles and temporal waveform was performed for the original and the proposed HYCR implementations, using a pulmonary perfusion application as an example.

Bo Zhao^1,2, Wenmiao Lu², and Zhi-Pei Liang^1,2
1Department of Electrical and Computer Engineering, University of Illinois at Urbana-Champaign, Urbana, IL, United States, ²Beckman Institute of Advanced Science and Technology, University of Illinois at Urbana-Champaign, Urbana, IL, United States

A new reconstruction method is proposed for accelerating parametric mapping using partial separability and sparsity constraints jointly. The joint use of the two constraints yields parameter maps with higher spatial resolution and SNR than several state-of-the-art methods.

Jan Aelterman¹, Hiep Luong¹, Steven Baete², Bart Goossens¹, Aleksandra Pizurica¹, and Wilfried Philips^1
1TELIN-IPI-IBBT, Ghent University, Ghent, Belgium, ²NYU Langone Medical Center, United States

Very often, clinicians will ‘zoom in’ on an MRI image, this means implicitely doing a bilinear, bicubic or sinc interpolation, which gives rise to artifacts. We propose an augmented Lagrangian based reconstruction method, which is akin to compressed sensing MRI reconstruction, to perform this same type of image resolution enhancement. The proposed method dramatically reduces the ringing artifact, while preserving sharpness. We also present a method for estimating the free parameter in the reconstruction formulation in order to accelerate the algorithm.

Huajun She¹, Rong-Rong Chen¹, Dong Liang^2,3, Edward DiBella⁴, and Leslie Ying^3
1Department of Electrical and Computer Engineering, University of Utah, Salt Lake City, Utah, United States, ²Shenzhen Institutes of Advanced Technology, Shenzhen, China, ³Department of Electrical Engineering and Computer Science, University of Wisconsin, Milwaukee, WI, United States, ⁴Department of Radiology, University of Utah, Salt Lake City, Utah, United States

This work investigates the blind multichannel under-sampling problem where both the channel functions and signal are reconstructed simultaneously. We propose a new approach to blind compressed sensing in the context of parallel imaging where the sensing matrix is not known exactly and needs to be reconstructed. The proposed method effectively incorporates the sparseness of both the desired image and coil sensitivities in reconstruction of both the coil sensitivities and image simultaneously from randomly-undersampled, multichannel k-space data. The proposed method is compared with Sparse SENSE and L1 SPIRiT and demonstrates a significant improvement in image quality at high reduction factors.

Kamlesh Pawar^1,2, and Arjun Arunachalam^1
1Electrical Engineering, Indian Institute of Technology Bombay, Mumbai, India, ²IITB Monash Research Academy, Mumbai, India

A new technique for accelerating a dynamic MRI scan through simultaneous image and k- space domain aliasing is demonstrated. Parallel Imaging (PMRI) and Compressed Sensing (CS) techniques provide an acceleration factor of A_PMRI/CS while A_kspace via k-space aliasing is provided by a new method that uses multiple RF excitation pulses and gradients to overlap distinct k-space points. Both types of aliasing are corrected during image reconstruction to restore the fidelity of the accelerated dataset. The total acceleration from such a combination is given by A_kspace*A_PMRI/CS as data is sub-sampled in both the image and k-space domains simultaneously during data acquisition.

Jun Liu¹, Jeremy Rapin¹, Ti-chiun Chang¹, Peter Schmit², Xiaoming Bi³, Alban Lefebvre¹, Michael Zenge², Edgar Mueller², and Mariappan S. Nadar^1
1Siemens Corporate Research, Princeton, NJ, United States, ²Siemens AG, Healthcare Sector, Erlangen, Germany, ³Siemens Medical Solutions USA Inc., Chicago, IL, United States

Non-contrast-enhanced 4D intracranial MR angiography (NCE 4D MRA) is a promising non-invasive technique for visualization of vascular anatomy. In this paper, a parallel imaging method is proposed that makes use of a regularization based on 4D redundant Haar wavelet transformation, which allows for incorporating both spatial and temporal structures. NCE 4D MRA images were reconstructed from k-space data that was under-sampled according to a spiral phyllotaxis pattern. Our results show that excellent results can be achieved with an acceleration rate of even 13.7.

Jonathan I. Sperl¹, Ek T. Tan², Marion I. Menzel¹, Kedar Khare², Kevin F. King³, Christopher J. Hardy², and Luca Marinelli^2
1GE Global Research, Garching n. Munich, BY, Germany, ²GE Global Research, Niskayuna, NY, United States, ³GE Healthcare, Waukesha, WI, United States

Diffusion spectrum imaging (DSI) acquisition can be accelerated by randomly undersampling q-space and subsequent compressed sensing (CS) reconstruction. This work presents several extensions to CS-DSI, namely the combination of the approximate message passing (AMP) and Nesterov updates, the application of AMP to total variation minimization, and random translations for wavelet based CS. All methods can be combined yielding superior convergence properties. Fiber simulations and in vivo brain data are analyzed demonstrating the improvements in terms of speed and accuracy.

Jon Furuyama¹, Brian Burns¹, Neil Wilson¹, and M. Albert Thomas^1
1Radiology, UCLA, Los Angeles, CA, United States

Four-dimensional spectroscopic imaging techniques (2 spatial, 2 spectral) can be accelerated by making use of an echo-planar spectroscopic imaging readout to interleave the collection of one spatial and one spectral dimension within a single TR. The two remaining dimensions must be incrementally collected, requiring scan times on the order of 20+ minutes, well above what is clinically acceptable. We show that the application of Compressed Sensing (CS) techniques can be implemented successfully in a clinical setting to accelerate the collection of the remaining two dimensions. Examples of actual CS collected and reconstructed J-resolved spectra from the prostate are shown.

Mariya Doneva¹, Holger Eggers¹, and Peter Börnert^1
1Philips Research Europe, Hamburg, Germany

In this work, we investigate the influence of the sampling pattern on the convergence behaviour of l1-regularized SENSE reconstruction at different reduction factors. In other words, we try to answer the question what improvement can CS-SENSE provide over l1-denoised SENSE?

Zhaoyang Jin¹, and Qing-Sang Xiang^2
1Institute of Information and Control, Hangzhou Dianzi University, Hangzhou, Zhejiang, China, ²Radiology Department, University of British Columbia, Vancouver, BC, Canada

Skipped phase encoding and edge deghosting (SPEED) is an effective method for scan time reduction. Previously, PE skip size was limited to be a prime number in order to avoid reconstruction difficulty in matrix inversion due to potential ghost phase degeneracy. This study revealed that while the prime number requirement is helpful, this restriction can be very much relaxed. Composite numbers combined with appropriately chosen PE shifts can also yield good reconstruction results. It was demonstrated that much more flexible data acquisition schemes can be used, offering significantly more freedom in practical implementations and applications.

Hao Gao^1,2, Stanislas Rapacchi³, Da Wang³, John Moriarty³, Conor Meehan³, James Sayre³, Gerhard Laub⁴, Paul Finn³, and Peng Hu^3
1Mathematics, UCLA, Los Angeles, CA, United States, ²Mathematics, UCI, Irvine, CA, United States, ³Radiology, UCLA, Los Angeles, CA, United States, ⁴Siemens

We propose a novel CS method for dynamic MRI applications using Prior Rank, Intensity and Sparsity Model (PRISM) and evaluate this technique for cardiac cine MRI. PRISM differs from the previous CS techniques in the ability to apply the sparsifying transform (TF) after background suppression using rank minimization. PRISM was tested on cardiac cine MRI data sets acquired on 6 healthy subjects. The data was fully sampled and retrospectively undersampled. Results show dynamic 2D MRI could be greatly accelerated using PRISM. PRISM provides good-quality image series even from highly undersampled kspace data when state-of-the art traditional compressed sensing fails.

Merry Mani^1,2, Mathews Jacob³, Arnaud Guidon⁴, Chunlei Liu⁵, Allen Song⁵, and Jianhui Zhong^2,6
1Electrical and Computer Engineering, University of Rochester, Rochester, NY, United States, ²Rochester Center for Brain Imaging, Rochester, NY, United States,³Electrical and Computer Engineering, University of Iowa, ⁴Biomedical Engineering, Duke University, ⁵Duke University, ⁶University of Rochester

Both high angular and high spatial resolution are highly desirable in diffusion imaging applications. However, the acquisition time gets prohibitively long as the resolution in both dimensions are increased. We propose a new acquisition strategy to simultaneously enhance the resolution in both dimensions of diffusion imaging in a reasonable scan time. We achieve this by under-sampling the combined k-q acquisition space of diffusion imaging and using a compressed sensing strategy for reconstruction. Results show that at least 6 fold acceleration in scan time is possible, making it feasible to achieve 1mm in-plane resolution and high angular resolution in around 8 minutes.

Suhyung Park¹, and Jaeseok Park^1
1Department of Brain and Cognitive Engineering, Korea University, Seoul, Seoul, Korea

Recently, several imaging techniques for combination of compressed sensing (CS) and parallel imaging (PI) have shown the possibility that can reduce total acquisition time and improve temporal and spatial resolution. Among them, approach proposed by Sung et al., which separated reconstruction methods in k-space, showed improved image quality compared to other methods. However, this method was vulnerable to image-degrading artifacts due to signal discontinuity occurring in boundary not only between high frequency sub-sections in k-space but also between CS and PI reconstruction data. In this work, we propose improved high frequency CS and new reconstruction method optimally synthesizing CS and PI data.

Xue Feng¹, Michael Salerno², Christopher M Kramer^2,3, and Craig H Meyer^1,3
1Biomedical Engineering, University of Virginia, Charlottesville, Virginia, United States, ²Medicine, University of Virginia, Charlottesville, Virginia, United States,³Radiology, University of Virginia, Charlottesville, Virginia, United States

In dynamic MRI, spatial and temporal parallelism are exploited to reduce scan time. A real-time reconstruction is sometimes necessary for timely feedback during the scan. In this study a Kalman filter model suitable for real-time reconstruction is used to increase the temporal resolution. The original application of the Kalman filter to dynamic MRI was limited to non-Cartesian trajectories; here we overcome this limitation and apply the model to the more commonly used Cartesian trajectory. Furthermore, we combine the Kalman model with spatial parallel imaging techniques to further increase the spatial and temporal resolution and SNR.

Mei-Lan Chu¹, Hsiao-Wen Chung¹, Yi-Ru Lin², and Tzu-Cheng Chao^3,4
1Graduate Institute of Biomedical Electronics and Bioinformatics, National Taiwan University, Taipei, Taiwan, ²Department of Electronic Engineering, National Taiwan University of Science and Technology, Taipei, Taiwan, ³Institute of Medical Informatics, National Cheng-Kung University, Tainan, Taiwan, ⁴Dept. of Computer Science and Information Engineering, National Cheng-Kung University, Tainan, Taiwan

The fidelity of training data has been one major limitation for k-t reconstruction method at present. The traditional acquisition of the training data is performed in a separate scan, which may not exactly follow the same procedure of the real acquisition stage. Another acquisition scheme is variable density k-t sampling pattern that is the lines of central k-space are fully sampled while the peripheral lines are under-sampled. The training data can be extracted from central k lines, and the temporal resolution is maintained. However, the number of acquired central k lines for training data implies a tradeoff between the net accelerating ratio and the quality of training data. In this work, we propose to solve the reconstructing problem by exploiting data-driven method to acquire the prior knowledge of imaged object. The method uses the idea that each x-f profile can be transformed into a linear combination of features, which are the principal components of the distribution of x-f profiles. The principal components can be extracted from several existing homogenous data, instead of the data acquired directly from the imaged object. This is the main difference between traditional k-t method and the proposed method. We demonstrate its feasibility with numerical simulations of cine cardiac imaging. The results show that the proposed method can achieve comparable temporal resolution and leads to reduced artifacts even from substantially down-sampled k-space data.

Mei-Lan Chu¹, Ping-Huei Tsai^2,3, Hsiao-Wen Chung¹, Hsu-Hsia Peng⁴, and Cheng-Wen Ko^5
1Graduate Institute of Biomedical Electronics and Bioinformatics, National Taiwan University, Taipei, Taiwan, ²Imaging Research Center, Taipei Medical University, Taipei, Taiwan, ³Department of Radiology, WanFang Hospital, Taipei, Taiwan, ⁴Department of Biomedical Engineering and Environmental Sciences, National Tsing Hua University, Hsinchu, Taiwan, ⁵Department of Computer Science and Engineering, National Sun Yat-sen University, Kaohsiung, Taiwan

Previous research of reconstructing dynamic MR imaging with arbitrary sampling trajectory was mainly rely on the traning data whcih extracted from central k-space. However, the training data is proned to streaking artifact from substantially under-sampled data. We propose a prior-data-driven method to address this problem. The method mainly rely on x-f principal components extracted from existing images with similar anatomical position, which make the prior information free from streaking artifact. We demonstrate the feasibility of the proposed method wtih simulation of radial imaging with simulation of radial imaging, and the results show that the method robustly reconstuct under-sampled dynamic images of arbitrary trajectory.

Ganesh Adluru¹, Liyong Chen^1,2, David Feinberg², Jeffrey Anderson¹, and Edward V.R. DiBella^1
1Radiology, University of Utah, Salt Lake City, UT, United States, ²Advanced MRI Technologies, Sebastopol, CA, United States

Image reconstruction using a rank penalty term is a promising way to remove undersampling artifacts in multi-image MRI. Exciting results have been reported in dynamic imaging situations where temporal signal changes are highly correlated. However, when the underlying true data have a lot of variation, a low rank constraint may not be the best choice. Here we propose a reordering technique to improve rank constrained reconstructions in such cases. Pixel intensities in the matrix of the multi-image estimate are reordered based on the sorting order of a prior. This results in a better match with the low rank model. Promising results are presented on undersampled multi-image diffusion data.

Zhaoyang Jin¹, and Qing-Sang Xiang^2
1Institute of Information and Control, Hangzhou Dianzi University, Hangzhou, Zhejiang, China, ²Radiology Department, University of British Columbia, Vancouver, BC, Canada

Skipped phase encoding and edge deghosting (SPEED) is an effective method for scan time reduction. Similar to compressed sensing, it uses a sparsifying operation in the reconstruction process. Previously, this was realized with a simple differential operation. This study investigated wavelet transform as another option. Results from differential, wavelet, and combined operations were compared. It was found that the wavelet approach can have reduced reconstruction error than the original differential approach, and when the two operations are combined in a series, even smaller reconstruction error was achievable.

Xiao Wang¹, Enhao Gong¹, Zhengwei Zhou¹, Sheng Fang², Kui Ying³, and Shi Wang^3
1Biomedical Engineering, Tsinghua University, Beijing, China, ²Institute of nuclear and new energy technology, Tsinghua University, Beijing, China, ³Engineering physics, Tsinghua University, Beijing, China

Compressed sensing (CS) is a newly developed method which can reduce the scan time of MR imaging. A new method based on coherence assumption and nonlocal operator (CORNOL) is proposed. We implemented CORNOL instead of Total-Variation (TV) on the reconstruction of CS. Thus only intrastructure intensity changes are penalized while the interstructure intensity changes are preserved. The result demonstrates both noise penalization and detail structure preserving character. Phantom simulation and in-vivo data shows the validity and advantages of our method.

Enhao Gong¹, Xiao Wang¹, Kui Ying², and Shi Wang^2
1Biomedical Engineering, Tsinghua University, Beijing, China, ²Engineering Physics, Tsinghua University, Beijing, China

Compressed sensing (CS) is an emerging acceleration technique and recently applied for MRI. Conventional CS reconstruction techniques are based on simplistic sparsity of signals and use uniform L1-norm penalty regardless of whether the coefficients contribute to significant information for pathological diagnosis. This results in reconstruction errors, like blurring details. We proposed a new algorithm that uses Hidden Markov Tree model to extract structural information in wavelet domain. Sparsity is regulated by exploiting statistical structural matrices, such that important coefficients are enhanced and artifacts are further reduced. Phantom simulation and in-vivo experiments show the validity and advantages of our algorithm.

David S Smith¹, Lori R Arlinghaus¹, Thomas E Yankeelov¹, and E Brian Welch^1
1Institute of Imaging Science, Vanderbilt University, Nashville, TN, United States

We show that it is possible to optimize the random sampling pattern in compressed sensing MRI using a target-independent measure before data acquisition. Despite the nonlinear and random nature of the CS reconstruction and the spatially variant PSF, we show on a T1-weighted complex image of the breast that the variance of the PSF of the sampling pattern could be a reliable predictor of the ultimate reconstruction quality¬. The linear correlation of the variance of the pattern PSF with the normalized mean square error of the reconstructed image was –0.55.

Yuqiong Ding^1,2, Yiu-Cho Chung^1,2, Leslie Ying³, and Dong Liang^1,2
1Paul C. Lauterbur Research Centre for Biomedical Imaging, Shenzhen Institutes of Advanced Technology, Shenzhen, Guangdong, China, ²Key Laboratory of Health Informatics, Chinese Academy of Sciences, Shenzhen, Guangdong, China, ³Department of Electrical Engineering and Computer Science, University of Wisconsin-Milwaukee, Milwaukee, WI, United States

As an emerging reconstruction technique, compressed sensing (CS) has demonstrated great potential to reconstruct high quality images from undersampled k-space data. However, the noise behavior of CS reconstruction in MRI remains largely unexplored. This work analyzes how noise is distributed and changed with increasing accelerations. We particularly focus on dynamic contrast-enhanced imaging (DCE-MRI), where the temporal and spatial noise behavior in CS-based DCE-MRI is characterized using the Marcenko-Pastur (MP)-Law method. The study provides a qualitative understand of the noise behavior in CS reconstructed DCE images. Such understanding can accelerate application of CS in clinical practice.

