Hao Sun¹, Jeffrey A. Fessler¹, Douglas C. Noll², and Jon-Fredrik Nielsen^2
1Electrical Engineering and Computer Science, the University of Michigan, Ann Arbor, MI, United States, ²Biomedical Engineering, the University of Michigan, Ann Arbor, MI, United States

In 3D tailored RF pulse design, one typically predetermines a k-space (gradient) trajectory and then designs the corresponding RF waveforms for a target excitation pattern. Recently, the KT-points method was proposed as an approach for jointly designing the trajectory and RF pulses for 3D flip-angle homogenization (B1 shimming). KT-points models the 3D pulse design as a sparse approximation problem and selects sparse phase encoding locations by either a greedy approach or a simple inverse Fourier transform ignoring transmit coil sensitivity and field inhomogeneity. However, with only a few discrete phase encoding locations, it is difficult to approximate a non-smooth target excitation pattern in 3D. Also, it is relatively inefficient to traverse 3D k-space by discrete gradient blips with no RF transmission along those blips. In this work, we extend the KT-points method to a joint optimization of the continuous k-space trajectory and the RF waveform by: (1) applying local minimization to further optimize those KT points, and (2) efficiently ordering those points and generating a fast gradient waveform to traverse those points. We evaluate our proposed joint design with and without local minimization, and compare them with a recently proposed continuous nonselective spiral (SPINS) trajectory for 3D cubic excitation.

Rita Schmidt¹ and Lucio Frydman^1
1Chemical Physics, Weizmann Institute of Science, Rehovot, Israel

Two-dimensional (2D) excitation pulses are often used for localization in spectroscopic imaging and for in-plane region-of-interest delineation in MRI. Recent research has shown that these RF manipulations can also be based on spatiotemporal encoding (SPEN) principles. SPEN is a spatiotemporal manipulation that has also been used for single-shot ultrafast MRI. Fast volumetric SPEN MRI acquisitions, however, are still challenged. The present work merges the benefits of both 2D SPEN-based excitation and 2D SPEN single-shot acquisitions, demonstrating a multi-slice ultrafast sequence. Experiments testing these ideas were demonstrated on phantom as well as on brain volunteer imaging experiments at 3 T.

Kangrong Zhu¹, Adam B. Kerr¹, and John M. Pauly^1
1Electrical Engineering, Stanford University, Stanford, CA, United States

Multiband RF pulses are central to the signal excitation in simultaneous multislice acquisitions. The peak amplitude has been a limiting factor in multiband RF design, especially in multiband spin-echo pulse design. In this work, nonlinear-phase multiband pulses, which have reduced peak power compared to linear-phase pulses, are designed. A pair of 90°-180° nonlinear-phase multiband pulses are applied to generate a linear-phase echo. An additional reference phase is applied to each individual excited band to further reduce the peak power of the multiband pulse.

Ruud B van Heeswijk^1,2, Kieran R O'Brien^2,3, Jean Delacoste^1,2, and Matthias Stuber^1,2
1Radiology, University Hospital (CHUV) and University of Lausanne (UNIL), Lausanne, Switzerland, ²Center for Biomedical Imaging (CIBM), Lausanne, Switzerland,³Radiology, University of Geneva, Geneva, Switzerland

An adiabatic T₂ Preparation module (T₂Prep) at high magnetic field normally requires too much energy to allow it to be combined with bSSFP imaging. Numerically optimized adiabatic pulses were therefore used to design a T₂Prep for cardiovascular imaging at high magnetic field. The energy efficiency of this optimized T₂Prep was established at 3T and compared to standard adiabatic T₂Prep. Finally, T₂-prepared bSSFP cardiovascular imaging and coronary MRA were demonstrated in healthy volunteers.

Cornelius Eichner^1,2, Robert Turner², Lawrence L Wald¹, and Kawin Setsompop^1
1Martinos Center for Biomedical Imaging, Massachusetts General Hospital, Charlestown, MA, United States, ²Max Planck Institute for Human Cognitive and Brain Sciences, Leipzig, Saxony, Germany

Simultaneous Multi Slice (SMS) acquisition enables increased temporal resolution and acquisition speed. However, at ultra high field strengths, SAR constraints of SMS RF pulses can enforce slower acquisition speed. We propose a novel MultiPINS RF pulse design that combines PINS and MultiBand pulses to achieve lower SAR. Slice profiles and off-resonance behavior of this pulse were evaluated using Bloch simulations. The MultiPINS pulse achieves similar slice profiles, but significantly reduced energy transmission and peak RF voltage. In-vivo high-resolution Blipped-CAIPI SMS diffusion MRI data with 3x multiband acceleration were acquired at 7T to show the usefulness of this new pulse design.