YUDONG ZHANG^1,2, BRADLEY PETERSON^1,2, and ZHENGCHAO DONG^1,2
1Brain Imaging Lab, Columbia University, New York, NY, United States, ²New York State Psychiatric Inst., New York, NY, United States

Compressed sensing was introduced to the field of magnetic resonance imaging in recent years as a promising method to significantly reduce scan time. The performance of CS depends on the sparsity of the image in the sparse domain, such as wavelet transform domain. In this report, we proposed a method to increase the sparsity of CS MRI by taking exponent of wavelet transform normalized to the range [0 1]. The method was tested on a digital phantom and in vivo MRI data, and the results show that EXP-WT can improve the quality of the reconstructed image or significantly speed up the reconstruction.

Mariya Doneva¹, and Peter Börnert^1
1Philips Research Europe, Hamburg, Germany

Fast reconstruction is crucial for the implementation of CS-SENSE on clinical scanners. Thus, improvements of the reconstruction speed both in terms of algorithms with improved convergence and parallel implementation are desirable. In this work, we propose a modified CS-SENSE reconstruction method based on the Nesterov’s optimal gradient scheme, which is less sensitive to inaccuracies in the coil sensitivities estimation and has an improved convergence speed.

Yihang Zhou¹, Yuchou Chang¹, and Leslie Ying^1
1College of Engineering and Applied Science, University of Wisconsin-Milwaukee, MILWAUKEE, WI, United States

We propose a new dynamic MRI reconstruction method that effectively combines the compressed sensing based dynamic imaging technique with parallel MRI technique. The method decouples the reconstruction process into two sequential steps. In the first step, a series of aliased dynamic images is reconstructed using a CS method from the highly undersampled k-space data. In the second step, the missing k-space data for the original image are reconstructed by the nonlinear GRAPPA technique. Experimental results using in vivo data demonstrate that the proposed method is able to preserve both spatial resolution and spatial variations at high accelerations.

Gregory R. Lee^1,2, Jeffrey L. Sunshine^1,2, and Mark A. Griswold^1,2
1Radiology, Case Western Reserve University, Cleveland, OH, United States, ²University Hospitals Case Medical Center, Cleveland, OH, United States

Non-Cartesian 3D acquisitions, when combined with parallel imaging and compressed sensing, have great potential for accelerating MR image acquisition. However, compressed sensing reconstructions of large 3D datasets remains computationally challenging. In the present work, a simple algorithm that alternates between iterations of a conjugate gradient SENSE algorithm and a recently proposed transform-domain denoising operation is proposed. The proposed technique does not require the tuning of any regularization parameters and requires only a background noise standard deviation as input to the denoising routine. High quality reconstructions are demonstrated for contrast-enhanced MRA images undersampled by a factor of 40-150.

Xiang Feng¹, Guoxi Xie¹, Chao Zou¹, Anthony G. Christodoulou², Xin Liu¹, and Bensheng Qiu^1
1Paul C. Lauterbur Research Centre for Biomedical Imaging, Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen, China,²Department of Electrical and Computer Engineering, University of Illinois at Urbana-Champaign, Urbana, IL, United States

The Partial Separability (PS) of spatiotemporal signals has been exploited effectively for sparse (k, t)-space sampling in dynamic MRI. It has been successfully applied to real-time cardiac MRI and myocardial perfusion. In the conventional PS model-based sparse sampling scheme, the navigator data are reordered in every navigator cycle by using a projection strategy. This reordering method assumes that temporal signal changes are negligible within the navigator cycle. However, this assumption may result in an suboptimal temporal estimation. To address this issue, we present a sliding window method for reordering the navigator data. In vivo cardiac imaging results demonstrated that the proposed method could produce much better reconstructions with higher temporal resolution.

Jun Deng¹, Yi Lu², and Wenmiao Lu^3
1Electrical & Electronic Engineering, Nanyang Tech. University, Singapore, SG, Singapore, ²Electrical and Computer Engineering, University of Illinois, Urbana-Champaign, Illinois, United States, ³Beckman Institute, University of Illinois, Urbana-Champaign, Illinois, United States

This work introduces a fast l_1-minimization algorithm for CS based on orthonormal expansion of sensing matrix. The proposed algorithm converges significantly faster than commonly used non-linear conjugate gradient method without compromising the reconstruction accuracy.

Md Mashud Hyder^1,2, Bradley Peterson^1,2, and Zhengchao Dong^1,2
1Brain Imaging Lab, Columbia University, New York, NY, United States, ²New York State Psychiatric Inst., New York, NY, United States

The theory of compressed sensing (CS) states that MRI images can be recovered efficiently from randomly undersampled k-space data under certain conditions. Many MRI images have sparse representation when we calculate their spatial finite-differences or wavelet coefficients. In this work we show that spatial finite-differences of a set of wavelet coefficients can increase the sparsity of MRI image, which results in improved recovery of CS MRI.

Stanislas Rapacchi¹, Robert X. Smith², Danny J.J. Wang², and Peng Hu^1
1Radiology, UCLA, Los Angeles, CA, United States, ²Neurology, UCLA, Los Angeles, CA, United States

We propose to investigate the potential implementation of compressed sensing (CS) to accelerate ASL acquisitions by undersampling using two different strategies: 1) To under-sample each acquisition and apply CS individually then subtract the reconstructed images. 2) To undersample both acquisitions identically in order to perform the subtraction in the k-space domain. ASL data were acquired on the brain of one volunteer with fully sampled k-space then under-sampled retrospectively. CS reconstruction was compared to reference images. Separated CS reconstruction of the 2 acquisitions seems more appropriate for ASL although it prevents CS to benefit from the sparsity of images after subtraction.

Srikant Kamesh Iyer^1,2, Tolga Tasdizen^1,2, Ganesh Adluru³, and Edward DiBella^3,4
1Electrical and Computer Engineering, University of Utah, salt lake city, Utah, United States, ²Scientific Computing and Imaging Institute (SCI), University Of Utah, salt lake city, Utah, United States, ³UCAIR, Department of Radiology, University of Utah, salt lake city, Utah, United States, ⁴Department of Bioengineering, University of Utah, salt lake city, Utah, United States

Incorporating priors about the order of signal intensities of pixels to modify the TV constraint has been shown to improve the quality of the images reconstructed from under sampled data. We propose to apply the pixel ordering information by dividing the data into smaller blocks to make the method more robust to motion. Comparisons with TCR and TCR with single block reordering show that that smaller block size help improve the robustness of the reconstruction to motion in the data and the reconstructions match the fully sampled data more faithfully.

Daniel Neumann¹, Nicole Seiberlich², Felix A. Breuer¹, Gregory Lee³, Philipp Ehses^4,5, Jeffrey L. Duerk^2,3, Peter M. Jakob⁶, and Mark A. Griswold^2,3
1Research Center for Magnetic Resonance Bavaria (MRB), Wuerzburg, Germany, ²Biomedical Engineering, Case Western Reserve University, Cleveland, OH, United States, ³Radiology, University Hospitals, Cleveland, OH, United States, ⁴Dept. for Neuroimaging, University Hospital, Tuebingen, Germany, ⁵Max Planck Institute for Biological Cybernetics, Tuebingen, Germany, ⁶Biophysics, University of Wuerzburg, Wuerzburg, Germany

One important requirement of Compressed Sensing is incoherent k-space sampling. This cannot be easily achieved for 2D imaging due to trajectory restrictions. We propose a new sampling scheme using radial projections covering the full square-shaped k-space in combination with blipped gradients of varying amplitude, frequency and phase shift. To show the applicability of the new sampling scheme, highly undersampled in vivo cardiac data has been acquired along the Moët trajectory using a TrueFISP sequence and reconstructed using a combination of through-time non-Cartesian GRAPPA and Compressed Sensing.

Xiaobo Qu¹, Di Guo², Bende Ning¹, Yingkun Hou³, Shuhui Cai¹, and Zhong Chen^1
1Department of Electronic Science, Fujian Key Laboratory of Plasma and Magnetic Resonance, Xiamen University, Xiamen, Fujian, China, ²Department of Communication Engineering, Xiamen University, Xiamen, Fujian, China, ³School of Information Science and Technology, Taishan University, Taian, Shandong, China

Undersampling the k-space data can speed up magnetic resonance imaging (MRI) at the cost of introducing the aliasing artifacts. In this paper, a patch-based directional wavelets(PBDW) is proposed to sparsify the magnetic resonance (MR) image in undersampled MRI reconstruction. First, a guide image is reconstructed from incomplete k-space data with conventional compressed sensing MRI method. Then, a parameter of PBDW, indicating the geometric direction of each image patch, is trained from the guide image and incorporated into the sparsifying transform to provide the sparse representation for the image to be reconstructed. Simulations demonstrate that trained PBDW leads to better edges than the convetional sparse MRI reconstruction methods do.

Jan Aelterman¹, Hiep Luong¹, Bart Goossens¹, Aleksandra Pizurica¹, and Wilfried Philips^1
1TELIN-IPI-IBBT, Ghent University, Ghent, Belgium

It is demonstrated that trajectory optimization is of crucial importance in compressed sensing MRI reconstruction. Variational Bayesian k-space trajectory optimization techniques is then analyzed from the point of view of variable density k-space sampling. It is shown that density compensation is imperative in this framework and we present new and computationally very efficient method to estimate the density compensation function.

Jordan P Hulet^1,2, Ganesh Adluru³, Julio C Facelli¹, Edward DiBella³, and Dennis L Parker^4,5
1Biomedical Informatics, University of Utah, Salt Lake City, Utah, United States, ²Utah Center for Advanced Imaging Research, Salt Lake City, Utah, United States, ³University of Utah, Salt Lake City, Utah, ⁴University of Utah, Salt Lake City, utah, ⁵Utah Center for Advanced Imaging Research, University of Utah, Salt Lake City, Utah, United States

In this work, the reconstruction time required for temporally constrained reconstruction of a large cardiac perfusion data set was reduced by utilizing 12 graphics processing units distributed across 6 different compute nodes. A speedup of over 30x was achieved compared to a 16 core CPU system resulting in a reconstruction time of only 48 seconds.

Jason K Mendes¹, and Dennis L Parker^1
1Radiology - UCAIR, University of Utah, Salt Lake City, Utah, United States

The temporal signal in any given voxel can be approximated by a combination of a finite number of basis functions (or principle components). As we keep fewer principle components, we can increase the amount of data undersampling. This can cause a problem, however, when two voxels exhibit highly correlated signals that differ by only a small delay in time. In such cases, keeping only a small number of principle components can inadvertently change the temporal characteristics of these voxels by matching them to the same principle components. The proposed method helps to address this concern by reducing the number of principle components needed to reconstruct an undersampled data set by allowing time shifted principle components.

David S Smith¹, Ryan Robison^1,2, and E Brian Welch^1
1Institute of Imaging Science, Vanderbilt University, Nashville, TN, United States, ²Philips Healthcare, Cleveland, OH, United States

We present a method using concepts from compressed sensing to automatically detect and eliminate radio frequency spike noise from MRI data sets. The spikes are located by their effect on the total variation of the image. The spikes are then deleted from the full data set, creating a very slightly undersampled data set, which is then reconstructed in a TV-regularized compressed sensing MRI reconstruction. Since the data is almost completely Nyquist sampled, this method introduces no artifacts and produces images with normalized mean square error two orders of magnitude smaller than both zeroing of the spiked data and Fourier interpolation.

Jørgen Avdal¹, Anders Kristoffersen², Asta Håberg³, and Pål Erik Goa^2
1Department of Circulation and Medical Imaging, MI Lab, NTNU, Trondheim, Sør-Trøndelag, Norway, ²MI Lab, Clinic of Radiology, St. Olavs Hospital, Trondheim, Norway, ³Department of Neuroscience, NTNU, Trondheim, Norway

A novel functional MRI imaging and analysis technique employing PRESTO and Compressed Sensing for accelerated frame rate is presented, where the fMRI signal is partially separated from the noise before the reconstruction step. The technique tested on in vivo data results in strong activation signals as well as good contrast between activated and non-activated voxels. Results also show that Compressed Sensing may be used to produce temporal noise variance maps based on undersampled data.

Zhong Chen¹, Changwei Hu¹, Xiaobo Qu¹, Lijun Bao¹, and Shuhui Cai^1
1Department of Electronic Science, Fujian Key Laboratory of Plasma and Magnetic Resonance, Xiamen University, Xiamen, Fujian, China

Undersampling k-space is an effective way to reduce acquisition time in MRI. However, this will introduce significant aliasing artifacts, and blur edges in the reconstructed magnetic resonance image. The edges usually contain important information for the clinical diagnosis. In this work, we propose a method to recover edges from undersampled MRI by incoporating a weighting matrix to the l₁ norm sparsity regularization term. Compared with conventional compressive sensing methods, the proposed method yields better edge recovery, and requires fewer k-space measurements to achieve acceptable reconstruction quality.

Iulius Dragonu¹, Guobin Li¹, Jeff Snyder¹, Jürgen Hennig¹, and Maxim Zaitsev^1
1Dept. of Radiology, Medical Physics, University Medical Center Freiburg, Freiburg, Germany

Compressed Sensing (CS) is a technique that allows accelerating data acquisition in the presence of sparse or compressible signals. This is accomplished by using a pseudo-random undersampling in the phase-encoding direction. The Point Spread Function (PSF) is a fundamental tool allowing the evaluation of the quality of reconstruction and the spatial resolution of images. Previously the concept of PSF approximation was extended to non-linear and non-stationary imaging systems. The PSF has different values in all imaging points due to the non-linearity and non-stationary proprieties of the CS algorithms. In this work, we propose a technique of evaluating the PSF of the CS reconstruction based on an acquisition pattern used in PSF for echo-planar imaging.

Brian L Burns^1,2, Jon K Furuyama³, Neil Wilson³, Nicki DeBruhl³, Alex Bui^2,3, and M. Albert Thomas^3
1Biomedical Engineering, UCLA, Los Angeles, CA, United States, ²Medical Imaging Informatics (MII), UCLA, Los Angeles, CA, United States, ³Department of Radiological Sciences, UCLA

2D Localized-Correlated Spectroscopy (L-COSY) has been shown to be a powerful diagnostic tool in breast cancer tumor classification. However, in order to reduce the scan time to a clinically acceptable level, the number of t1 increments used to construct the indirect spectral dimension must be greatly reduced, sacrificing spectral resolution and reducing its diagnostic potential. By under-sampling the indirect spectral dimension to only 22 t1 lines and reconstructing the fully-sampled spectra using Maximum Entropy image reconstruction, a 2x acceleration of the in vivo L-COSY sequence has been achieved, which shows under-sampled spectra with similar spectral characteristics to fully sampled spectra.

Eric Y. Pierre¹, Nicole Seiberlich¹, Stephen Yutzy², Vikas Gulani³, and Mark A. Griswold^1,3
1Dept. of Biomedical Engineering, Case Western Reserve University, Cleveland, OH, United States, ²Dept. of Radiology, University of Pittsburgh, PA, United States,³Dept. of Radiology, University Hospitals of Cleveland, OH, United States

A general ABSINTHE framework that makes use of prior information to help reduce the amount of data needed to generate an image of a given object is presented. It can be used with any undersampled data reconstruction technique and with any form of prior information compatible with the acquisition scheme. Results are presented for an implementation using SENSE and a database of previously acquired images with identical coil configuration, identical resolution, similar anatomical positioning, and similar image contrast as the signal to reconstruct. A regularization criteria for reconstruction based on a priori knowledge was introduced which enforces sparsity.

Matthew Ethan MacDonald^1,2, David Adair^2,3, Parviz Dolati^2,4, Jerome Yerly^2,3, and Richard Frayne^2,4
1Biomedical Engineering, University of Calgary, Calgary, AB, Canada, ²Seaman Family MR Research Centre, Foothills Medical Centre, Alberta Health Services, Calgary, AB, Canada, ³Electrical and Computer Engineering, University of Calgary, Calgary, AB, Canada, ⁴Radiology and Clinical Neurosciences, University of Calgary, Calgary, AB, Canada

Real time imaging is used to visualize devices during endovascular procedures clinically with x-ray imaging. Magnetic resonance (MR) imaging has been used for applications that require imaging but implementations have been generally limited to 2D or bi-planar systems. In this work we present a hardware implementation and software solution that is capable of performing 3D reconstruction in real time. Imaging is accelerated with random undersampling trajectories and used to visualize endovascular devices and contrast agent inflow in anthropomorphic phantoms.

Yong Pang¹, and Xiaoliang Zhang^1,2
1Radiology and Biomedical Imaging, University of California San Francisco, San Francisco, CA, United States, ²UCSF/UC Berkeley Joint Graduate Group in Bioengineering, San Francisco & Berkeley, CA, United States

Sparse MRI has been introduced to reduce the acquisition time and raw data size by significantly undersampling the k-space. We propose an interpolation method to improve the signal to noise ratio or imaging speed for multi-slice two-dimensional MRI. The raw data of each slice is multiplied by a weighting function and then used to estimate the missed k-space data of the neighboring slice, which helps improve the SNR of the neighboring slice. In-vivo MR of human feet has been used to investigate the feasibility of the proposed method, showing obvious improved SNR of the neighboring slice.