Wei Feng¹, Yang Xuan², and E Mark Haacke^3
1Radiology, Wayne State University, Detroit, Michigan, United States, ²Radiology, Wayne State University, MI, United States, ³Wayne State University, MI, United States

A novel multi-dimensional spokes RF pulse design method is proposed to compensate for bulk susceptibility-induced phase variations. Under the small flip angle regime, there is a Fourier relationship between the excitation pattern and the RF and gradient waveforms, which traverses the excitation k-space. The conventional spokes pulse design approach is modified such that the cost function incorporates both magnitude and phase constraints inside the desired region of interest (ROI) based on susceptibility field map, while the phase is allowed to vary arbitrarily outside the ROI. Numerical Bloch simulations and imaging experiments were performed for 1D, 2D and 3D pulse design applications. It is shown that the proposed method is viable and could have significant potential in susceptibility-related imaging applications.

Fabian Zimmer¹, Frank G Zöllner¹, and Lothar R Schad^1
1Computer Assisted Clinical Medicine, Medical Faculty Mannheim, Heidelberg University, Mannheim, Germany

For pulsed arterial spin labeling (ASL) the quality of the slice-selective inversion is decisive because of the inherent low perfusion contrast. Commonly, adiabatic inversion pulses are used. However, the adiabatic condition has to be fulfilled, leading to a compromise between the maximum available coil voltage and profile quality, i.e. pulse parameters. Especially the high RF power demand of adiabatic dual-band inversion pulses normally leads to inversion profiles that are unusable for ASL. We present a dual-band FOCI inversion pulse with reduced RF power requirements that conserves the high quality of a single-band FOCI pulse.

Rebecca Emily Feldman¹, Haisam Islam², and Priti Balchandani^1
1Translational and Molecular Imaging Institute, Icah School of Medicine at Mount Sinai, New York, New York, United States, ²Bioengineering, Stanford University, California, United States

High field MR can be challenging due to limited slice coverage, B₁-inhomogeneity. Adiabatic pulses can limit sensitivity to B₁-inhomogeneity, however adiabatic pulses deposit quadratic phase that is difficult to refocus and are SAR intensive. Similarly, simultaneous multi-slice imaging can accelerate image acquisition at high fields but RF pulses created as the superposition of multiple slice RF pulses can rapidly exceed safe SAR limits. Using a 'Power Independent of Number of Slices' (PINS) technique, multiple slices can be excited simultaneously at lower power. We implemented a adiabatic PINS refocusing pulse with a matched phased PINS excitation pulse.

Nii Okai Addy¹ and Dwight G Nishimura^1
1Electrical Engineering, Stanford University, Stanford, CA, United States

A recent method for a flexible, numerical spiral imaging trajectory design can be adapted to RF pulse design. With this general framework, various VERSE RF pulses can be designed in one or multiple dimensions. This work presents results for slab-selection and spiral excitation.

Christopher J. Hardy¹, Seung-Kyun Lee¹, and Michael J. Wittbrodt^1
1GE Global Research, Niskayuna, NY, United States

Slice-select pulses can be made quieter by derating them, i.e. by reducing gradient slew rate and/or amplitude (along with RF bandwidth), but this increases minimum echo time. The variable-rate principle is used here to design slice–select pulses with improved acoustic signature and with identical slice profiles on-resonance, without lengthening pulse duration.

Christoph Stefan Aigner¹, Christian Clason², Armin Rund², and Rudolf Stollberger^1
1Institute of Medical Engineering, Graz University of Technology, Graz, Austria, ²Institute for Mathematics and Scientific Computing, University of Graz, Graz, Austria

Optimal control (OC) is a flexible framework for the design of RF pulses with arbitrary slice profiles, even in the presence of relaxation effects and field inhomogeneities, and is therefore well suited for simultaneous multislice (SMS) imaging. We demonstrate the ability of this approach to generate RF pulses with arbitrary (large) flip angles, slice thickness, slice gaps and slice numbers. The results for two and three slices of 4mm thickness are validated on a 3T MR scanner and indicate the applicability of the proposed method.

Emre Kopanoglu¹, Leo K. Tam¹, and Robert Todd Constable^1
1Dept. Diagnostic Radiology, Yale University School of Medicine, New Haven, Connecticut, United States

The effect of using nonlinear gradient fields (NLGFs) on excitation fidelity is studied, specifically on a multi-dimensionally selective excitation scheme. For this purpose, three nonlinear and two linear gradient fields (LGFs) are used. Two-dimensionally selective RF pulses are designed utilizing more than two fields simultaneously, using a multi-dimensional k-space approach. Using simulations, it is shown that increasing the number of k-space dimensions beyond the number of spatial coordinates may yield excitation profiles with lower error, compared to the target profile. It is also shown that, NLGFs may improve excitation fidelity, even for profiles that are more compatible with LGFs.

Joep Wezel¹, Maarten Versluis¹, Andrew Webb¹, Matthias van Osch¹, and Peter Börnert^1,2
1C.J. Gorter center for high field MRI, Radiology, Leiden University Medical Center, Leiden, Netherlands, ²Philips Research Europe, Hamburg, Germany

Neuroimaging at 7 tesla is complicated by the high degree of B1-inhomogeneity within the brain. Spezialized RF pulses that take the B1 distribution into account can compensate for the inhomogeneous field. These pulses are generally longer than the regular pulses, leading to increased sensitivity to B0 deviations. To counter this trend we apply high permittivity dielectric pads that reduce the severity of the flip angle voids. This potentially leads to the design of shorter RF pulses to compensate for the remainder of the inhomogeneities. We have simulated three pulse lengths with and without the bags and verified this in-vivo.

Wenwen Jiang^1,2, Michael Lustig³, John Pauly⁴, and Peder E.Z. Larson^5
1Graduate Group in Bioengineering, University of California, Berkeley, Berkeley, California, United States, ²University of California, San Francisco, San Francisco, California, United States, ³Electrical Enigneering and Computer Science, University of California, Berkeley, California, United States, ⁴Electrical Enigneering, Stanford University, California, United States, ⁵Radiology and Biomedical Imaging, University of California, San Francisco, California, United States

2D spiral excitation pulses are potentially valuable for bolus tracking and reduced FOV imaging. But 2D excitation pulses are usually long, given the FOV and resolution requirements, which results in off-resonance blurring of the spatial profile. Subsampled spiral trajectories could shorten the duration of the pulse but resulting in aliasing sidelobes in the excitation profile. In analogy to the subsampled data acquisition, subsampled excitation profiles can be designed to produce incoherent sidelobes. These can be further reduced by the fact that spin-echoes square the linear excitation dramatically shrinking the sidelobes. With the design of appropriate variable density spiral trajectories, this method will effectively suppress aliasing sidelobes while resulting in shorter excitation pulses.

Mathias Davids^1,2, Bastien Guérin², Lothar R. Schad¹, and Lawrence L. Wald^2,3
1Computer Assisted Clinical Medicine, Medical Faculty Mannheim, Heidelberg University, Mannheim, BW, Germany, ²Martinos Center for Biomedical Imaging, Dept. of Radiology, Massachusetts General Hospital, Charlestown, MA, United States, ³Harvard-MIT, Division of Health Sciences and Technology, Cambridge, MA, United States

The k-space trajectory, in addition to the RF, possesses powerful degrees of freedom to enhance 3D parallel selective excitations. A novel approach on rapidly designing arbitrarily shaped time-optimal trajectories was used to simultaneously optimize the trajectory and RF pulse. The trajectory was defined by shape parameters that were optimized for a cubic ROI and brain only excitation. Two trajectories were optimized – a 3D Cross and a 3D Concentric Shells trajectory – with durations of less than 7 ms each. The excitation RMSE could be reduced by up to 60% in an eight channel 7T setup, yielding applicable 3D selective excitation pulses.

Nicolas Boulant¹, Andres Hoyos-Idrobo¹, Pierre Weiss², Aurelien Massire¹, and Alexis Amadon^1
1Neurospin, CEA, Saclay, Ile de France, France, ²ITAV, CNRS, Toulouse, Midi-Pyrénées, France

Despite the importance of the magnitude least squares problem in parallel transmission pulse design and the availability of other powerful numerical optimization methods, this problem has been faced almost exclusively with the so-called variable exchange method. Here, we investigate various two stage strategies and incorporate directly the SAR and power constraints. Different schemes such as sequential quadratic programming, interior point methods, semi-definite relaxation and magnitude squared least squares relaxations are studied in the small and large flip angle regimes with B1 and DB0 maps obtained in-vivo on a human brain at 7 Tesla.

Filiz Yetisir¹, Bastien Guerin², Lawrence L. Wald^2,3, and Elfar Adalsteinsson^1,3
1Dept. of Electrical Engineering and Computer Science, Massachusetts Institute of Technology, Cambridge, MA, United States, ²Martinos Center for Biomedical Imaging, Dept. of Radiology, Massachusetts General Hospital, Charlestown, MA, United States, ³Harvard-MIT Division of Health Sciences Technology, Cambridge, MA, United States

Explicit SAR constraints have been proven useful for slice selective small tip angle pulse design. We propose a nonselective pulse design method that explicitly constrains local SAR and RF peak amplitude at large tip angle and demonstrate that when local SAR is controlled directly rather than via control of peak RF voltage, safer pulses with better excitation profiles are obtained. Our method is more practical than Tikhonov regularized strategies since it only requires one run to ensure that all the limits (SAR and RF) are satisfied.

Daniel Brenner¹, Desmond H. Y. Tse², Eberhard D. Pracht¹, Thorsten Feiweier³, Rüdiger Stirnberg¹, and Tony Stöcker^1
1German Center for Neurodegenerative Diseases (DZNE), Bonn, Germany, ²INM-4, Research Centre Jülich GmbH, Jülich, Germany, ³Siemens AG, Healthcare Sector, Erlangen, Germany

A 3D variant of the DREAM sequence, with a spiral phase encode view ordering, is utilized for B1 mapping at 7T. Together with short non-selective preparation and imaging RF pulses this enables whole volume B1 mapping of the human head in 15s or even a single shot - which only lasts 3s – at negligible SAR levels (1%). Good agreement is found with a reference AFI dataset with degraded quality in a low tip angle regime due to the low SNR of the STE* image.