Brian Burns^1,2, Jon K Furuyama³, Neil Wilson⁴, Nicki DeBruhl⁴, and M. Albert Thomas^4
1Department of Radiological Sciences, UCLA, Los Angeles, CA, United States, ²Medical Imaging Informatics (MII), UCLA, ³Radiological Sciences, UCLA,⁴Department of Radiological Sciences, UCLA

The EPCOSI sequence acquires 2-spatial and 2-spectral dimensions in a single scan. The kx and t2 dimensions are acquired simultaneously, but the ky and t1 dimensions are acquired indirectly. By under-sampling the indirect spectral and spatial dimensions and reconstructing with Maximum Entropy, we have shown that a 4x reduction factor can be used to reduce the scan time to a clinically acceptable 5 minutes with equivalent spatial and spectral results.

Sylvain Louis Merlet¹, Michael Paquette², Rachid Deriche¹, and Maxime Descoteaux^2
1Athena Project-Team, INRIA, Sophia Antipolis, Méditerranée, France, ²Computer Science Departement, Université de Sherbrooke, Québec, Canada

Sparsity is one of the key ingredient in Compressed Sensing recovery. In Diffusion MRI, few studies have been proposed to characterize the sparsity of the Ensemble Average Propagator which captures the water diffusion phenomenon. We propose a fair comparison of two classes of representations : The discrete representations, via the Haar, Daubechie-Cohen-Fauveau (DCF) 5-3, DCF 9-7 wavelets bases, and the continuous representations, via Spherical Polar Fourier (SPF) and 3D Simple Harmonic Oscillation Reconstruction and Estimation (SHORE) bases. We study the advantages and disadvantages of these discrete and continuous representations EAP for the first time.

Yu Li¹, Jean Tkach¹, Suhas Kallapur¹, Machiko Ikegami¹, Alan Jobe¹, and Charles L. Dumoulin^1
1Imaging Research Center, Radiology Department, Cincinnati Children's Hospital Medical Center, Cincinnati, Ohio, United States

The presented work addresses a clinical challenge to neonatal MRI on a small bore scanner within the Neonatal Intensive Care Unit (NICU): The parallel imaging acceleration is limited by the restricted spatial encoding capability of the small coil arrays sized to match the neonatal anatomy on the NICU scanner. In this work, we propose an approach to accelerating multi-slice 2D imaging using image-space correlation introduced by naturally existing MRI k-space data sparsity. The proposed approach is demonstrated in an extreme case: Neonatal MRI performed with a single-channel volume birdcage coil that is incapable of parallel imaging.

Christine Gnahm¹, Michael Bock^1,2, Wolfhard Semmler¹, and Armin M. Nagel^1
1Dept. of Medical Physics in Radiology, German Cancer Research Center (DKFZ), Heidelberg, Germany, ²Radiology - Medical Physics, University Hospital Freiburg, Freiburg, Germany

The image quality of ²³Na-MRI can be substantially improved by incorporating basic anatomical information from ¹H-MRI into the reconstruction. An iterative reconstruction algorithm is presented that combines a total variation (TV) constraint with object boundary information derived from ¹H data. The algorithm is evaluated on phantom data and compared to both conventional gridding and TV-regularized iterative reconstruction. With the ¹H constraint, an up to 150% increase in SNR over gridding reconstruction is seen, and spatial resolution is preserved better than with a TV-constraint alone. Furthermore, the image contrast is maintained which is also demonstrated in brain images of a volunteer.

Peng Lai¹, Shreyas S Vasanawala², and Anja C.S Brau^1
1Global Applied Science Laboratory, GE Healthcare, Menlo Park, CA, United States, ²Radiology, Stanford University, Stanford, CA, United States

This work analyzes the source of temporal blurring of the conventional DC removal method for tGRAPPA and ktGRAPPA and presents a modified image sparsification scheme to address the problem. The proposed method automatically identifies dynamic tissue locations and removes DC signals from static tissues only before parallel imaging reconstruction. Our invivo results from cardiac cine MRI demonstrates that this new method can reduce aliasing artifacts and improve SNR for high acceleration tGRAPPA and ktGRAPPA without losing temporal fidelity.

Mei-Lan Chu¹, Jia-Shuo Hsu¹, Hsiao-Wen Chung¹, Shang-Yueh Tsai^2,3, and Yi-Ru Li^4
1Graduate Institute of Biomedical Electronics and Bioinformatics, National Taiwan University, Taipei, Taiwan, ²Graduate Institute of Applied Physics, National Chengchi University, Taipei, Taiwan, ³Research Center for Mind, Brain and Learning, National Chengchi University, Taipei, Taiwan, ⁴Department of Electronic Engineering, National Taiwan University of Science and Technolo, Taipei, Taiwan

Phase-correlation motion estimation (ME) and motion compensation (MC), which is an essential part of video compression technique, has been successfully applied to reconstructing under-sampled cine cardiac imaging. For dynamic MR imaging outside of the cardiac region, such as dynamic contrast-enhanced (DCE) perfusion lung imaging, higher temporal resolution for wider slice coverage is still highly desirable. In this abstract, we demonstrate that the drawbacks of temporal-smoothing and baseline overshoot can be overcome by using phase-correlation ME / MC with 2-fold to 6-fold acceleration. The experimental results show that the proposed method successfully reconstructs full-resolution dynamic frames at substantially reduced acquisition data without the disadvantages of overshooting in the initial time frames and the undesired smoothing effects in the presence of abrupt temporal variations.

Armin M. Nagel¹, Marc-André Weber², Maya B. Wolf^2,3, and Wolfhard Semmler^1
1Dept. of Medical Physics in Radiology, German Cancer Research Center (DKFZ), Heidelberg, Germany, ²University Hospital of Heidelberg, Heidelberg, Germany,³Dept. of Radiology, German Cancer Research Center (DKFZ), Heidelberg, Germany

This study employed a 3-D density adapted projection reconstruction sequence, which samples a rectangular cuboid. It was found to attain a full anisotropic resolution and FOV. ²³Na-MRI of the human calf muscles and knee cartilage was conducted using a 7-Tesla whole body MR system with high in-plane resolution. The newly implemented sequence allowed for anisotropic FOVs and resolutions. All in all, it provided a better performance than spherical k-space sampling.

Gregory R. Lee^1,2, and Mark A. Griswold^1,2
1Radiology, Case Western Reserve University, Cleveland, OH, United States, ²University Hospitals Case Medical Center, Cleveland, OH, United States

In dynamic imaging applications it is desirable to order the projections of a 3D radial acquisition so that they are approximately equally distributed over k-space at any time scale. The authors propose a numerical algorithm based on electrostatic charge repulsion with modifications to encourage uniform sample spacing at various time scales. In comparison to an existing 3D golden angle approach, the proposed method results in improved sampling uniformity over the full set of projections. While the golden angle approach is only compatible with linear projections, the proposed technique also applies when multiple radial projections are acquired in a single shot.

Sana Vaziri¹, and Michael Lustig^1
1Electrical Engineering and Computer Sciences, UC Berkeley, Berkeley, CA, United States

We present a method to compute the fastest possible gradient waveforms for a given k-space trajectory. In our design, we exploit the fact that each gradient set has its own constraints. The resulting trajectory is therefore not rotationally invariant and a redesign is needed when rotated. However, our algorithm is fast and non-iterative and can compute waveforms on-the-fly. Results of 5-10% waveform time reduction are presented for spiral and circular trajectories.

Haisam Islam¹, and Gary Glover^2
1Bioengineering, Stanford University, Stanford, CA, United States, ²Radiology, Neurosciences, Biophysics, Stanford University, Stanford, CA

We propose a new fMRI method for doubling the temporal resolution by simultaneously exciting pairs of adjacent slices in quadrature, and reconstructing them based on their phase distributions. The phase distributions are collected during a calibration scan, in which slices are individually excited with the same phase as in the quadrature excitation. The results show that the method is accurate in uniform brain regions, but fails in regions near large magnetic susceptibilities due to off-resonance and greater signal dropout. The effects of slice thickness, echo time, and number of simultaneously excited slices on reconstruction accuracy still need to be investigated.

Noam Ben-Eliezer¹, Lucio Frydman², and Daniel K Sodickson^1
1Center for Biomedical Imaging, New-York University Medical Center, New-York, NY, United States, ²Chemical Physics, Weizmann Institute of Science, Rehovot, Israel

In recent years a conceptually different approach for collecting MR data has emerged, based on progressive Spatiotemporal-Encoding (SPEN) of the magnetic spins. This approach extends the capabilities MRI, allowing users to overcome sizable macro- and microscopic field inhomogeneities by altogether freeing the spins’ evolution from T₂^* effects, while owing to its lack of aliasing enables the use of limited FOVs. This work demonstrates the potential of SPEN for imaging at ultra high spatial resolutions – a regime that is hard to realize using standard k-space encoded Gradient-Echo or Spin–Echo protocols owing to T₂^* related broadening of their acquisition point-spread-functions.

Ethan M Johnson¹, and John M Pauly^1
1Electrical Engineering, Stanford University, Stanford, CA, United States

Reconstruction techniques for data encoded by nonlinear, non-bijective magnetic fields have thus far used receive coil array sensitivity maps to resolve the non-bijective-encoding ambiguity. However, if the acquired signals are compared to a pair of calibration signals encoded by a bijective field, the ambiguity can be resolved. From this, a practical autocalibrating method that does not require sensitivity maps can be identified, and an image reconstructed.

Leo K Tam¹, Jason P Stockmann¹, Gigi Galiana², Dana C Peters², and Robert Todd Constable^1,2
1Biomedical Engineering, Yale University, New Haven, CT, United States, ²Diagnostic Radiology, Yale University, New Haven, CT, United States

Recent nonlinear encoding strategies have provided robust parallel imaging strategies using highly undersampled datasets. Here, we introduce an iterative optimization of the spatial basis in order to achieve efficient encoding and preserve smoothness (reduced dB/dt). To focus the optimization landscape, a library is constructed from the achievable phase profiles from the available gradient strength and phasor combinations. The search prioritizes orthogonality between encoding bases as well as smoothness in transition. Preliminary results with a second order spherical harmonic gradient system suggest significant reductions in dB/dt as compared with parallel echo planar imaging.

Leo K Tam¹, Jason P Stockmann¹, Gigi Galiana², Dana C Peters², and R. Todd Constable^2
1Biomedical Engineering, Yale University, New Haven, CT, United States, ²Diagnostic Radiology, Yale University, New Haven, CT, United States

Recent nonlinear gradient encoding strategies have demonstrated highly accelerated parallel imaging for reconstruction of a single slice. In the present study, we explore the unique 3D spatial localization available to spherical harmonic encoding gradients. To design the encoding gradients, the Null Space Imaging (NSI) strategy is used to create readout gradients complementary to the receiver coil sensitivities. As a first approach to nonlinear gradient projection imaging, the proposed method demonstrates the benefits of coil complementarity and reduced dB/dt extend to volume imaging.

Jakob Assländer¹, Jason Stockmann², Martin Blaimer³, Felix Breuer³, and Maxim Zaitsev^1
1Dep. of Radiology, Medical Physics, University Medical Center, Freiburg, Baden-Württemberg, Germany, ²Biomedical Engineering, Yale University, New Haven, Connecticut, United States, ³Magentic Resonace Bavaria, Würzburg, Bavaria, Germany

VSOP COGNAC is a acquisition strategy, that reduces the FOV by combining phase encoding with a z2-gradient and sinusoidal gradient pulses on the Cartesian gradients. This acquisition scheme intrinsically dephases the spin ensemble at the center of the FOV, thus a reduction of the phase encoding steps can be achieved without violating the Nyquist criterion. This potentially allows faster imaging e.g. of the cortex, when the center of the FOV is of no interest.

Jason P Stockmann¹, Gigi Galiana², and R. Todd Constable^1,2
1Biomedical Engineering, Yale University, New Haven, CT, United States, ²Diagnostic Radiology, Yale University, New Haven, CT, United States

Quadratic phase has been shown to permit spatial localization by "phase scrambling" signals from outside a given region of interest. This has been applied in the through-slice direction using quadratic-phase RF pulses as well as in-plane using nonlinear spatial encoding magnetic fields. In this work, we use a quadratic-phase RF pulse to precompensate the through-plane phase applied by a 3-D magnetic encoding field, permitting acquisition of the target region-of-interest using a Z2 spherical harmonic field. We further show that GradLoc can be combined with parallel imaging to enable highly-accelerated acquisitions of the region of interest without aliasing.

Walter R.T. Witschey^1,2, Sebastian Littin³, Daniel Gallichan³, Christian A. Cocosco³, Anna Masako Welz³, Hans Weber³, Gerrit Schultz³, Andrew Dewdney³, Felix Wehrli¹, Juergen Hennig³, and Maxim Zaitsev^4
1Department of Radiology, University of Pennsylvania, Philadelphia, PA, United States, ²Department of Surgery, University of Pennsylvania, Philadelphia, PA, United States, ³Department of Radiology, University Medical Center Freiburg, Freiburg, Germany, ⁴Department of Radiology, University Medical Center Freiburg, Freiburg, United States

Spatial selection is achieved by spoiling the MR signal at the boundaries of anatomical shapes. This controlled spoiling is achieved here using a new 3-channel, flat gradient coil integrated with the patient table.

Kelvin J. Layton^1,2, Daniel Gallichan³, Chris A. Cocosco⁴, Anna M. Welz³, Frederik Testud³, Christoph Barmet⁵, Klaas Prüssmann⁵, and Maxim Zaitsev^3
1Electrical & Electronic Engineering, University of Melbourne, Melbourne, Victoria, Australia, ²National ICT Australia, Melbourne, Victoria, Australia, ³Dept. of Radiology, Medical Physics, University Medical Center Freiburg, Freiburg, Germany, ⁴Radiology, Medical Physics, University Medical Center Freiburg, Freiburg, Germany, ⁵Biomedical Engineering, University & ETH Zürich, Zürich, Switzerland

Nonlinear encoding gradients inherently produce MR images with a spatially varying resolution. In this work, we propose a method for the automated design of trajectories that improve the resolution in a target region-of-interest. Our technique finds the optimal trajectory satisfying safety and hardware constraints given an arbitrary configuration of encoding fields. In particular, we design single-shot trajectories that use a combination of linear and quadrupolar encoding fields to provide a localised two-fold improvement in resolution. The trajectories are demonstrated using simulated and in-vivo data.

Anna-Katinka Bracher¹, Daniel Kammer¹, Erich Hell², Johannes Ulrici², and Volker Rasche^1
1Department of Internal Medicine II, University Hospital of Ulm, Ulm, BW, Germany, ²Sirona Dental Systems GmbH, Bensheim, Germany

In UTE imaging the k-space center is acquired during ramp-up of the read-out gradient. This leads to a quadratic T2 decay during acquisition of k-space center which effects the image quality. By applying a small prephasing gradient the k-space center is sampled at a higher read-out gradient and hence much faster than in the conventional UTE technique. The faster sampling causes less T2 decay during k-center sampling with a concomitant reduction of image blur due to T2 and off-resonance effects.

Ali Caglar Ozen^1,2, Koray Ertan¹, and Ergin Atalar^1,2
1Electrical and Electronics Engineering, Bilkent University, ANKARA, Turkey, ²UMRAM, Bilkent University, ANKARA, Turkey

Concurrent RF transmission and signal reception is significant for applications including imaging species with ultra short T2 values, and measurement of spin properties during excitation. In this study, magnetic field decoupling of 75dB between transmit coils and the receive coil is accomplished using a transmit array system. In order to detect concurrent MR signals, linearly polarized electromagnetic fields are utilized to generate a plane with zero perpendicular magnetic field components coplanar with the receiver coil.

Markus Weiger^1,2, David Otto Brunner³, Benjamin Emanuel Dietrich³, Colin Felix Müller³, and Klaas Paul Pruessmann^3
1Bruker BioSpin AG, Faellanden, Switzerland, ²Bruker BioSpin MRI GmbH, Ettlingen, Germany, ³Institute for Biomedical Engineering, University and ETH Zurich, Zurich, Switzerland

MRI with zero echo time and hard-pulse excitation (ZTE) enables efficient, silent, and robust 3D imaging of short-T2 samples. Data obtained with this scheme are slightly incomplete in the k-space centre due to the initial dead time that is caused by the finite length of the RF pulse, transmit-receive (T/R) switching, and digital filtering. In the present work, we report the first implementation of the ZTE approach for human applications, relying on rapid T/R switching in the µs range.

Axel J. Krafft¹, Ruitian Song¹, Ralf B. Loeffler¹, Matthew D. Robson², and Claudia M. Hillenbrand^1
1Radiological Sciences, St. Jude Children's Research Hospital, Memphis, TN, United States, ²MRS Unit, Oxford University Centre for Clinical Magnetic Resonance Research, Oxford, United Kingdom

2D UTE imaging employs half pulse excitation and radial sampling to realize ultra-short echo times and allows for detection of very short T2* species. However, the half pulse slice profile is very sensitive to gradient imperfections and the imaging steady state so that accurate T2* measurements based on 2D UTE imaging remains challenging. Here, a recently presented concept using out-of-slice saturation to suppress unwanted background signals in 2D UTE imaging is extended with a technique for preservation of the imaging steady state. The sequence could be demonstrated as a simple and robust approach for improved 2D UTE slice selectivity.