Qi Duan¹, Peter van Gelderen¹, Souheil J. Inati², and Jeff H. Duyn^1
1AMRI, LFMI, National Institute of Neurological Disorders and Stroke, National Institutes of Health, Bethesda, Maryland, United States, ²FMRIF, National Institute of Mental Health, National Institutes of Health, Bethesda, Maryland, United States

This abstract investigates fast simultaneous B₀/B₁⁺ mapping by Bloch-Siegert shift via lowering the off-resonance frequency of this pulse, as theoretical analysis indicated that the sensitivity of Bloch-Siegert based B1+ mapping can be substantially improved when irradiating closer to resonance. Using optimized irradiation pulse shape and gradient crushers to minimize direct excitation effects, in vivo experiments on human brain at 7T confirmed the improved sensitivity available with this approach operating with peak B₁⁺ much larger than the frequency offset. This improved sensitivity translated into an 80% reduction in B₁⁺ estimation errors, without increasing tissue heating.

Parnian Zarghamravanbakhsh¹, Christopher Ellenor¹, John M Pauly¹, and Greig Scott^1
1Electrical engineering, Stanford university, Stanford, CA, United States

We propose a method to predict B1 filed using coil geometry and location in imaging coordinate system without doing B1 mapping. Most of B1 mapping are done without any assumption about the coil geometry and location. Coils location are found by placing markers on coil conductors, then by acquiring three sets of 1D projections to localize the markers ,coil plane can be detected. Having known the coil location and geometry, B1 field distribution can be obtained by computational analysis .To validate the method, we compare simulated with measured results. Field prediction can be used in auto-calibration and RF pulse design.

Kalina V Jordanova¹, Dwight G Nishimura¹, and Adam B Kerr^1
1Electrical Engineering, Stanford University, Stanford, California, United States

We redesign the BEAR B₁ mapping method to use HSn pulses, which have lower adiabatic thresholds. By optimizing the HSn pulse parameters, we can reliably acquire B₁maps for lower nominal peak B₁ than with the original BEAR method. We validate the performance of BEAR with HSn pulses via simulation and in vivo at 3T, with average errors from the original BEAR method of less than 3%. This method will be useful for reliably acquiring B₁ maps for lower B₁ values.

Daniel J. Park¹, Neal K. Bangerter^1,2, and Glen R. Morrell^2
1Department of Electrical and Computer Engineering, Brigham Young University, Provo, Utah, United States, ²Department of Radiology, University of Utah, SLC, Utah, United States

B₁ mapping mapping is an important component of quantitative MRI and parallel transmission. Although many B₁ mapping methods have been introduced and analyzed, there is no clear superior method. One method, the Bloch-Siegert shift method, has potential for improved B₁ mapping of parallel transmit arrays through separation of excitation and the B₁ encoding pulse. We introduce a modification to the Phase Sensitive (PS) method that allows similar improvement by decoupling the compound excitation pulse in the PS method. We introduce a brief Monte Carlo based statistical analysis (mean bias and standard deviation) which illustrates the potential of this method.

Qinwei Zhang¹, Yi-Xiang J Wang¹, Heather Ting Ma², Queenie Chan³, and Jing Yuan^1,4
1Department of Imaging and Interventional Radiology, The Chinese University of Hong Kong, Shatin, N.T., Hong Kong, ²Harbin Institute of Technology Shenzhen Graduate School, Shenzhen, Guangdong, China, ³Philips Healthcare, Hong Kong, Hong Kong, ⁴CUHK Shenzhen Research Institute, Shenzhen, Guangdong, China

T1� imaging has been proved to be a sensitive biomarker for disc degeneration while suffering from long scan time. Large field of view and respiratory motion further adversely affect the mapping results. We proposed a 2DRF pulse with novel high-low EP excitation trajectory to realize reduced field of view (rFOV) T1� imaging in lumbar spine region on a 3T clinical scanner. The scan time was halved and motion artifact was eliminated. Good consistency between rFOV and full FOV T1� maps was observed. The proposed 2DRF has potential to be used for high-resolution spine T1� imaging in routine clinical scan.

Yuval Zur^1
1GE Healthcare, Tirat Carmel, Israel

Two dimensional (2D) RF pulses with EPI excitation trajectory are extremely sensitive to system imperfections such as eddy currents and waveform distortions. These imperfections cause stop band excitation and pass band saturation. A method to correct these 2D RF pulses is presented. The correction is done by adding a phase to even sub pulses and gradient blips to the oscillatory excitation gradient. The added phase and the area of the blips are determined by a calibration done at system installation. The method was applied successfully to spectral spatial RF pulses at oblique and non oblique slice orientations.

M Louis Lauzon^1,2 and Richard Frayne^1,2
1Radiology, Hotchkiss Brain Institute, University of Calgary, Calgary, AB, Canada, ²Seaman Family MR Research Centre, Calgary, AB, Canada

The flip angle term in an image can be determined using the TR-interleaved AFI (actual flip angle imaging) sequence. AFI makes an approximation, which leads to an inherent bias (a measure of accuracy). Noise propagation, however, produces an uncertainty bias and an increase in the variance (a measure of precision). Here, we analytically and numerically determine the accuracy/precision of the inherent and noise-induced errors of the cosine of the flip angle as a function of various acquisition parameters, and provide an overall error.

Rainer Schneider^1,2, Jens Haueisen², and Josef Pfeuffer^1
1MR Application Development, Siemens Healthcare, Erlangen, Bavaria, Germany, ²Institute of Biomedical Engineering and Informatics, TU Ilmenau, Ilmenau, Thuringia, Germany

Shaped saturation human in-vivo experiments were realized for the first time on the basis of multidimensional spatially selective RF pulses in parallel transmission. For this purpose, a variable-density 2D spiral trajectory design was introduced, which offers inherent RF power efficiency by incorporating the a-priori information of the target pattern and available B1 magnitude. The design was evaluated in sagittal head and t-spine experiments, saturating different patterns at 3T. The proposed approach was shown to offer up to 32% improved spatial accuracy and saturation performance under given RF hardware and SAR constraints.

Tobias C Wood¹, Samuel A Hurley², Gareth J Barker¹, and Steven C R Williams^1
1Neuroimaging, King's College London, Institute of Psychiatry, London, London, United Kingdom, ²Medical Physics, University of Wisconsin, Wisconsin, United States

We present an SSFP signal equation that includes finite RF pulse lengths and exchanging components, and then use this to calculate Myelin Water Fraction maps using mcDESPOT.

Tiffany Jou¹, Steve Patterson², John M. Pauly¹, and Chris Bowen^2
1Electrical Engineering, Stanford University, Stanford, CA, United States, ²Physics and Atmospheric Science, Dalhousie University, Halifax, NS, Canada

Alternating-SSFP (alt-SSFP) suppresses banding artifacts and allows for whole-brain fMRI in a single run by using RF catalyzation to alternate between RF phase-cycling steady states. However, bright fat signal remains a problem because of the short TR used. Without an effective fat suppression technique, artifacts like fat chemical shift and off-resonance signal instability result. Our goal was to design a spectral-spatial (SPSP) pulse short enough to be used for alt-SSFP fMRI, with reduced fat chemical shift artifacts and improved temporal SNR (tSNR) time-courses. In this study, we propose our SPSP fat-suppression method as the first practical implementation of alt-SSFP and demonstrate good BOLD sensitivity in both breath-hold and visual paradigms.

Jonathan Phillips¹ and Sophie Schirmer^2
1Institute of life Science, College of Medicine, Swansea University, Swansea, Wales, United Kingdom, ²Physics Department, Swansea University, Swansea, Wales, United Kingdom

It has been suggested that the UDD sequence, originally developed to extend the lifetime of qubits in quantum computing, may be used as an imaging sequence to extend the coherence lifetime in tissues to elucidate different tissue structures. We performed phantom studies to investigate these claims. We find evidence of lifetime extension although the interpretation of the images is complicated by steady state effects i.e. banding. However, the sequence may elucidate three-dimensional structure near tissue boundaries as well enhancing fat-water contrast at large repetition times.

Guobin Li¹, Maxim Zaitsev¹, Matthias Weigel², Esther Meyer³, Dominik Paul³, Jan Korvink^4,5, and Jürgen Hennig^1
1University Medical Center Freiburg, Freiburg, Baden-Württemberg, Germany, ²Radiological Physics, University of Basel Hospital, Switzerland, ³Siemens Healthcare, Germany, ⁴Department of Microsystems Engineering — IMTEK, University of Freiburg, Germany, ⁵Freiburg Institute of Advanced Studies (FRIAS), University of Freiburg, Germany

SPACE employs non-selective RF pulses to achieve short echo spacing in the imaging with long echo trains. However, the use of non-selective RF pulses results in a single slab acquisition in SPACE imaging. In typical SPACE protocols, the duration of the echo train is only about 20% of the TR, which means 80% of the time is purely for waiting. In order to improve the time utilization ratio, in this work, we investigate the feasibility of acquiring a second contrast in the waiting time of the TR in single slab SPACE imaging. Some preliminary results of the hybrid imaging are shown.

Bing Wu¹ and Yongchuan Lai^1
1GE healthcare, Beijing, Beijing Municipality, China

A new pulse sequence named FIESTA-flex is proposed. Comparing to FIESTA(bSSFP), it removes the banding artifacts and allows flexibility image contrast at the sacrifice of signal level compared to FIESTA. In vivo feasibility studies have been performed, and qualitative inspection agrees with theoretical expectation.