Suchandrima Banerjee¹, Atsushi Takahashi¹, Yuval Zur², Ajit Shankaranarayanan¹, and Douglas A.C. Kelley^3
1Global Applied Science Lab, GE Healthcare, Menlo Park, California, United States, ²GE Healthcare, Tirat Carmel, Haifa, Israel, ³Global Applied Science Lab, GE Healthcare, San Francisco, California, United States

Multiband echo planar imaging (EPI) is an important enabler for achieving high temporal and spatial resolution for brain connectivity studies. Determination of the slice separation coefficients is generally performed using single slice acquisitions. However, at high fields such as 7 Tesla, care must be taken to avoid errors from geometric distortion and susceptibility effects. In this work we propose a calibration method which uses equivalent modulation schemes for both the calibration and the actual acquisition to guarantee that off-resonance and other effects present in the actual acquisition, are also similarly present in the calibration and thereby compensated.

Lars Kasper^1,2, Maximilian Haeberlin¹, Bertram Jakob Wilm¹, Klaas Enno Stephan^2,3, and Klaas Paul Pruessmann^1
1ETH/UZH Zurich, Institute for Biomedical Engineering, Zurich, Zurich, Switzerland, ²University of Zurich, Laboratory for Social and Neural Systems Research, Zurich, Zurich, Switzerland, ³University College London, Wellcome Trust Centre for Neuroimaging, London, London, United Kingdom

MR images are commonly processed after image reconstruction, e.g. to suppress Gibbs ringing artifacts. Following the matched filter theorem, we optimize a regular gradient echo acquisition to maximize the signal-to-noise ratio (SNR) in the final, smoothed images. Our approach enables a 2D-matching of the acquisition to the desired smoothing filter by combining variable phase encoding line spacing with gradient amplitude modulation in readout direction. This doubles the benefits of a matched filter acquisition compared to sole 1D-matching of the desired smoothing kernels yielding SNR increases of up to 50 %.

Andrew J Wheaton^1
1Toshiba Medical Research Institute USA, Mayfield Village, OH, United States

Prevailing Dixon-based fat-water separation methods use multiple images acquired at different echo times. In FSE-based implementations, the requirement of a range of echo times restricts its use to relatively long echo spacings. Moreover, since the data are acquired at different echo times, time-dependent sources of background phase (i.e. eddies) are not uniform in each dataset. This study proposes using a binomial prepulse to evolve fat and water isochromats. The prepulse can be coupled to any readout, including short echo-space FSE (i.e. SPACE, CUBE, HASTE). The feasibility of using such a scheme is demonstrated in short-echo space 3D FSE.

Mathias Engström¹, Roland Bammer², and Stefan Skare^1
1Clinical Neuroscience, Karolinska Institute, Stockholm, Sweden, ²Radiological Sciences Laboratory, Stanford University, Palo Alto, CA, United States

Phase navigated multi-slab acquisitions has been proposed as an alternative way of doing 3D diffusion weighted imaging. Combining multi-slab readout with Stejskal-Tanner diffusion preparation requires sequence modifications, intricate phase correction and slab profile optimizations in order to work. Here we present methods and modifications improving the outcome of 3D multi-slab DWI.

Morwan Choli¹, Martin Blaimer¹, and Peter M. Jakob^1,2
1Research Center Magnetic Resonance Bavaria e.V (MRB), Wuerzburg, Germany, ²Department of Experimental Physics 5, University of Würzburg, Wuerzburg, Germany

The TSE-EPI-CAT hybrid approach integrates TSE and EPI-modules in a sequential fashion and is due to its low SAR an interesting candidate for high-field MRI. The free choice of the λ-factor describing the relative size of the TSE and EPI-modules was so far limited for arbitrary T2-contrast imaging. The proposed technique overcomes this limitation by using a combination of T2-preparation and TSE-EPI-CAT with the potential of significantly reduced λ and thus SAR. T2-weighted head images obtained with a SAR reduction up to 55% compared to a conventional TSE sequence show no significant loss of image quality with increasing EPI-module size.

Hoonjae Lee¹, and Jaeseok Park^1
1Department of Brain and Cognitive Engineering, Korea University, Seoul, Korea

Fluid-attenuated inversion recovery (FLAIR) imaging (1), which employs long inversion recovery (IR) magnetization preparation to selectively nullify cerebrospinal fluid (CSF) signals while retaining a clinically useful T2-weighted contrast, has been widely used for lesion detection in brain. However, since the long time of IR (~ 2000 ms) is required for CSF suppression due to its long T1-relaxation time (~ 4500 ms), imaging time in conventional FLAIR is prohibitively prolonged and spatial resolution is substantially limited. Hence, it is challenging to achieve high-resolution isotropic whole-brain fluid-attenuated imaging for detecting very small lesions in a clinically acceptable imaging time. To resolve this problem, in this work we develop a highly efficient rapid combo acquisition technique for sub-millimeter isotropic FLAIR, wherein instead of long IR preparation rapid gradient echo (PSIF) data (2) is acquired in the periphery of k-space while turbo/fast spin echo (TSE) (3) is encoded in the central region of k-space, simultaneously collecting both low (TSE) and high (PSIF) spatial frequency signals in the ky and kz directions and thus speeding up imaging time.

Alto Stemmer¹, and Berthold Kiefer^1
1Healthcare Sector, Siemens AG, Erlangen, Germany

In this work a new Turbo Spin Echo (TSE) sequence is introduced, which refocuses m adjacent excited slices simultaneously and thereby reduces SAR by approximately a factor m. Contrary to earlier simultaneously refocused TSE sequences the new sequence utilizes the signal of all pathways and is able to sustain a long echo train required for T2 weighted imaging.

Hyunyeol Lee¹, Suhyung Park¹, and Jaeseok Park^1
1Brain and Cognitive Engineering, Korea University, Seoul, Korea

As an alternative to time-consuming conventional DIR imaging, a highly efficient pulse sequence is developed for high-resolution isotropic whole-brain gray matter (GM) imaging, named DUal-Echo Single-Slab 3D Turbo Spin Echo (DUESS-TSE). Without long inversion recovery preparation, the proposed method generates sub-millimeter isotropic whole-brain GM images within a clinically acceptable imaging time.

Junqian Xu¹, Steen Moeller¹, John Strupp¹, Edward J Auerbach¹, Liyong Chen^2,3, David A Feinberg^2,3, Kamil Ugurbil¹, and Essa Yacoub^1
1Center for Magnetic Resonance Research, University of Minnesota, Minneapolis, MN, United States, ²Helen Wills Neuroscience Institute, University of California, Berkeley, CA, United States, ³Advanced MRI Technologies, Sebastopol, CA, United States

Using a blipped-controlled-aliasing multiband EPI sequence with k-space center (k₀) alignment (ABC-MB EPI), up to 12 simultaneously excited slices at 3T using a 32-channel head coil were achieved. The k-space center (k₀) alignment strategy significantly improved image quality for high MB factors compared to using only a balanced blip approach. Spectral analysis of highly accelerated data revealed un-aliased sampling of physiological fluctuations, demonstrating the robustness and utility of the technique, which can be applied to any EPI based application, including fMRI, diffusion, or perfusion imaging.

Daniel Posfai¹, J. Andrew Derbyshire², and Daniel A. Herzka^1
1Department of Biomedical Engineering, Johns Hopkins School of Medicine, Baltimore, Maryland, United States, ²Tornado Medical Systems, Toronto, Ontario, Canada

Time-efficient gradient waveforms with pre-defined 1st moments (M1) can be difficult to design. These gradients are at the core of many techniques such as phase contrast or flow-compensated acquisitions. A new method of designing hardware-optimized gradient waveforms is presented. The design strategy works in both logical and physical coordinate systems, pushing gradient hardware to limitations. The method works directly on a rasterized time scale, and yields complete control of the 0th (M0), or M0 and M1 of waveforms. Design of an efficient balanced steady-state free precession (bSSFP) sequence is demonstrated, along with partially and fully M1-nulled (M1=0 mT/m*us2) variants useful for flow-compensated bSSFP.

Jon-Fredrik Nielsen¹, Daehyun Yoon², and Douglas C Noll^1
1Biomedical Engineering, University of Michigan, Ann Arbor, Michigan, United States, ²Electrical Engineering, University of Michigan, Ann Arbor, Michigan, United States

We introduce a new spoiled steady-state sequence that incorporates a non-slice-selective tailored RF pulse after each data readout segment. This "tip-up", or "fast recovery", pulse is designed to tip the spins in the desired imaging region back toward the longitudinal axis. Out-of-slice signal is removed using RF-spoiling. Under ideal conditions, the proposed sequence produces images that are similar to balanced SSFP, but without banding artifacts or transient fluctuations. We assess the feasibility of using this sequence for T2/T1-weighted brain imaging, using a short spiral tip-up pulse and a conventional quadrature transmit/receive headcoil.

Steve Patterson^1,2, Erin Mazerolle^1,2, Steven Beyea^1,2, and Chris Bowen^1,2
1Institute for Biodiagnostics (Atlantic), National Research Council Canada, Halifax, NS, Canada, ²Dalhousie University, Halifax, NS, Canada

In this study we demonstrate whole-brain SSFP fMRI at 3T in a rat model from a single run of the functional paradigm (hypercapnia challenge). Banding artifact-free images were generated by combining two images with complementary RF phase cycling increments. These image pairs were collected sequentially in time, by alternating the RF phase cycling increment between images, and using RF catalyzation to minimize signal fluctuations. The sensitivity of this alternating SSFP (altSSFP) technique was compared to conventional pbSSFP in pbSSFP high-signal (pass-band) and low-signal (stop-band) regions. Results show comparable sensitivity in pass-band regions and significantly enhanced sensitivity in stop-band regions.

Kenneth O. Johnson¹, and Craig H Meyer^1
1Biomedical Engineering, University of Virginia, Charlottesville, VA, United States

Interferometry is a powerful technique that is capable of accelerating imaging methods. Interferometry is applied to MR spectroscopic data, reducing the amount of necessary data. This abstract analyzes the information sources that provide this acceleration. These sources are dependence on spectral sparsity and phase ambiguity. Understanding the sources of information clarifies the algorithm and its limitations.

Andreas J Hopfgartner¹, David M Grodzki², Julian Boldt³, Kurt Rottner³, Ernst Jürgen Richter³, and Peter Michael Jakob^1
1Experimental Physics 5, University of Würzburg, Würzburg, Germany, ²Siemens, Erlangen, Germany, ³Prosthodontics, Dental School, University of Würzburg, Würzburg, Germany

Dental imaging can offer new applications for MRI. Without ionizing radiation, carious lesions and the shape of teeth can be detected with MRI, as it has recently been shown in a couple of publications. The T2 relaxation in rigid bodies is very short (enamel 14-240µs and dentine 12µs-1ms, at 1.5T). In this work, an ultra-short echo time sequence (PETRA) is investigated to characterize its performance in extracted teeth at 1.5T . An in-vivo measurement is presented as well.

Holger Eggers¹, Marko K Ivancevic², Gillian M Newstead³, and Gwenael Herigault^4
1Philips Research, Hamburg, Germany, ²Philips Healthcare, Cleveland, OH, United States, ³University of Chicago Medical Center, Chicago, IL, United States,⁴Philips Healthcare, Best, Netherlands

Dixon methods have been shown to improve the homogeneity and the extent of fat suppression in breast imaging. However, the traditional choice of in- and out-of-phase echo times leads to restrictions on the achievable spatiotemporal resolution. In this work, a recently proposed dual-echo Dixon method with flexible echo times is employed to overcome these restrictions. It is demonstrated to permit bilateral breast imaging with sub-millimeter resolution in less than a minute at 3 T. While an accurate modeling of the fat signal is found to be crucial, a fixed seven-peak spectral model is shown to be sufficient for an excellent fat suppression.

Valentina Taviani¹, Sidharta Nagala², Andrew N. Priest³, Mary A. McLean⁴, and Martin J. Graves^3
1Department of Radiology, University of Wisconsin, Madison, WI, United States, ²Department of Otolaryngology, Addenbrooke’s Hospital, Cambridge, United Kingdom, ³Department of Radiology, University of Cambridge, Addenbrooke's Hospital, Cambridge, United Kingdom, ⁴Cancer Research UK Cambridge Research Institute, Li Ka Shing Centre, Cambridge, United Kingdom

Diffusion-weighted (DW) echo planar imaging (EPI) in the thyroid can benefit from reduced field of view (r-FOV) techniques to reduce off-resonance-induced distortion and blurring due to T₂* decay. In this work, we used a 2D spatially selective RF pulse integrated into a dual spin echo DW single-shot (SS) EPI pulse sequence to perform contiguous, multi-slice, fat-suppressed, r-FOV, DW imaging of the thyroid gland at 3T. The r-FOV technique was compared with the conventional full-FOV DW-SS-EPI and the beneficial effects of an optimal B₁ reconstruction strategy on the measured apparent diffusion coefficients were investigated in phantom experiments and healthy volunteers.

Daniel P. Fiege¹, Sandro Romanzetti¹, Fernando E. Boada², Stephen R. Yutzy², Yongxian Qian², Jörg Felder¹, and N. Jon Shah^1,3
1Institute of Neuroscience and Medicine - 4, Forschungszentrum Jülich, Jülich, Germany, ²Radiology and Bioengineering, University of Pittsburgh, Pittsburgh, PA, United States, ³JARA - Faculty of Medicine, RWTH Aachen University, Aachen, Germany

Sodium levels are tightly regulated in the human body. Therefore monitoring total sodium and triple-quantum filtered sodium (as a measure of intracellular sodium) are of great interest for a variety of diseases of the central nervous system. A sequence that enables acquisition of tissue sodium concentration weighted and triple-quantum filtered sodium images in the same run is presented, using two different TPI readouts. Phantom and in vivo results from 9.4T show very good resolution and SNR.

Misung Han¹, Ko Chiba^1,2, Suchandrima Banerjee³, and Roland Krug^1
1Radiology and Biomedical Imaging, University of California - San Francisco, San Francisco, CA, United States, ²Orthopaedic Surgery, Nagasaki University School of Medicine, Nagasaki, Japan, ³GE Healthcare, Applied Science Lab, Menlo Park, CA, United States

High-resolution imaging of the central body regions is usually hampered by long scan time in order to acquire the complete data without aliasing. High-resolution trabecular bone MRI allows assessment of trabecular bone structure in the context of osteoporosis in vivo, but the application to the proximal femur has been limited due to its deep-seated location. In this work, we propose a high-resolution imaging technique by combining variable flip-angle 3D fast spin-echo sequence and outer volume suppression at 3T. Ex-vivo and in-vivo experiments demonstrate that our technique can depict trabecular structure at the proximal femur reliably without visible aliasing artifacts.

Frank Riemer¹, Bhavana Solanky¹, Matthew Clemence², Xavier Golay³, and Claudia AM Wheeler-Kingshott^1
1NMR Research Unit, Department of Neuroinflammation, Institute of Neurology, London, United Kingdom, ²Philips Clinical Science Group, Philips Healthcare, Guildford, United Kingdom, ³Department of Brain Repair and Rehabilitation, UCL Institute of Neurology, London, United Kingdom

High field MRI,new pulse sequences and improvements in non-cartesian sampled k-space reconstruction has drawn attention to Sodium MRI, a technique directly susceptible to most pathological changes in the brain due to the tight regulation of intra- and extracellular 23Na gradient of which a change alters the sodium signal. It could give new insight into diseases such as Multiple Sclerosis, were disease progression has been postulated to be associated with rise and reduction if intracellular sodium content. In this work, we compare 3 common pulse sequences for SNR and resolution in the brain; as previous studies have focused on phantoms with different properties such as absence of short T2 blurring.

Mo Kadbi¹, Hui Wang¹, Lizette Warner², Melanie Traughber², Motaz Alshaher¹, Andrea Yancey³, Jens Heidenreich¹, and Amir A Amini^1
1University of Louisville, Louisville, Kentucky, United States, ²Philips healthcare, Cleveland, Ohio, United States, ³VA hospital, Louisville, Kentucky, United States

UTE phase contrast may be a more accurate technique for measurment of blood flow with higher Reynolds numbers. Additional potential application areas for UTE PC MRI is measurement of blood flow in the presence of atherosclerotic disease where there is turbulent and disturbed flow distal to a stenosis. UTE can remedy intravoxel dephasing which leads to signal loss and reduce errors in velocity measurment . These delays can result in phase errors that may affect the accuracy and reliability of flow measurements. In this work we present a UTE-PC sequence to measure the disturbed blood flows and show that correction of these phase errors improves the result of UTE-PC technique dramatically. Here, the proposed sequence is utilized to quantitatively measure blood velocity in the carotid bifurcation which is associated with disturbed blood flow and results are compared to standard PC MRI sequences.

Atsushi M Takahashi¹, Suchandrima Banerjee¹, Yuval Zur², Robert F Dougherty³, Aviv Mezer⁴, Jason D Yeatman⁴, Brian A Wandell^3,4, Douglas Kelley¹, and Ajit Shankaranarayanan^1
1Applied Science Laboratory, GE Healthcare, Menlo Park, California, United States, ²GE Healthcare, Tirat Carmel, Haifa, Israel, ³Cognitive and Neurobiological Imaging, Stanford University, Stanford, California, United States, ⁴Psychology, Stanford University, Stanford, California, United States

Multi-Band excitation pulses and parallel imaging reconstruction methods are applied to making quantitative measurements of the NMR properties of tissue. Specifically, in this study, T1 measurements of brain were made using multi-band excitation pulses for decreasing acquisition time.