Maxwell L Wong¹, Eric Z.C. Wu², and Eric C Wong^3
1UC San Diego, La Jolla, California, United States, ²University of Southern California, California, United States, ³Department of Radiology and Psychiatry, UCSD, La Jolla, California, United States

In this study, we simulate and attempt to optimize parameters for MR Fingerprinting. We looked for ways to characterize flip angle and TR schedules that produce the best estimated T1, T2 and frequency maps. This was done using frequency filters applied to random schedules, and comparison to DESPOT1 and DESPOT2 to assess the effectiveness of MRF. We confirmed higher sensitivity of MRF compared to DESPOT, and identified frequency characteristics in the flip angle schedules that were beneficial.

Nahal Geshnizjani¹, Kenneth A. Loparo¹, Dan Ma², Debra McGivney³, Vikas Gulani^2,3, and Mark A. Griswold^2,3
1Electrical Engineering and Computer Science, Case Western Reserve University, Cleveland, Ohio, United States, ²Biomedical Engineering, Case Western Reserve University, Cleveland, Ohio, United States, ³Radiology, University Hospitals of Cleveland and Case Western Reserve University, Cleveland, Ohio, United States

The purpose of this work is to design a system that is able to extract basic MR sequence parameters such as FA and TR from TrueFIsp signal evolutions. Artificial Neural Networks are used as the main tool because of their ability to be trained and learn and then solve complicated mathematical equations. We use an efficient method to predict FAs of TrueFISP signal evolutions one excitation at a time using the magnetization preceding and following the excitation. ANNs are trained by arbitrary initial magnetizations and random flip angles.

Sebastian Littin¹, Feng Jia¹, Hans Weber¹, Frederik Testud¹, Anna Welz¹, and Maxim Zaitsev^1
1Dept. of Radiology, Medical Physics, University Medical Center Freiburg, Freiburg, Germany

Singular Value Decomposition (SVD) is a very usefull tool to asses encoding fields from matrix gradient coils.

David Manuel Grodzki¹ and Bjoern Heismann^1,2
1Magnetic Resonance, Siemens Healthcare, Erlangen, Bavaria, Germany, ²Pattern Recognition Lab, Friedrich-Alexander-University of Erlangen-Nuremberg, Erlangen, Bavaria, Germany

Acoustic noise is one of the main reasons for patient discomfort during an MRI examination. High noise levels are caused by fast-switching gradients during the scan. By reducing the gradient switching and optimizing gradients, significant acoustic-noise reductions can be achieved. In this work, we present an automated gradient conversion algorithm that optimizes the gradient shape of any incoming sequence on the fly. Noise reductions of up to 12 dB(A) were reached. Incorporating further careful protocol adaptions, further noise reduction of up to 10 dB(A) was achieved, without sacrificing diagnostic image quality.

Xiaobo Qu¹, Ying Chen¹, Xiaoxing Zhuang¹, Zhiyu Yan¹, Di Guo², and Zhong Chen^1
1Department of Electronic Science, Xiamen University, Xiamen, Fujian, China, ²School of Computer and Information Engineering, Xiamen University of Technology, Xiamen, Fujian, China

To accelerate imaging, compressed sensing MRI (CS-MRI) suggests performing randomly undersampling to reduce the coherence between the encoding matrix and the sparsity bases. Spread spectrum (SS) is recently introduced to improves the reconstruction by reducing this coherence. But SS is achieved via a second order shim coil which limits modulation intensity and is not convenient to be operated. In this work, we propose a chirp radio frequency (RF) pulses to easily control the spread intensity by choosing a proper bandwidth and apply in CS. Simulation on the sampled data implies that the reconstruction error reduces if a proper bandwidth is provided.

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

The use of a spatial-spectral pulse in the presence of a static non-linear shim gradient has been proposed for reduced FOV excitation, in particular for exciting thin disc-shaped regions at isocenter. In this work, we extend the method to excite regions at an arbitrary locations and of arbitrary slice thicknesses, allowing for accelerated 2D or 3D high-resolution imaging.

Bryant T. Svedin^1,2 and Dennis L. Parker^1
1Utah Center for Advanced Imaging Research, Radiology, University of Utah, Salt Lake City, Utah, United States, ²Physics, University of Utah, Salt Lake City, Utah, United States

Measurements of T1 using the Variable Flip Angle method are subject to errors introduced by inaccuracy in the flip angle used. Simulations were performed to test the effects of the slice excitation profile on the dependence of the measured signal on flip angle. Excitation profiles for several TBP and T1 values were simulated using the steady state flash equation. Calculated T1 values are compared with the true values.

Xiaodong Guo^1
1Brain Research Imaging Center, The University of Chicago, Chicago, IL, United States

The slice thickness of slices in a small region of the human brain, where large MRI signal loss was experienced due to susceptibility difference at the air-tissue interfaces, was set to half the thickness of slices located at other regions where the static magnetic field was homogeneous. The MRI signal loss was dramatically recovered. Compared with z-shimming technique, images acquired by this volume-selective thin slice thickness technique have lesser temporal signal to noise ratio. However, thin slice thickness technique worked better for slices close to brain stem which cover amygdala and hippocampus areas.