José P. Marques^1,2, Tobias Kober¹, Florent Eggenschwiler¹, Tobias Kober^1,3, and Rolf Gruetter^1,4
1CIBM, EPFL, Lausanne, Vaud, Switzerland, ²CIBM, University of Lausanne, Lausanne, Vaud, Switzerland, ³Advanced Clinical Imaging Technology, Siemens Medical Solutions, Lausanne, Vaud, Switzerland, ⁴CIBM, University of Lausanne and Geneva, Switzerland

In this work we evaluate the improvement of jointly estimating T1 and B1 maps using the MP2RAGE and Sa2RAGE sequences.

Ranjit Singh Hira¹, Usha Sinha¹, shantanu Sinha², and Todd Pawlicki^3
1Physics, San Diego State University, San Diego, CA, United States, ²Radiology, University of California San Diego, San Diego, CA, United States, ³Radiation Therapy, University of California San Diego, San Diego, CA, United States

Gel based MR relaxometry provides accurate 3D dosimetric mapping. The routinely used multi spin echo (MSE) sequence has limitations. Our aim was to evaluate a stepped TE segmented EPI sequence for 3D dose monitoring at 1.5T and 3T. The EPI image quality at 1.5T was comparable to the MSE sequence. The dose calibration factor evaluated with the EPI sequences was comparable at 1.5T and 3T and close to the MSE sequence. Acquisition time of EPI is less than half of MSE sequence with more flexibility in choice of TEs .The study establishes EPI as a novel sequence for dose mapping.

Klaus H. M. Möllenhoff^1,2, Fabian Keil¹, and N. Jon Shah^1,3
1Institute of Neuroscience and Medicine - 4, Forschungszentrum Jülich, Jülich, Deutschland, Germany, ²Department of Cognitive Neuroscience, University Maastricht, Maastricht, Netherlands, ³JARA - Faculty of Medicine, RWTH Aachen University, Aachen, Germany

T1 relaxation times are shown to vary in various diseases and thus fast and accurate mapping of relaxation times is of clinical interest. Here a 3D method, which gives rise to more accurate results due to a higher SNR, based on the 2D TAPIR approach is presented.

Frank Riemer¹, Bhavana Solanky¹, Matthew Clemence², Claudia AM Wheeler-Kingshott¹, and Xavier Golay^3
1NMR Research Unit, Department of Neuroinflammation, Institute of Neurology, London, United Kingdom, ²Philips Clinical Science Group, Philips Healthcare, Guildford, United Kingdom, ³Department of Brain Repair and Rehabilitation, UCL Institute of Neurology, London, United Kingdom

Diseases and pathological changes such as Multiple Sclerosis (MS) or stroke involve changes in the sodium gradient across the brain, which can be observed with Sodium MRI. However due to its low abundance and reduced resonant frequency compared to 1H, spatial resolution is limited. In addition, the relaxation properties are different due to 23Na’s 3/2 spin and quadrupole dominated interactions. In particular, changes of the intra- to extracellular sodium gradient are of interest to be observed but involve toxic chemical shift reagents or use low resolution multiple quantum filtering. The T2 properties of 23Na could give insight into the different sodium compartments and in this work we measured the components of the bi-exponential short and long T2* decay and present the first high SNR short T2*-map of the brain.

Heiko Schmiedeskamp¹, Matus Straka¹, Thomas Christen¹, Greg Zaharchuk¹, and Roland Bammer^1
1Department of Radiology, Stanford University, Stanford, CA, United States

Precision and accuracy of R2 and R2* estimation with spin- and gradient-echo (SAGE) EPI was assessed using 3 different processing methods.

Axel Hartwig¹, Mathias Engström¹, and Stefan Skare^1
1Clinical Neurosience, Karolinska Institute, Stockholm, Sweden

In post contrast T1w SE imaging, spectrally selective RF pre-pulses are normally used to achieve fat suppression, which increases SAR and reduces the number of slices per TR. This work evaluates two alternative methods for fat saturation that do not involve a pre-pulse. Both methods, using either different transmit bandwidths and/or Slice-Selection Gradient Reversal (SSGR), were found to perform equally well or better w.r.t. fat suppression, acquisition time and T1-w contrast compared to two commercially available spectrally selective pre-pulses.

Luca Y Li¹, Cheryl R McCreary^2,3, Fiona Costello^4,5, and Richard Frayne^2,3
1Seaman Family MR Research Centre, Calgary, Alberta, Canada, ²Radiology, University of Calgary, Calgary, Alberta, Canada, ³Seaman Family MR Research Centre, Foothills Medical Centre, Calgary, Alberta, Canada, ⁴Clinical Neurosciences, University of Calgary, Calgary, Alberta, Canada, ⁵Surgery (Opthalmology), University of Calgary, Calgary, Alberta, Canada

Accumulation of iron in deep grey matter structures has been shown to correlate with physical and cognitive and dysfunction in multiple sclerosis. MR susceptibility imaging could provide an objective surrogate marker for accumulation of iron. We compared normalized signal, SNR and CNR of the caudate, putamen, globus pallidus, and thalamus between T2* gradient echo and susceptibility-weighted angiography (SWAN) sequences on our 3 T scanner. Eleven subjects with multiple sclerosis and 5 healthy controls participated. Results suggest that although the SNR was lower for T2* GRE it may be better for detection of changes because it provides better CNR and lower variance in SNR than SWAN.

Sinyeob Ahn¹, Daniel Huddleston², Xiangchuan Chen¹, Govind Bhagavatheeshwaran¹, and Xiaoping Hu^1
1Biomedical Engineering, Georgia Institute of Technology/Emory University, Atlanta, GA, United States, ²Neurology, Emory University, Atlanta, GA, United States

Neuromelanin pigment in locus coeruleus (LC) contributes to generation of image contrast. It decreases as neurodegeneration progresses in some pathological conditions such as Parkinson’s disease. Neuromelanin-sensitive turbo spin echo (TSE) imaging has been used to generate neuromelanin contrast in recent years. TSE imaging with magnetization-transfer contrast (MTC) has been used to increase paramagnetic contrast. However, TSE imaging with MTC increases specific absorption rate (SAR). In this study, T1-weighted gradient echo imaging with MTC is examined for imaging LC of 5 participants while a reduced SAR is sought.

Edzer Lienson Wu¹, Yun-An Huang², Tzi-Dar chiueh², and Jyh-Horng Chen^2
1Biomedical Engineering, National Taiwan University, Taipei, Taiwan, Taiwan, ²electrical engineering, National Taiwan University, Taipei, Taiwan

 

Matthew V Gettemy¹, Amanda R Gearhart², Jeffrey Vesek³, David R Little¹, Jianli Wang³, Joeseph P Stitt², Rick O Gilmore⁴, Anna S Engels^2,4, Qing X Yang³, and Susan K Lemieux^2
1Physics, Penn State, University Park, PA, United States, ²Social, Life, & Engineering Sciences Imaging Center, Penn State, University Park, PA, United States,³Center for NMR Research, Penn State Hershey Medical School, Hershey, PA, United States, ⁴Psychology, Penn State, University Park, PA, United States

Johannes Lindemeyer¹, Ana-Maria Oros-Peusquens¹, and N.-Jon Shah^1,2
1Institute of Neuroscience and Medicine - 4, Forschungszentrum Jülich, Jülich, Germany, ²Department of Neurology, Faculty of Medicine, JARA, RWTH Aachen University, Aachen, Germany

This abstract describes a multi-step algorithm to efficiently remove B0 background field shifts from high-resolution phase data at 9.4T. Influences from static field inhomogeneities are accounted for as well as distortions induced by susceptibility distributions with arbitrary geometry residing outside and inside the field-of-view, such as air/water interfaces. The presented algorithm is successfully applied to phase images of the cerebellum in a post mortem brain.

Alexandru Vlad Avram^1,2, Arnaud Guidon^2,3, Wei Li³, Chunlei Liu^3,4, and Allen W Song^3,4
1Section on Tissue Biophysics and Biomimetics, NICHD, National Institutes of Health, Bethesda, Maryland, United States, ²Biomedical Engineering Department, Duke University, Durham, NC, United States, ³Brain Imaging and Analysis Center, Duke University Medical Center, Durham, NC, United States, ⁴Radiology Department, Duke University Medical Center, Durham, NC, United States

We investigate the contribution of the nuclear paramagnetic susceptibility to the frequency shifts between brain tissues in the 3D SPGR experiment and its dependence on sequence parameters. Our findings indicate that, depending on sequence parameters, differences in steady state nuclear magnetization between gray and white matter (WM) and cerebrospinal fluid and WM can result in both paramagnetic and diamagnetic frequency shifts within a range of a few parts per billion. We hope that our preliminary results help design improved protocols for quantitative mapping of iron and myelin concentrations and for investigating the contrast mechanism underlying phase and susceptibility imaging.

Karsten Sommer¹, Ferdinand Schweser¹, Andreas Deistung¹, and Jürgen Rainer Reichenbach^1
1Medical Physics Group, Dept. of Diagnostic and Interventional Radiology I, Jena University Hospital, Jena, Germany

In this contribution, a novel technique for separating susceptibility and non-susceptibility contributions in GRE phase data is presented.

Sagar Buch¹, Saifeng Liu², E. Mark Haacke^3,4, and Jaladhar Neelavalli^3
1McMaster University, Hamilton, Ontario, Canada, ²McMaster University, ³The Magnetic Resonance Imaging Institute for Biomedical Research, Detroit, MI, United States, ⁴Wayne State University

A simulated 3D model is produced to understand the study phase behavior of the geometries inside the brain and validate the processing techniques. The simulated phase images show agreement with the real phase data. The results of the simulated data, when compared with the real data, show that the ring around red nucleus structure can be an artifact or phase behavior of the geometry, instead of the general theory of having a capsule surrounding the boundary of red nucleus with different susceptibility value. This 3D model can be used in other applications wherever phase images are used in real data.

Jan Sedlacik¹, and Ferdinand Schweser^2
1Neuroradiology, University Medical Center Hamburg-Eppendorf, Hamburg, Germany, ²Medical Physics Group, Department of Interventional and Diagnostic Radiology, Jena University Hospital, Jena, Germany

Signal-phase contrast in gradient echo imaging delineates anatomic brain structures in great detail. While the phase contrast in the deep gray matter nuclei has been attributed to iron, the origin of the phase contrast between gray and white matter is not yet fully understood. He and Yablonskiy proposed a Lorentzian cylinder approximation which predicts phase effects due to the (sub-)cellular diffusion anisotropy in the vicinity of myelin bundles. Liu et al. found considerable anisotropy of the magnetic susceptibility of myelin-lipids. Purpose of this work was to investigate both phenomena more closely using phantom experiments with well-defined magnetic and structural properties.

Xiaolei Song^1,2, Raag Airan^1,2, Dian R Arifin^1,2, Segun Bernard^1,2, Assaf A Gilad^1,2, Peter C.M. van Zijl^1,3, Michael T Mcmahon^1,3, and Jeff W.M. Bulte^1,2
1Russell H. Morgan Department of Radiology and Radiological Sciences, Johns Hopkins University School, Johns Hopkins University, Baltimore, MD, United States, ²Cellular Imaging Section, Institute for Cell Engineering, The Johns Hopkins University, Baltimore, MD, United States, ³F.M.Kirby Research Center for Functional Brain Imaging, Kennedy Krieger Institute, Baltimore, MD, United States

Mucin-1, a large cell surface marker of epithelial cell lines, is found in underglycosylated form (uMUC-1) in almost all human epithelial cell adenocarcinomas and non-epithelial cancers, and is one of the early hallmarks of tumorigenesis. We used chemical exchange saturation transfer (CEST) MRI to specifically detect uMUC-1 expressing cells. Encapsulated cells and in vivo mouse tumor models of two cancer cell lines, LS174T (uMUC-1+) and U87 (uMUC-1-) show that uMUC-1-expressing cells display differential CEST contrast. This glycosyl (OH)-related CEST MRI may be potentially used for non-invasive phenotyping of tumors based on uMUC-1 expression.

Sai Chilakapati¹, Sanjana Reddy¹, Mohammad Haris¹, Anup Singh¹, Kejia Cai¹, Kavindra Nath², Hari Hariharan¹, and Ravinder Reddy^1
1Radiology, University of Pennsylvania, Philadelphia, Pennsylvania, United States, ²Molecular Imaging Lab, University of Pennsylvania, Philadelphia, Pennsylvania, United States

In the current study, we characterized the chemical exchange saturation transfer effect (CEST) effect from glucose at brain physiological concentration (<10mM) using different saturation pulse power and saturation duration and image them at high spatial resolution in in vitro on 7T human scanner. The CEST peak was observed at ~1 ppm and showed linear relation with glucose concentration. The next step is to monitor the change in the brain glucose concentration after intravenous injection of glucose using the optimized saturation parameters.

Sheng-Min Huang¹, Yun-Hsuan Lin¹, Yi-Chun Wu², Yi-Jui Liu³, and Fu-Nien Wang^1
1Department of Biomedical Engineering and Environmental Sciences, National Tsing Hua University, Hsinchu, Taiwan, ²Molecular Imaging Center, Chang Gung Memorial Hospital, Taoyuan, Taiwan, ³Department of Automatic Control Engineering, Feng Chia University, Taichung, Taiwan

CEST MRI has been showed for the ability of detecting the change of chemical microenvironment, such as pH level. In this study, dynamic amide proton transfer CEST imaging was first utilized for pharmacological MRI on a normal rat model. The Z-spectra were dynamically scanned after administration of methamphetamine. Different responses were discovered between cortex and striatum during methamphetamine challenge. Dynamic CEST imaging was demonstrated for the feasibility of observation of the drug effect and providing additional information.

Karin Shmueli^1,2, Peter van Gelderen¹, and Jozef H Duyn^1
1Advanced MRI Section, Laboratory of Functional and Molecular Imaging, National Institute of Neurological Disorders and Stroke, National Institutes of Health, Bethesda, MD, United States, ²Department of Medical Physics and Bioengineering, University College London, London, United Kingdom

It is unknown which molecules are primarily responsible for tissue exchange-induced frequency contrast (f_e). So far, f_e measurements have relied on internal frequency reference chemicals but these interact, causing confounding frequency shifts. Therefore, we developed a reference-chemical-free phantom-based method to measure f_e and applied it to cerebrosides. The phantom is an interchangeable cylindrical tube inside a saline-filled sphere. Frequency images were acquired with the (cerebroside- or saline-filled) tube parallel and perpendicular to B₀ to separate orientation-dependent susceptibility-induced contributions from f_e. Precise cerebroside susceptibility and f_e values were obtained at 7 Tesla and the phantom could be filled with different chemicals.

Phillip Zhe Sun¹, Charng-Ming Liu¹, Philip K Liu¹, and Renhua Wu^2,3
1Department of Radiology, Athinoula A. Martinos Center for Biomedical Imaging, Charlestown, MA, United States, ²Department of Radiology, 2nd Affiliated Hospital of Shantou University Medical College, Shantou, Guangdong, China, ³Provincial Key Laboratory of Medical Molecular Imaging, Shantou, Guangdong, China

CEST MRI measures dilute CEST agents and microenvironment properties, holding promise for in vivo applications. Because of confounding concomitant RF irradiation and relaxation effects, CEST-weighted MRI contrast, however, may not fully characterize the underlying CEST phenomenon. We postulated that the accuracy of quantitative CEST MRI could be improved if the labeling efficiency and RF spillover factor were estimated and taken into account. The simulation was confirmed with a multi-compartment Creatine CEST phantom of serial concentration. The proposed solution provides simplified yet reasonably accurate quantification of the underlying CEST system, which may help guide the ongoing development of quantitative CEST MRI.

Tao Jin¹, and Seong-Gi Kim^1
1Department of Radiology, University of Pittsburgh, Pittsburgh, Pennsylvania, United States

The amide proton transfer (APT) effect is sensitive to the mobile protein/peptide levels but its imaging contrast in vivo is quite low because the amide water exchange is very slow. Recently, it has been proposed that a novel chemical exchange contrast based on the faster amine-water proton exchange (APEX) may also be sensitive to protein/peptide concentrations. Here we measured labile amide, amine and hydroxyl protons at 9.4 T using different protein and peptide phantoms and showed that the APEX contrast may potentially be a more sensitive biomarker than APT for protein/peptide imaging at a high magnetic field.

Xiaopeng Zong¹, Tao Jin¹, Ping Wang¹, and Seong-Gi Kim^1
1Department of Radiology, University of Pittsburgh, Pittsburgh, PA, United States

The amine-proton is present in abundant free amino acids, as well as in amino-acid residues of protein and peptide side chains. Recently, the amine-water proton exchange (APEX) effect has been proposed as a sensitive molecular imaging contrast at high magnetic field. In this work, we performed multi-parametric MRI and spectroscopic experiments in a rat MCAO model to investigate its potential applications in stroke. Our results show that the APEX signals are sensitive to rapid tissue pH changes following stroke and are also correlated with more gradual metabolite concentration (e.g. Glu) changes, and thus is a promising technique for stroke applications.

Francisco Torrealdea¹, Marilena Rega¹, Mark Lythgoe², David Thomas¹, Simon Walker Samuel², and Xavier Golay^1
1UCL Institute of Neurology, University College London, London, London, United Kingdom, ²UCL Center of Advanced Biomedical Imaging, University College London, United Kingdom

Detection of CEST signal in vivo from the hexoses and pentoses of the Glycolytic pathway would allow distinguishing the GlucoCEST signal is coming from extracellular or intracellular origin. This would open the possibility of exploring the kinetics of the cancer metabolism non-invasively under the MRI. In this preliminary study we show that both hexoses (glucose and glucose-6-phosphate) give an enhanced CEST signal compared to the pentoses (fructose-6-phosphate and fructose-1,6-biphosphate). Despite this could be a limitation of the extend of this technique, the high CEST signal from fructose-1,6-biphosphate suggests the feasibility of detecting intracellular metabolic activity.

Mohammad Haris¹, Kejia Cai¹, Anup Singh¹, Ravi Prakash Reddy Nanga¹, C Damodar Reddy¹, Hari Hariharan¹, and Ravinder Reddy^1
1Radiology, University of Pennsylvania, Philadelphia, Pennsylvania, United States

In the current study, our objective was to map the change in creatine (Cr) level in response to mild exercise at high spatial resolution in calf muscles by exploiting the chemical-exchange-saturation-transfer effect between Cr amine proton and bulk water protons (CrEST). After exercise an increase in CrEST contrast was observed in each calf muscles, which return to baseline following relaxation. We conclude that high resolution imaging of Cr is possible without any contamination from PCr. This technique may provide an opportunity to map any alteration in the [Cr] and creatine kinase reactions at high spatial resolution in muscle pathologies.

Mohammad Haris¹, Anup Singh¹, Kejia Cai¹, Ravi Prakash Reddy Nanga¹, Catherine Debrosse¹, Hari Hariharan¹, and Ravinder Reddy^1
1Radiology, University of Pennsylvania, Philadelphia, Pennsylvania, United States

Exchange rate (kex) and logarithmic dissociation constant (pKa) of amine protons of creatine (Cr), phosphocreatine (PCr) and adenosine-triphosphate (ATP) were calculated with an objective to explore the chemical-exchange-saturation-transfer-effects (CEST) of these metabolites. The kex of amine protons in Cr was found to be 7-8 times higher than PCr and ATP. Higher kex in Cr is associated with the low pKa value suggesting the faster dissociation of its amine protons compared to PCr and ATP. CEST imaging of these metabolites showed predominant CEST contrast from Cr. These results provide a new method to perform high resolution proton imaging of Cr without any contamination from PCr.

Scott D. Swanson¹, and Dariya I. Malyarenko^1
1Department of Radiology, University of Michigan, Ann Arbor, Michigan, United States

Mixtures of surfactants, water, and alcohols form well characterized systems that mimic many properties of biological membranes. We have found that these lyotropic lamellar liquid crystals (as surrogate membranes) generate MT between the water and lipid phases and allow numerous molecular permutations to help disentangle MT properties. We report here studies of the mole fraction of decanol and weight percent of water influence on MT parameters, such as the estimated solid component (Mb0), cross-relaxation rate (Rt), and solid component T2 (T2b), in lamellar liquid crystals composed of sodium dodecyl sulfate (SDS), decanol, and water.

Sean Foxley¹, Way Cherng Chen¹, and Karla L Miller^1
1FMRIB Centre, University of Oxford, Oxford, United Kingdom

In this present study, we focus on measurements that are sensitive to the distribution of frequencies within a voxel: multi-GRE magnitude spectra and BSSFP profile asymmetries. The latter method detects asymmetries in the BSSFP frequency profile, which are driven by asymmetries in the underlying frequency distribution (as collected with multi-GRE). We compare multi-GRE- and BSSFP-based metrics for tracts with approximately parallel or perpendicular orientations, demonstrating orientation dependence that suggests a single underlying source. Calculation of the multi-GRE spectral and BSSFP profile asymmetries corroborate previously reported orientational dependence of the signal T₂^* decay, but also suggest a more complicated underlying effect.

Samuel James Wharton¹, and Richard Bowtell^1
1Sir Peter Mansfield Magnetic Resonance Centre, School of Physics and Astronomy, University of Nottingham, Nottingham, United Kingdom

Recent studies at high magnetic field strengths have shown that there is a direct link between the orientation of white matter fibers to the main magnetic field and the contrast observed in magnitude and phase images acquired using gradient echo MRI. Understanding the origin of this link could offer researchers access to a new diagnostic tool for investigating tissue microstructure. Here, we show that fibre-orientation-dependent contrast in magnitude and phase images can be fully explained by modelling nerve fibers as long, hollow cylinders of myelin, within which: (i) myelin-water generates a rapidly-decaying signal and (ii) the magnetic susceptibility is anisotropic.

Zhao Li¹, Chaohsiung Hsu¹, Ryan Quiroz¹, Raymond Ngo¹, Clifton Shen², Mark Girgis², Vay L. Go³, Yu-Hao Chen⁴, Lian-Ping Hwang⁴, and Yung-Ya Lin^1
1Chemistry and Biochemistry, UCLA, Los Angeles, CA, United States, ²Crump Molecular Imaging Institute, UCLA, Los Angeles, CA, United States, ³Center for Excellence in Pancreatic Diseases, UCLA, Los Angeles, CA, United States, ⁴Chemistry, National Taiwan University, Taipei, Taiwan

Early detection of high-grade malignancy using enhanced MRI techniques significantly increases not only the treatment options available, but also the patients’ survival rate. For this purpose, a conceptually new approach, termed “Active-Feedback MR”, was developed. An active feedback electronic device was homebuilt to implement active-feedback pulse sequences for spin avalanching amplification. With this new approach, early detection of pancreatic cancers can be achieved through sensitively imaging the dipolar fields induced by targeted magnetic nanoparticles. Early detection of orthotopic glioblastoma multiforme with 4-5 times improved contrast than conventional images can be reached through sensitively imaging susceptibility variations due to irregular water contents and deoxyhemoglobin.

Uzi Eliav¹, Michal Komlosh^2,3, Peter J. Basser², and Gil Navon^1
1School of Chemistry, Tel Aviv University, Tel Aviv, Israel, ²STBB/PPITS/NICHD/NIH, Bethesda, MD, United States, ³CNRM and Henry Jackson Foundation, Bethesda, MD, United States

In the present work it is shown that by combining the double quantum filtered, magnetization transfer and the ultra-short TE MRI methods it is possible to obtain contrast between tissue compartments based on the following factors: (a) the residual dipolar coupling interaction within the macromolecules, which depends on their structure, (b) residual dipolar interactions within the water molecules, and (c) the magnetization exchange rate between macromolecules and water. The technique is demonstrated in rat tail specimen where the collagenous tissue, such as tendons and the annulus pulposus of the disc are highlighted in these images, and their macromolecular properties along with those of bones and muscles can be characterized.

Janggeun Cho^1,2, Jee-Hyun Cho^1,2, Sangdoo Ahn², and Chulhyun Lee^1
1Division of Magnetic Resonance Research, Korea Basic Science Institute, Ochang 363-883, Chungbuk, Korea, ²Department of Chemistry, Chung-Ang University, Seoul 156-756, Korea

As a new method for contrast enhancement in MRI, the intermolecular multiple quantum coherences (iMQCs) MR imaging has recently attracted considerable attention mostly because of the intrinsic sensitivity of the iMQCs to changes the magnetization and susceptibility structures. Relaxation times (T1 and T2) play an important role in creating and determining susceptibility structures of the matters for MRI detection. In this study, we have experimentally and numerically investigated the relaxation time dependence on the intensity profiles and/or contrasts of the intermolecular double quantum coherences (iDQCs) MR images by fixing and varying T1 and T2, respectively and vice versa.

Stefanie Remmele¹, Alois Martin Sprinkart², Andreas Müller², Frank Träber², M von Lehe³, Jürgen Gieseke^2,4, Sebastian Flacke^2,5, Winfried A Willinek², Julien Sénégas¹, Jochen Keupp¹, Hans H Schild², and Petra Mürtz^2
1Philips Research, Hamburg, Germany, ²Department of Radiology, University of Bonn, Bonn, Germany, ³Department of Neurosurgery, University of Bonn, Bonn, Germany,⁴Philips Healthcare, Best, Netherlands, ⁵Department of Radiology, Tufts University Medical School, MA, United States

This work describes a novel approach of how to analyze the multi-parametric response to a respiratory challenge in order to differentiate normoxic from non-normoxic tissue. The technique is based on the simultaneous measurement of the change of two relaxation parameters (R1 and R2*) during inhalation of oxygen-enriched air. (dR2*,dR1) scatter-plots are used to identify tissue voxels with an response behavior presumably related to abnormal oxygen function or oxygenation. We present abnormal response maps that visualize these non-normoxic voxels for 4 patients with intracranial tumors.

Yoonho Nam¹, Bonsuk Koo¹, Dosik Hwang¹, and Dong-Hyun Kim^1
1Electrical and Electronic Engineering, Yonsei University, Seoul, Korea

Quantitative myelin water imaging can be a useful tool for studying white matter diseases such as multiple sclerosis. Myelin water imaging via multi-echo gradient echo based T2* imaging has recently been introduced as an alternative to multi-echo spin echo based T2 imaging. T2* based myelin water imaging has some technical benefits such as lower SAR, larger volume coverage. In this work, the potential for whole brain quantitative myelin water imaging using T2* decay analysis was investigated at 3T.

Florence Colliez¹, Julie Magat¹, Vincent Denolin², Thierry Duprez³, Bénédicte F Jordan¹, and Bernard Gallez^1
1Biomedical Magnetic Resonance Group, Université Catholique de Louvain, Brussels, Belgium, ²Philips Health Care Benelux, Belgium, ³Institute of Clinical and Experimental Research, Université Catholique de Louvain, Brussels, Belgium

The purpose of the current work was to translate the MOBILE technique, a method developed to map variations in oxygenation based on the changes in the relaxation properties of the tissue lipids by exploiting the higher solubility property of oxygen in lipids than in water, on a 3T clinical MR scanner. MOBILE consists in the selective measurement of the lipids relaxation rate with water suppression. In order to validate the technique, we compared the sensitivity of MOBILE in the brain of healthy volunteers (n=4) with changes in R2* and R1 of water with respect to an hyperoxic breathing challenge.

Denis Grenier¹, and Olivier Beuf^1
1IRM & Optique, Creatis UMR CNRS 5220, Lyon, France

MR signal loss pertinent to dipolar interaction process is estimated in vivo in the rat brain tissues. This work enable the calculation of dipolar-free and dipolar-induced transverse relaxation (T2) components

Uvo Christoph Hoelscher¹, Arne Gruenewald², Martin Blaimer¹, and Peter Michael Jakob^2
1Research Center for Magnetic Resonance Bavaria, Wuerzburg, Germany, ²Experimental Physics 5 (Biophysics), University of Wuerzburg, Wuerzburg, Germany

Relaxation dispersion contrast is a new method to acquire images where signal is proportional to the magnetic field dependence of the relaxation rate. The image intensity indicates whether the relaxation dispersion rate of the tissue experiences strong or weak changes when going from 1.4T to 1.6T. The resulting images yield a completely new relaxation weighted MR contrast which is yet to be interpreted but has proven in spectroscopic measurements to be very interesting.

Way Cherng Chen¹, and Karla Miller^1
1Centre for Functional MRI of the Brain (FMRIB), University of Oxford, Oxford, Oxfordshire, United Kingdom

A phantom made of densely-packed fibers has been constructed to study the relationship of T2* and GRE phase maps to the orientation to B0. By minimizing the contribution of bulk magnetic susceptibility effects, signal changes can be more definitively attributed to the microstructural properties of the phantom. We observe a similar trend in the fiber phantom to that previously reported in white matter fibre tracts supporting the idea that phase and T2* variation in white matter fibers is related to magnetic susceptibility shifts associated with axonal micro-geometry.

Hai Zheng¹, Tiejun Zhao², Yongxian Qian³, Tamer Ibrahim^1,3, and Fernando Boada^1,3
1Bioengineering, University of Pittsburgh, Pittsburgh, Pennsylvania, United States, ²Siemens Medical Solutions USA, Pittsburgh, Pennsylvania, United States, ³Radiology, University of Pittsburgh, Pittsburgh, Pennsylvania, United States

Susceptibility-induced artifacts lead to geometric distortions and/or signal loss near some important functional regions. This signal loss increases at ultra-high-field and is often located at the sites being investigated with fMRI and must, therefore, be compensated in order to obtain robust activation. In this study, previously reported PTX 3DTRF pulse design is extended to restore the lost signal over the whole brain and increase the BOLD contrast to brain activation in ultra high field. In addition, the compromise between the performance of signal recovery and computational time is also investigated through the effect of trajectory design.

Clare Poynton¹, Elfar Adalsteinsson^1,2, Edith V. Sullivan³, Adolf Pfefferbaum³, and William Wells III^1,4
1Health Sciences and Technology, Harvard - MIT, Cambridge, MA, United States, ²Department of Electrical Engineering and Computer Science, MIT, Cambridge, MA, United States, ³Department of Psychiatry & Behavioral Sciences, Stanford University School of Medicine, Stanford, CA, United States, ⁴Department of Radiology, Harvard Medical School, Brigham and Women's Hospital, Boston, MA, United States

Quantifying magnetic susceptibility in the brain from the phase of the MR signal provides a non-invasive means for measuring the accumulation of iron during aging and neurodegenerative disease. Phase observations from local susceptibility distributions, however, are corrupted by external biasfields, only limited observations are available, and the inversion is ill-posed. We describe an atlas-based variational approach to susceptibility estimation that incorporates a tissue-air atlas to resolve ambiguity in the forward field model, while eliminating biasfields through application of the Laplacian. Results show an age-dependent increase in susceptibility in sub-cortical structures and strong correspondence with postmortem iron concentrations.

Sagar Buch¹, E. Mark Haacke^2,3, Saifeng Liu¹, and Jaladhar Neelavalli^2
1School of Biomedical Engineering, McMaster University, Hamilton, Ontario, Canada, ²The Magnetic Resonance Imaging Institute for Biomedical Research, Detroit, Michigan, United States, ³Academic Radiology, Wayne State University, Detroit, Michigan, United States

Homodyne high pass filter and SHARP method, which are commonly used to remove the background field variations from the real phase data, are used to process the phase images produced from a simulated 3D model of the brain, which includes the geometries of basal ganglia and grey matter/white matter. Susceptibility maps (SMs) of the processed simulations are compared with that of the real phase images. A negative susceptibility region, seen in both simulated and real phase data, is discussed as a possible artifact caused by the processing techniques after comparing the simulated SMs produced from unprocessed and processed phase data.

Ferdinand Schweser¹, Karsten Sommer¹, Andreas Deistung¹, and Jürgen Rainer Reichenbach^1
1Medical Physics Group, Dept. of Diagnostic and Interventional Radiology I, Jena University Hospital, Jena, Germany

Quantitative susceptibility mapping (QSM) is a novel imaging technique that determines tissue magnetic susceptibility from the phase of complex gradient-echo data. Due to the ill-posed nature of this inverse problem, regularization strategies are required to reduce streaking artifacts on the computed susceptibility maps. This paper presents an improved QSM algorithm, Homogeneity Enabled Incremental Dipole Inversion (HEIDI), which utilizes a sophisticated problem-specific incremental inversion procedure and a priori information on the homogeneity of the susceptibility distribution.

Ferdinand Schweser¹, Martin Stenzel², Andreas Deistung¹, Karsten Sommer¹, Hans-Joachim Mentzel², and Jürgen Rainer Reichenbach^1
1Medical Physics Group, Dept. of Diagnostic and Interventional Radiology I, Jena University Hospital, Jena, Germany, ²Section Pediatric Radiology, Dept. of Diagnostic and Interventional Radiology I, Jena University Hospital, Jena, Germany

Quantitative susceptibility mapping (QSM) is a novel imaging technique that employs the complex-phase of conventional gradient-echo sequences to obtain quantitative maps of bulk tissue magnetic susceptibility. Although QSM is currently regarded as the successor of the well-established Susceptibility Weighted Imaging (SWI) technique, only few studies have hitherto investigated the potential of applying QSM to pathologic conditions in a clinical scenario, thus impressing the perception of QSM being a research tool only so far. This paper presents initial results of applying QSM in routine pediatric neuroradiology to demonstrate the impressive potential of this novel imaging technique.

Ningzhi Li¹, Wen-tung Wang², Dzung L Pham¹, and John A Butman^3
1Image Processing Core, Center for Neuroscience and Regenerative Medicine, Bethesda, MD, United States, ²Human Imaging Core, Center for Neuroscience and Regenerative Medicine, Bethesda, MD, United States, ³Radiology Department, National Institutes of Health, Bethesda, MD, United States

A novel approach for choosing echo time in susceptibility weighted imaging (SWI) is introduced in this study. A model for SWI signal is derived, based on which optimal echo time can be chosen to yield maximum susceptibility contrast enhancement for a selected tissue type. Use of this approach requires information about tissue T2* values and phase characteristics, both of which can be easily obtained from multi-echo dataset. We find that optimal echo times are close to the T2* of the tissue.

Ferdinand Schweser¹, Andreas Deistung¹, Karsten Sommer¹, and Jürgen Rainer Reichenbach^1
1Medical Physics Group, Dept. of Diagnostic and Interventional Radiology I, Jena University Hospital, Jena, Germany

Magnetic susceptibility is an intrinsic physical tissue property, which recently became accessible in vivo by quantitative susceptibility mapping (QSM). The practical applicability of QSM, however, is currently hampered by the enormous numerical complexity and computational cost associated with the QSM reconstruction. In this study, we present and analyze a novel QSM framework, Superfast Dipole Inversion (SDI), that not only allows reconstruction of susceptibility maps from raw gradient-echo phase data in near real-time, but also enables processing of phase images with open-ended fringe lines, which often occur in practice due to improper combination of multi-channel images and which have hitherto hindered QSM.