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

High resolution 3D MRI is becoming increasingly desirable for many research and clinical applications. However, certain MRI pulse sequences (such as EPI) can only be performed in multi-slice mode, typically with inadequate through-plane resolution. We present a technique that allows scalable through-plane resolution enhancement for multi-slice acquisitions. The method employs multi-slice acquisitions staggered along Z refined with novel deblurring algorithms. Isotropic resolution in 3D is achievable when sufficient data are available.

Rita Schmidt¹, Amir Seginer¹, and Lucio Frydman^1
1Chemical Physics, Weizmann Institute of Science, Rehovot, Israel

Recent studies have shown the benefits of a single-shot spatiotemporal encoding (SPEN): reducing magnetic field inhomogeneity distortions and delivering chemical shift information. The present work demonstrates that SPEN can also exhibit substantial motion immunity, even when executed in interleaved multi-shot schemes. Shots of SPEN data can be co-processed to generate full FOV images without requiring extra reference scans. Since no aliasing is involved, a phase correction from SPEN shots can be estimated, resolving ghost-free SPEN images, and eliminating transient phase shifts or in-plane movement between shots. Preliminary tests confirm the advantages on phantom and human scans, including functional MRI experiments.

Anuj Sharma¹, Manus Donahue^2,3, V. Andrew Stenger⁴, and William A. Grissom^1,2
1Biomedical Engineering, Vanderbilt University, Nashville, TN, United States, ²Radiology, Vanderbilt University, TN, United States, ³Psychiatry, Vanderbilt University, TN, United States, ⁴Medicine, University of Hawaii, HI, United States

An SNR-efficient method to mitigate signal loss artifacts in single-slice and simultaneous multi-slice long echo time gradient echo acqisitions is presented. Signal improvement comes from the use of spectral spatial pulses to selectively excite regions that are refocused with Z-shim. Scan time is minimized by treating the different z-shim acquisitions as additional slices in a multi-slice stack. Phantom and in-vivo experiments at 7T and 3T demonstrate the effectiveness of the proposed method in reducing signal loss artifacts in both single-slice and multiband exams.

Tangi Roussel¹ and Lucio Frydman^1
1Department of Chemical Physics, Weizmann Institute of Science, Rehovot, Israel

Since 2010, ultrafast NMR is applied for MRI giving birth to several ultrafast single-shot SPatio-temporally ENcoded (SPEN) imaging sequences. Besides important scan time reduction, SPEN experiments are especially robust regarding high-field artifacts such as B0 inhomogeneities and susceptibility effects. Zooming abilities are also built-in into this kind of experiments. In this paper, we present a SPEN method developed for Bruker MRI systems. The method includes single-shot single-slice, multi-slice SPEN and RASER sequencing options; all with an online reconstruction and fully integrated in Bruker Paravision as a “method”.

Danny Kyejun Lee¹, Sean Pollock¹, Peter Greer², Taeho Kim¹, and Paul Keall^1
1Radiation Physics Laboratory, Sydney Medical School, The University of Sydney, Camperdown, NSW, Australia, ²The University of Newcastle, NSW, Australia

The dynamic keyhole method using respiratory signals has been demonstrated to reconstruct MR images with the considerably small amount of central phase encoding lines, linked to real-time thoracic imaging with minimal image intensity loss on tumor. These results suggest that the dynamic keyhole method could be a desirable technique for image-guided radiation therapy and MRI-guided radiotherapy that require real-time MR monitoring in thoracic region.

Xingfeng Shao¹, Xucheng Zhu¹, Feiyu Chen², and Kui Ying^3
1Department of Engineering Physics, Tsinghua University, Beijing, China, ²Department of Biomedical Engineering, Tsinghua University, Beijing, China, ³Department of Engineering physics, Tsinghua University, Beijing, China

In abdominal imaging, non-rigid and continuous motion artifacts can be introduced because of breathe and intestinal peristalsis movement. To reduce this kind of motion artifacts, we use COCOA to detect and discard motion corrupted data under PROPELLER trajectory, then use SPIRiT to reconstruct the k-space. To evaluate our method, simulated data was generated by adding non-rigid continuous motion to a reference image. After applying both proposed method and traditional COCOA to the simulated data, results show that simulated motion can be effectively reduced with our method comparing to traditional COCOA. Conclusion can be made that combining COCOA with PROPELLER and SPIRiT can effectively reduce non-rigid continuous motion in abdominal imaging, which is better than traditional COCOA.