Samuel James Wharton¹, and Richard Bowtell^1
1Sir Peter Mansfield Magnetic Resonance Centre, School of Physics and Astronomy, University of Nottingham, Nottingham, United Kingdom

Recent work has demonstrated that the magnetic susceptibility of white matter has a measurable anisotropy. The likely source of this anisotropy is the myelin sheath that surrounds axons and specifically, the highly ordered aliphatic chains in the myelin. Here, we derive analytical expressions which characterise the field perturbation due to a model of the myelin sheath consisting of a hollow cylinder in which the magnetic susceptibility is anisotropic and the principal component of the susceptibility tensor is radially directed. The expressions are validated by imaging a cylindrical phantom incorporating a pyrolytic graphite sheet with a known, highly anisotropic, susceptibility.

Sung-Min Gho¹, Wei Li², and Dong-Hyun Kim^1
1Electrical and Electronic Engineering, Yonsei University, Sinchon-dong, Seoul, Korea, ²Brain Imaging and Analysis Center, Duke University, Durham, NC, United States

Susceptibility Weighted Imaging (SWI) uses susceptibility differences between tissues, (i.e. phase image) to enhance the contrast of the magnitude image, such as venography. However, quantitative value of image phase is hard to estimate due to its non-local and orientation dependent properties. Limitations such as restriction of resolution to achieve optimum phase shift in SWI arise from these reasons. In this abstract, therefore, we propose a new SWI method (QSWI) for venography using quantitative susceptibility mapping (QSM) which allows reliable quantification of magnetic susceptibility (i.e. intrinsic property of tissue) to improve the contrast and mitigate above limitations.

Viktor Vegh¹, and David C Reutens^1
1Centre for Advanced Imaging, University of Queensland, Brisbane, Queensland, Australia

The research aims to develop a novel magnetic susceptibility imaging method applicable on high-field MRI scanners. Current MRI methods used to assess local susceptibility utilise image signal phase. In this project we examine whether additional susceptibility information may be extracted by using the total raw signal data. We compare our new method to T1-weighted magnitude images, susceptibility weighted images and phase contrast images.

Karen Mok¹, Jaladhar Neelavalli², Saifeng Liu³, and E. Mark Haacke^4,5
1Mcmaster University, Hamilton, ON, Canada, ²Department of Radiology, Wayne State University, ³McMaster University, ⁴Wayne State University, ⁵School of Biomedical Engineering, McMaster University

SWI is a widely used clinical tool for venography in the brain [1]. It is a unique imaging methodology which combines the magnitude and phase information, where phase is assumed to be directly related or proportional to the magnetic susceptibility of the structure. While this is true, the actual phase effects are dependent on tissue orientation, as well as the size and shape of the vessels and paramagnetic structures inside the brain. The phase images are also affected by local and global susceptibility effects arising from tissue interfaces between air and bone. At typical imaging resolutions of high in-plane and low through-plane resolution (i.e. voxel aspect ratios from in-plane to slice direction varying from 1:2 to 1:5), due to phase integration effects, the sign of the venous vessel phase turns out to be quite robustly consistent, independent of the vessel orientation. This is partly the reason why SWI typically acquired axially is so successful in depicting the veins exquisitely. However, at isotropic resolutions, when there are minimal phase integration effects, veins have different signs in the phase images depending on their orientation. This creates a problem in generating a proper phase based mask for susceptibility weighting. There have been attempts to use bi-directional masks which use both positive and negative signed phase values in their mask creation [2]. Although this may provide good venograms, it can also lead to artifacts due to remnant phase wraps. In this abstract, we evaluate the possibility of using the susceptibility map, instead of phase, to create the mask for susceptibility weighting. Since susceptibility maps are orientation independent, they provide more accurate representation of the brain as well as reducing the enlargement effect of the vessels in the brain due dipole effects.

Sharon K Schreiber¹, Bradley D Clymer², Michael V Knopp³, and Petra Schmalbrock^3
1Biomedical Engineering, The Ohio State University, Columbus, Ohio, United States, ²Electrical and Computer Engineering, The Ohio State University, Columbus, Ohio, United States, ³Radiology, The Ohio State University, Columbus, Ohio, United States

This work supports the contention that susceptibility dominates phase contrast in dense white matter fiber bundles, such as the corpus callosum. It suggests that this contrast at high fields may be used to distinguish subtle differences in white matter microarchitecture.

Christian Langkammer^1,2, Ferdinand Schweser³, Nikolaus Krebs², Andreas Deistung³, Walter Goessler⁴, Eva Scheurer², Karsten Sommer³, Gernot Reishofer⁵, Kathrin Yen⁶, Franz Fazekas¹, Juergen R Reichenbach³, and Stefan Ropele^1
1Department of Neurology, Medical University of Graz, Graz, Austria, ²Ludwig Boltzmann Institute for Clinical-Forensic Imaging, Graz, Austria, ³Medical Physics Group, Department of Diagnostic and Interventional Radiology I, Jena University Hospital, Jena, Germany, ⁴Institute of Chemistry - Analytical Chemistry, University of Graz, Graz, Austria, ⁵Department of Radiology, Medical University of Graz, Graz, ⁶Institute of Forensic Medicine, University of Heidelberg, Heidelberg, Germany

In this study quantitative susceptibility mapping (qSM) was performed in postmortem brain in situ and correlated with iron concentration as assessed by ICP mass spectroscopy. A strong linear correlation between iron concentration and magnetic susceptibility was found which became significantly weaker when considering white matter only. These results suggest that qSM is a reliable tool to study iron levels in gray matter. Iron mapping in white matter seems to be less reliable unless the magnetic susceptibility of myelin and its orientation is not accounted for.

Timo Liimatainen¹, Silvia Mangia², Andrew E Tyan², Hanne Hakkarainen¹, Djaudat Idiyatullin², Dennis Sorce², Michael Garwood², and Shalom Michaeli^2
1A.I. Virtanen Institute for Molecular Sciences, University of Eastern Finland, Kuopio, Finland, ²Center for Magnetic Resonance Research, University of Minnesota, Minneapolis, MN, United States

Series of amplitude and frequency modulation functions were designed using frequency swept pulses based on sine and cosine amplitude and frequency modulations to allow generation of MRI tissue contrast in the rotating frames of rank 1 ≤ n ≤ 5. The method is an extension of previously introduced technique RAFF (Relaxations Along a Fictitious Field), where the fictitious fields generated during non-adiabatic rotation lead to a locking magnetic field. The current method provides reduced SAR as compared to conventional and adiabatic rotating frame methods, and generates locking magnetic fields with much greater amplitudes than the maximum power of the pulses.

Audrey P Fan^1,2, Berkin Bilgic^1,2, Keith A Heberlein², Bruce R Rosen^2,3, and Elfar Adalsteinsson^1,2
1Electrical Engineering and Computer Science, Massachusetts Institute of Technology, Cambridge, MA, United States, ²Radiology, Athinoula A. Martinos Center for Biomedical Imaging, Charlestown, MA, United States, ³Harvard Medical School, Boston, MA, United States

We substantially improved the phase-based regional oxygen metabolism (PROM) MRI method to quantify absolute cerebral metabolic rate of oxygen (CMRO₂) in vivo across the cortex. Quantitative susceptibility maps were reconstructed from high-resolution 3D phase images using L1-regularization, from which venous oxygen saturation (Y_v) was quantified. Improved removal of background fields in the measured phase allowed for less than 5% error Y_v even in vessels perpendicular to B₀. Y_v was averaged for 31 cortical regions per hemisphere after anatomical registration, representing 380% increased coverage from the previous version of PROM. With an additional perfusion map from arterial spin labeling, we measured mean Y_v=62.3±4%, CBF=48.7±10ml/100g/min, and CMRO₂=134±32μmol/100g/min across the cortex, which lie in the normal physiological range.

Cheng-Chieh Cheng^1,2, Lawrence Panych¹, and Bruno Madore^1
1Department of Radiology, Brigham and Women's Hospital, Boston, MA, United States, ²Graduate Institute of Biomedical Electronics and Bioinformatics, National Taiwan University, Taipei, Taiwan

Phase information from MR images can prove useful to detect field variations across the imaged object, such as in susceptibility-weighted imaging. A pulse sequence is proposed that can make efficient use of the entire TR period to encode susceptibility-related information. While a 'fast imaging with steady-state precession' (FISP) signal may require relatively long TE settings to accurately detect susceptibility-related dephasing, an inverted FISP (or PSIF) signal acquired early in the TR interval can help improve SNR in the generated phase maps.

Wei He¹, Ling Xia¹, Yiyuan Cheng¹, Feng Liu², and Stuart Crozier^2
1Department of Biomedical Engineering, Zhejiang University, Hangzhou, Zhejiang, China, ²The School of Information Technology and Electrical Engineering, The University of Queensland, Brisbane, Australia

A discrete Particle Swarm Optimization (dPSO) is proposed for solving the branch-cut phase unwrapping problem. All the residues are first grouped by dividing the phase image into sub-regions. Then, the dPSO is performed region-by-region to match the opposite polarity residues that are connected by branch cuts afterward. The nearest-neighbor algorithm is employed to place branch cuts among any residues left. Finally, a flood-fill method is implemented to unwrap the phases over the whole image avoiding the branch cuts. The performance of the proposed algorithm was tested on sample MRI wrapped-phase images. The experiment shows that it can achieve better results compared to a previously published algorithm.

Martin Krämer¹, Tim Sprenger¹, Karl-Heinz Herrmann¹, and Jürgen R Reichenbach^1
1Medical Physics Group, Department of Diagnostic and Interventional Radiology I, Jena University Hospital, Jena, Germany

One dimensional phase unwrapping can be sensitive to noise or inhomogeneities in the phase data. Reliable analysis of 1D phase data however, often is a crucial pre-processing step for many image reconstruction algorithms (e.g.a EPI ghost correction, multi-echo field maps, ...). In our work we present a method that does not require phase unwrapping for fitting a known model to any measured 1D phase distribution. The only preliminary information required for the method to work is the function of the fit model for which the fit parameters are to be extracted from the phase data.

Petros Martirosian¹, Gerd Grözinger², Isabel Rauscher², Christian Würslin¹, Rolf Pohmann³, Fabian Springer², and Fritz Schick^1
1Section on Experimental Radiology, University of Tübingen, Tübingen, Baden-Württemberg, Germany, ²Diagnostic and Interventional Radiology, University of Tübingen, Germany, ³Magnetic Resonance Center, Max Planck Institute for Biological Cybernetics, Tübingen, Germany

Within the last years a variety of pulse sequences have been published which visualize musculoskeletal tissues such as menisci, tendons, or ligaments with positive contrast despite their relatively short T2-values. For optimization of sequence parameters by numerical simulations based on Bloch’s equations, knowledge of tissue relaxation times is essential. So far there has been no comprehensive MRI study regarding the systematic measurement of their inherent relaxation properties. Therefore, the purpose of this study was to exemplarily quantify the T1, T1, T2, and T2* relaxation times of rapidly relaxing tissues in the human knee joint at 3T in an acceptable examination time.

Milos Ivkovic¹, Thanh Nguyen¹, Dushyant Kumar¹, Susan Gauthier¹, and Ashish Raj^1
1Weill Cornell Medical College, New York City, NY, United States

We propose a spatially adaptive filter, based on the Diffusion-Weighted MRI, that regularizes T2 relaxometry signal with respect to neuronal direction -information otherwise unavailable in T2 relaxometry. Similar approaches can be used with other myelin content estimation techniques.

Jinnan Wang¹, Niranjan Balu², Thomas S Hatsukami², Chun Yuan², and Peter Börnert^3
1Philips Research North America, Briarcliff Manor, NY, United States, ²University of Washington, ³Philips Research Europe

Although MR has been shown to provide accurate and reliable atherosclerotic plaque components identification, the detection of high risk components is usually compromised when they coexist with calcification. Speckled calcification, which presents in over 35% of advanced plaques, usually coexists with high-risk plaque components at an arbitrary fraction. As a result, the MR signal from high risk components is masked by reduced signal from speckled calcification. In an aim to potentially correct for this signal drop caused by speckled calcification, a serial UTE based approach is proposed to accurately quantify the calcification fraction on a voxel level.

Ileana Ozana Jelescu¹, Nicolas Boulant¹, Denis Le Bihan¹, and Luisa Ciobanu^1
1NeuroSpin, CEA/DSV/I2BM, Gif sur Yvette, France

As opposed to standard diffusion-weighted sequences, the DESIRE (Diffusion Enhancement of SIgnal and REsolution) method gains signal through diffusion. We implemented 2D DESIRE with 60 µm spatial resolution and evaluated enhancement in three media: water (D≈1.5x10^-3 mm²/s), thin and thick silicone oils (D≈1.8x10^-4 mm²/s and D≈9x10^-6mm²/s, respectively), following a saturation time of 336 ms. Average enhancement in the DESIRE images was 3 in water, 1.9 in thin oil and zero in thick oil, confirming that the stronger the diffusion, the greater the enhancement. The application of the technique to imaging barriers and multi-compartment phantoms is in progress.

Julie Magat¹, Patrice D Cani², Nathalie Delzenne², Bernard Gallez¹, and Bénédicte F Jordan^1
1Biomedical Magnetic Resonance Group, Université Catholique de Louvain, Brussels, Belgium, ²Metabolism and nutrition Group, Université Catholique de Louvain, Brussels, Belgium

Techniques able to assess quantitative peripheral pO2 are mandatory to verify the efficacy of therapeutic approaches aimed to salvage limbs in diabetic patients with critical limb ischemia. We recently developed a method able to map variations in oxygenation based on the changes in the relaxation properties of the tissue lipids by exploiting the higher solubility property of oxygen in lipids than in water. The aim of the current work was to apply the MOBILE technique in order to map peripheral tissue oxygenation in a mouse model of peripheral ischemia.

Lauri Juhani Lehto^1,2, Djaudat Idiyatullin², Michael Garwood², Olli Gröhn¹, and Curt Corum^2
1A. I. Virtanen Institute for Molecular Sciences, University of Eastern Finland, Kuopio, Eastern Finland, Finland, ²Center for Magnetic Resonance Research, University of Minnesota, Minneapolis, Minnesota, United States

SWIFT is an almost zero acquisition delay sequence. Recently it has been shown to exhibit phase contrast even though there is no echo time for phase accumulation as in gradient recalled echo (GRE) type of sequences. SWIFT phase contrast was investigated by imaging an ex vivo rat brain in three different orientations of the fiber bundles of corpus callosum in comparison to B0. SWIFT and GRE showed similarities in their phase contrast. SWIFT phase contrast was more high-pass filtered than GRE phase contrast due to SWIFT’s self high-pass filtering property. SWIFT may provide novel phase contrast of short T2 components.

Katharina Goebel¹, Nicoleta Baxan¹, Oliver Gruschke², Johannes S. Kern³, Cristina Has³, Jan G. Korvink^2,4, Leena Bruckner-Tudermann^3,4, Juergen Hennig¹, and Dominik von Elverfeldt^1
1Department of Radiology, University Medical Center Freiburg, Freiburg, Baden-Württemberg, Germany, ²IMTEK, Laboratory of Simulation, University of Freiburg, Freiburg, Baden-Württemberg, Germany, ³Department of Dermatology, University Medical Center Freiburg, Freiburg, Baden-Württemberg, Germany, ⁴Freiburg Institute for Advances Studies (FRIAS), University of Freiburg, Freiburg, Baden-Württemberg, Germany

If spatial resolution, contrast and sensitivity are sufficient, MR-microscopy can be used as a powerful instrument for in vivo investigation of the human skin as an alternative to histology. The key obstacle is the SNR decrease due to the low sensitivity of MR-microscopy. A reason for that is the layered structure of the skin demanding an extremely high-resolution over an enhanced FOV. A phased-array MR-microcoil was developed to alleviate such limitations combining the advantages of large FOV and high SNR. The presented coil was characterized in terms of B0 homogeneity. Feasibility studies were performed revealing high-resolution images of human skin.

Stefan Wintzheimer¹, Michael Ledwig¹, Toni Drießle¹, Ralf Kartäusch¹, Peter Michael Jakob², and Florian Fidler^1
1Research Center Magnetic-Resonance-Bavaria, Würzburg, Bavaria, Germany, ²Research Center Magnetic-Resonance-Bavaria

In this work a dedicated small imaging unit is proposed consisting of a sensitive rf-coil, an optimized microscopy gradient system and a standalone MRI console. This makes this portable small animal imaging unit an independent system, which can be used in any clinical scanner magnet. It has been build for 1.5 T and first high resolution mice images are presented.

Miriam Scadeng¹, Lauren DuBois², Christy Simeone², Stephanie Costelow², David Dubowitz¹, and Judy St Leger^2
1Radiology, University Of California San Diego, La Jolla, CA, United States, ²SeaWorld, San Diego, CA, United States

This study describes the techniques for in-ovo, in-vivo anatomical MR imaging of penguin eggs, including anesthetized developing penguin chicks. It illustrates the features of penguin egg viability throughout development. MRI has potential use as an in-vivo assay for environmental toxicity in eggs, and as a non destructive method of determining the normal anatomical appearances of chick development for scientific and educational purposes.

Stefan Henning¹, Daniel Edelhoff¹, Sabine Leick², Heinz Rehage², and Dieter Suter^1
1Exp. Physics III, TU Dortmund, Dortmund, Germany, ²Physical Chemistry II, TU Dortmund, Dortmund, Germany

We present methods to evaluate the functionality of microcapsules used for controlled and site specific delivery of drugs. By quantitative measurements of time dependent contrast agent concentration in the capsules with micro scale MR imaging, we can obtain diffusion coefficients revealing how different capsule preparations decrease the permeability. In addition, we can also show how the capsules dissolve in intestinal media whereas they remain stable in gastric media by using a micro scale FLASH sequence.

Linqing Li¹, Karla Miller¹, and Peter Jezzard^1
1FMRIB, Clinical Neurology Department, University of Oxford, Oxford, United Kingdom

DANTE (a rapid series of low flip angle RF pulses interspersed with gradients)pulse trains may be employed for flow fluid signal suppression module. During application of DANTE pulse trains, longitudinal magnetization of flowing spins is largely (or fully) attenuated due to phase dispersion accrued while flowing along the applied gradient. This is in contrast to static tissue, whose longitudinal magnetization is mostly preserved. This progressive saturation of flowing spins is insensitive to velocity (above a threshold). This work is a theoretical framework for motion sensitized and motion suppressed quantitative MRI when DANTE is applied.

R. Reeve Ingle¹, Joëlle K. Barral², Marcus Björk³, Erik Gudmundson^4,5, Petre Stoica³, and Dwight G Nishimura^1
1Electrical Engineering, Stanford University, Stanford, California, United States, ²Heart Vista, Inc., Los Altos, California, United States, ³Department of Information Technology, Systems and Control, Uppsala University, Uppsala, Sweden, ⁴Centre for Mathematical Sciences, Lund University, Lund, Sweden, ⁵Signal Processing Lab, ACCESS Linnaeus Center, KTH – Royal Institute of Technology, Sweden

Due to the high SNR efficiency of bSSFP, estimation of T1, T2 and other parameters from phase-cycled bSSFP images seems promising at first glance. We show that the conditioning of the bSSFP signal equation is such that accurate estimation of T1 and T2 using a moderate number of phase-cycled bSSFP images requires prohibitively high SNR. We derive theoretical limits for the minimum variance of T1 and T2 estimates using the Cramér Rao Bound, and we show that the SNR required to accurately estimate T1 and T2 is higher than what is typically obtained in vivo.

Philipp Krämer¹, and Lothar R. Schad^1
1Computer Assisted Clinical Medicine, Heidelberg University, Mannheim, Germany

A twisted projection imaging sequence is presented. The properties of the k-space trajectory are used to reduce scan time in 3D imaging. The sequence was implemented as a spoiled gradient recalled echo (SPGR) sequence and as a fully balandec steady-state free precession (bSSFP) sequence. The different contrast behaviors of these sequences are used for fast 3D mapping of T₁ and T₂ with a variable flip angle approach.

Martin Ott¹, Martin Blaimer¹, Philipp Ehses², Peter Michael Jakob^1,3, and Felix Breuer^1
1Magnetic Resonance Bavaria e.V, Würzburg, Bayern, Germany, ²Max Planck Institute for Biological Cybernetics, Tübingen, Germany, Tübingen, Germany, ³Department of Experimental Physics V, University of Würzburg, Würzburg, Bayern, Germany

A novel procedure is proposed to extract T₁, T₂,  and M₀ from the signal of several phase-cycled(PC)-bSSFP measurements at two flip-angles. The measured data are fitted on ellipse equations which are equal to complex bSSFP steady-state equations. The resulting offresonance free data for one flip angle as well as the ratio of two flip-angles can be fitted to obtain T₁, T₂ and M₀. In phantom measurements T₁ and T₂ values derived by this procedure show good agreement with those from single point reference measurements. in-vivo measurements of T₁ and T₂ are presented.

Nikola Stikov¹, Ives R Levesque², Christine L Tardif¹, Joëlle K Barral³, and G Bruce Pike^1
1Montreal Neurological Institute, Montreal, Quebec, Canada, ²Stanford University, Stanford, CA, United States, ³HeartVista, Inc., Los Altos, CA, United States

T₁ mapping is critical for most quantitative MRI, yet there is a large variation of reported values in vivo, an inconsistency that highlights the issue of reproducibility and accuracy. We compare the three most common T₁ mapping techniques (Inversion Recovery, Look-Locker and Variable Flip Angle), and show that despite good agreement in phantoms, there is a significant variation in brain T₁ values when different techniques are applied to the same healthy subject. We compute the white matter T₁ peak in 10 healthy subjects and observe a trend consistent with literature, with Look-Locker underestimating, and VFA overestimating the inversion recovery T₁ values. Our findings suggest that phantom studies are not sufficient for validation of T₁ mapping techniques.

Sofia Chavez^1
1Research Imaging Centre, Centre for Addiction and Mental Health, Toronto, Ontario, Canada

The Method of Slopes (MoS) has been proposed for simultaneous 3D B1 and T1 mapping. It relies on the acquisition of 3D SPGR signal at four different nominal flip angles and TR=40ms for brain. In this work, the B1 and T1 mapping steps are separated, allowing for an optimal trade-off between TR and resolution at each sampled point, resulting in robust B1 and T1 mapping with minimal scan time. The MoS is shown to result in more precise and accurate T1 compared to the VFA . The robustness of the method allows parallel imaging to further reduce scan time.

Sohae Chung¹, Elodie Breton², and Leon Axel^1
1Center for Biomedical Imaging, Radiology Department, NYU Langone Medical Center, New York, NY, United States, ²LSIIT - AVR, University of Strasbourg - CNRS, Strasbourg, France

Most cirrhotic livers have qualitatively inhomogeneous hepatic texture in contrast-enhanced MR images; this is related to the degree of liver necrosis, inflammation, and fibrosis. Quantitative liver T1 mapping could potentially provide additional useful information on liver abnormalities. However, conventional T1 mapping approaches, using a multi-point inversion recovery imaging sequence, have long acquisition times (≥ 20s) and are sensitive to abdominal organ or respiratory motion; this can lead to T1 fitting errors. In this work, a single-point T1 mapping method was used to calculate the liver T1 map with just two images acquired in a short (2s) acquisition time.

Anne Menini^1,2, Hélène Clique^1,2, Marine Beaumont^3,4, Jacques Felblinger^1,2, and Freddy Odille^1,2
1IADI, Nancy-Université, Nancy, France, ²U947, INSERM, Nancy, France, ³CIT801, INSERM, Nancy, France, ⁴CHU de Nancy, Nancy, France

T₁ mapping is useful to analyze myocardial fibrosis or liver perfusion. Motion artifacts and misregistration that can occur during the examination affect the quantification. GRICS is an adaptive reconstruction method that takes into account signals from physiological sensors to correct for motion artifacts. The variant of GRICS presented in this work has been adapted to T₁ and 3D. On a gadolinium phantom, T₁ maps were obtained using the variable flip angle method, with and without motion. Corrected T₁ map was also reconstructed with our method. It significantly improves the quantification compared with T₁ obtained without correction.

Cheukkai Hui¹, and Ponnada Narayana^2
1cheukkai.hui@uth.tmc.edu, Houston, Texas, United States, ²Radiology, UT Health, Houston, Texas, United States

A truly centric scheme along both phase encoding directions in the k-space for three-dimensional Look-Locker (LL) sequence for fast T1 acquisition is implemented. In addition, a multi-step fitting algorithm which utilizes a ² weighted angle maps for improving the accuracy of T1 measurements is presented. The performance of this technique is evaluated both on a gel phantom and rodent brain. The proposed method can improve both the acquisition and analysis of T1 values.

Daniel Polders¹, Alexander Leemans², Peter Luijten¹, and Hans Hoogduin^3
1Radiology, UMC Utrecht, Utrecht, Netherlands, ²Image Sciences Institute, UMC Utrecht, Utrecht, Netherlands, ³Rudolf Magnus Institute of Neuroscience, Department of Neurology and Neurosurgery, UMC Utrecht, Utrecht, Netherlands

This study utilizes a slice shifted EPI based T₁ mapping technique to calculate T₁maps over the human brain, sampling the inversion curve at 23 inversion times in 4 minutes 10 seconds. Then, wild bootstrapping was applied to determine the uncertainty of the fitted T₁values. This approach allows for simultaneous mapping of both T₁ and the uncertainty in T₁. Knowledge of regional T1 and its uncertainty is essential for interpretation if differences between regions of interest or subjects.

Michael J. Wilhelm¹, Henry H. Ong², and Felix W. Wehrli^2
1Department of Chemistry, Temple University, Philadelphia, PA, United States, ²Laboratory for Structural NMR Imaging, Department of Radiology, University of Pennsylvania School of Medicine, Philadelphia, PA, United States

Characterizing the T₂^* distribution of myelin is key to developing optimal ultra-short echo time (UTE) methods for direct myelin imaging. Previous attempts have used multi-exponential fitting of the FID, which theory shows is incorrect. Instead, myelin is a liquid crystalline lipid system that is described by a super-Lorentzian lineshape. Here, we use a super-Lorentzian framework to calculate T₂^* distributions from fits of ¹H NMR spectra of myelin lipid extract and intact rat spinal cord. Both distributions are similar and have an approximate T₂^* range from 10μs to 10ms. Despite this large range, ~50% of the myelin lipid signal exhibits T₂^*≤20μs

Kelvin J. Layton^1,2, Mark Morelande¹, Peter M. Farrell¹, Bill Moran¹, and Leigh A. Johnston^1,3
1Electrical and Electronic Engineering, University of Melbourne, Melbourne, Victoria, Australia, ²National ICT Australia, Melbourne, Victoria, Australia, ³Florey Neuroscience Institutes (Parkville), Melbourne, Victoria, Australia

The underlying distribution of relaxation times contributing to an MR signal can provide useful information about the structure of brain tissue. Recently, this distribution has been modelled by discrete T₂ components, which are estimated using a gradient-based least-squares optimisation algorithm. In this work, we demonstrate that this algorithm highly dependent on initialisation and SNR levels, often producing inaccurate estimates of the T₂ components. We propose a Bayesian algorithm that overcomes these limitations and provides reliable T₂ estimates at clinically achievable SNR. The algorithm is demonstrated through simulated and experimental data.

Se-Hong Oh^1,2, Young-Bo Kim², Zang-Hee Cho², and Jongho Lee^1
1Department of Radiology, University of Pennsylvania, Philadelphia, Pennsylvania, United States, ²Neuroscience Research Institute, Gachon University of Medicine and Science, Incheon, Korea

The contribution of tissue iron and the effect of temperature have been investigated for R2* (= 1/T2*) and R2* contrasts. The results suggest that iron has large contribution on R2* but not on R2* whereas temperature has significant effects on both contrasts

Ashish Raj¹, Kyoko Fujimoto², Thanh Nguyen³, and Susan Gauthier^2
1Radiology, Weill-Cornell Medical College, New York, NY, United States, ²Neurology, Weill-Cornell Medical College, New York, ³Radiology, Weill-Cornell Medical College, New York

T2-relaxometry, the numerical separation of differently relaxing tissue components in the brain, is a challenging problem due to ill-posedness, instability and extremely demanding requirement for SNR. The resulting myelin water fraction maps are noise, unstable and frequently diagnostically uninterpretable. Here, we completely side-step the inverse problem and instead separate the differentially relaxing components in the brain by principal components analysis (PCA), which is popular in fMRI analysis but has never before been used in T2 Relaxometry. This approach greatly improves computation speed, noise, spatial variations and definition of white matter in the brain.

Tilman Johannes Sumpf¹, Florian Knoll², Jens Frahm¹, Rudolf Stollberger², and Andreas Petrovic^2,3
1Biomedizinische NMR Forschungs GmbH am Max-Planck-Institut fuer biophysikalische Chemie, Goettingen, Germany, ²Institute of Medical Engineering, University of Technology, Graz, Austria, ³Ludwig Boltzmann Institute for Clinical Forensic Imaging, Graz, Austria

Quantitative evaluation of the T2 relaxation time is of high importance for diagnostic MRI. Standard T2 mapping procedures rely on the time-demanding acquisition of fully-sampled MSE datasets. Nonlinear inversion strategies allow for T2 mapping from undersampled data by exploiting a mono-exponential signal model. However, true MR data usually deviates from the idealized model which yields erroneous T2 estimations. A more accurate model of the MSE signal has been recently proposed with the generating function formalism (GF). This work evaluates the combination of the GF with a nonlinear inversion approach to allow for accurate T2 reconstructions from highly undersampled Cartesian data.

Se-Hong Oh^1,2, Sung-Yeon Park², John A. Detre¹, Young-Bo Kim², Zang-Hee Cho², and Jongho Lee^1
1Department of Radiology, University of Pennsylvania, Philadelphia, Pennsylvania, United States, ²Neuroscience Research Institute, Gachon University of Medicine and Science, Incheon, Korea

In this study, we developed a new approach to generate a Δ(0°-90°)R2* map (= R2*@parallel to B0 - R2*@perpendicular to B0 in each voxel) from one rotation and one DTI scans.

Tilman Johannes Sumpf¹, Amir Moussavi^1,2, Martin Uecker^1,3, Susann Boretius^1,4, and Jens Frahm^1
1Biomedizinische NMR Forschungs GmbH am Max-Planck-Institut fuer biophysikalische Chemie, Goettingen, Germany, ²DFG Research Center for Molecular Biology of the Brain (CMPB), Goettingen, Germany, ³Electrical Engineering and Computer Science, University of California, Berkeley, United States, ⁴Klinik für Diagnostische Radiologie, Universitätsklinikum Schleswig-Holstein, Kiel, Germany

Quantitative evaluation of the T2 relaxation time is of high importance for diagnostic MRI. Standard T2 mapping procedures rely on the time-demanding acquisition of several fully-sampled k-space data sets. We recently evaluated a new method that allows for the reconstruction of spin-density and T2 maps from highly undersampled Cartesian data by exploiting data redundancy in parameter space. However, the effect of periodic phase alternations due to unavoidable motion in in-vivo experiments can cause severe artifacts in reconstructions from undersampled data. This work explains the origin of these artifacts and demonstrates the impact on reconstructions from undersampled high-field animal MRI data.

Harsh Kumar Agarwal¹, Julien Sénégas², Baris Turkbey³, Marcelino Bernardo³, Tim Nielsen², Jochen Keupp², and Peter L Choyke^3
1Philips Research NA, Briarcliff Manor, New York, United States, ²Philips Research Europe, Hamburg, Germany, ³National Institute of Health, Molecular Imaging Program, National Cancer Institute, Bethesda, Maryland, United States

T2-weighted MRI has successfully been used in diagnostic prostate MRI. However, a clinically feasible T2 mapping technique is essential to develop a quantitative MR imaging framework for prostate oncology. This work proposes a novel image acquisition and image reconstruction protocol which exploit the temporal redundancy in the multi-echo fast spin echo imaging sequence used for T2 mapping along with coil sensitivity profile and partial-Fourier MRI to generate T2 maps over whole prostate on the scanner in clinically feasible time of 5min and 55 seconds.

Diego Hernando¹, and Scott B Reeder^1,2
1Department of Radiology, University of Wisconsin-Madison, Madison, WI, United States, ²Department of Medical Physics, University of Wisconsin-Madison, Madison, WI, United States

R2* measurement has multiple applications in MRI, including assessment of tissue iron overload. However, apparent R2* values are affected by the presence of fat. Fat introduces additional modulations in the acquired signal and results in large errors in R2* if not corrected. These errors can be addressed by including fat-water separation in the R2* measurement. In this work, we show that (i) fat correction is needed for robust (protocol-independent) R2* mapping, and (ii) including fat-water separation in the R2* measurement does not result in SNR penalty for R2* measurement over a wide range of true R2* values.

Debra E. Horng^1,2, Diego Hernando¹, Jens-Peter Kühn^1,3, and Scott B. Reeder^1,2
1Radiology, University of Wisconsin-Madison, Madison, WI, United States, ²Medical Physics, University of Wisconsin-Madison, Madison, WI, United States, ³Radiology and Neuroradiology, Ernst Moritz Arndt University Greifswald, Greifswald, Germany

R₂* relaxometry is a promising technique for liver iron quantification. However, measured R₂* values are affected by the presence of macroscopic B₀ inhomogeneities due to susceptibility effects. Susceptibility effects are modeled and measured to describe the B₀ distribution. Fieldmaps are measured, and simulated from susceptibility distributions (including air/water/fat) derived from patient data. Segments 4A, 7 and 8 (near dome) exhibit rapid field variation along ẑ, both in measured and simulated field maps. Sagittal acquisitions may be preferable to axial if R₂* measures near the liver dome are required. These methods may be used to optimize R₂* mapping acquisitions for liver iron quantification.

Fumiyuki Mitsumori¹, Hidehiro Watanabe¹, Nobuhiro Takaya¹, Michael Garwood², Edward Auerbach², Shalom Michaeli², and Silvia Mangia^2
1National Institute for Environmental Studies, Tsukuba, Ibaraki, Japan, ²University of Minnesota

We conducted measurements of apparent transverse relaxation rate at 1.5, 1.9, 3, 4.7, and 7T using a MASE sequence. The regional distribution of the relaxation rate was well described by a linear combination of the regional iron concentration [Fe], and the macromolecular mass fraction f_M defined as 1 - water fraction. The B₀-dependence of the [Fe] and f_M terms was linear and quadratic, respectively. The linear dependence of the [Fe] term was similar to that reported for ferritin solution. The quadratic dependence of the macromolecular term was well explained with an exchange model of water protons with those in macromolecules.