Guobin Li¹, Maxim Zaitsev¹, Martin Büchert¹, Esther Meyer², Dominik Paul², Jan Korvink^3,4, and Jürgen Hennig^1
1University Medical Center Freiburg, Freiburg, Baden-Württemberg, Germany, ²Siemens Healthcare, Germany, ³Department of Microsystems Engineering — IMTEK, University of Freiburg, Baden-Württemberg, Germany, ⁴Freiburg Institute of Advanced Studies (FRIAS), University of Freiburg, Baden-Württemberg, Germany

3D turbo spin echo sequences (e.g. SPACE) suffer from long acquisition times and are therefore prone to motion artifacts. Two features are introduced into SPACE imaging: a) integrated motion detection, which judges whether a patient’s movement is tolerable or not, and stops the acquisition immediately when unacceptable motion is detected; b) a dedicated sampling strategy, which optimizes the image quality in cases that an acquisition is interrupted. Comprehensive in vivo experiments have been conducted to evaluate the performance of the method.

Yajun Ma¹, Wentao Liu¹, Yang Fan¹, Huanjie Li¹, and Jia-Hong Gao^1
1MRI Research Center and Beijing City Key Lab for Medical Physics and Engineering, Peking University, Beijing, Beijing, China

Date processing techniques such as multiple average methods and COCOA have been developed recently for motion artifacts reducing. They employed convolution of k-space data to reduce localized data inconsistencies. These data processing techniques can be incorporated with other techniques, such as respiratory gating, navigator echoes.And then, better image quality is obtained. And they can also be used alone for free breath imaging. A new data processing strategy is introduced in this work to optimize the date convolution procedure and take care of different motion characters exist in multi-coil images to protect the image SNR.

Judith Biermann¹, Martin Krämer¹, and Jürgen R Reichenbach^1
1Medical Physics Group, Institute of Diagnostic and Interventional Radiology I, Jena University Hospital - Friedrich Schiller University Jena, Jena, TH, Germany

To perform multi-slice multi-echo abdominal imaging during free breathing a radial golden angle self-gating technique was used applying 1D projection navigators in an automatically added slice. 1D navigators were utilized to generate a breathing trigger signal through correlation analysis. Since each time point of the trigger signal is directly related to a radial readout in all slices, respiratory phases could be excluded from image reconstruction for obtaining breathing corrected images. Image blurring caused by respiratory motion was highly reduced.

R Reeve Ingle¹, Juan M Santos², William R Overall², Bob S Hu^1,3, and Dwight G Nishimura^1
1Electrical Engineering, Stanford University, Stanford, California, United States, ²HeartVista, Inc., Menlo Park, California, United States, ³Palo Alto Medical Foundation, Palo Alto, California, United States

A technique for fat-suppressed self-gated cardiac cine imaging is demonstrated, which enables cardiac cine imaging without the need for external electrocardiogram (ECG) gating. Fat suppression is achieved using an alternating repetition time (ATR) balanced steady-state free precession (bSSFP) pulse sequence. By redesigning the slice-select rephaser gradients, one-dimensional projection navigators can be acquired during the unused short TR interval. Volunteer and patient results are presented and compared with ECG-gated ATR and bSSFP acquisitions.

Malek I Makki¹, Christoph Ruegger², Cyrille Capel³, Catherine Gondry-Jouet⁴, and Olivier Baledent^5
1MRI Research Center, University Children Hospital of Zurich, Zurich, Switzerland, ²Neonatalogy, University Hospital of Zurich, Zurich, Switzerland, ³Neurosurgery, University Hospital, Amiens, France, ⁴Radiology, University Hospital, Amiens, France, ⁵Image Processing, University Hospital, Amiens, France

An accurate CSF flow measurement through the pontine cistern is difficult to achieve because of blood flow artifacts from the basilar artery and surrounding vessels. We developed a dedicated PCMRI sequence with pre-saturation double sided bands to suppress signal from blood with no compromise neither on SNR nor on spatio-temporal resolution and validated this on 18 patients with hydrocephalus. We compared the results of the developed sequence in the pontine cistern and foramen of Magendi which is free of blood flow artifact and demonstrated its accuracy and reliability to measuring the stroke volume and the minimum and maximum flow.

Ville Renvall^1,2, Thomas Witzel^1,2, Marta Bianciardi^1,2, and Jonathan R. Polimeni^1,2
1Athinoula A. Martinos Center for Biomedical Imaging, Massachusetts General Hospital, Charlestown, MA, United States, ²Department of Radiology, Harvard Medical School, Boston, MA, United States

Multi-contrast inversion-recovery EPI (MI-EPI) was implemented on a 7T system and was used to study brain activations simultaneously using T₁-, BOLD-, and multiple inversion-recovery-time-signal changes following visual stimulation at a 3-s sampling period. The different contrast images yielded similar activation maps, with extent of activation roughly proportional to the signal levels, but less activation with CSF suppressed rather than gray matter suppressed contrasts. T₁ values were found to significantly increase during visual stimulation concordant with blood volume increases and/or inflow effects.

Liyong Chen^1,2, Alexander Beckett^1,2, Ajay Verma³, and David Feinberg^1,2
1Helen Wills Neuroscience Institute, University of California, Berkeley, California, United States, ²Advanced MRI Technologies, LLC, Sebastopol, California, United States,³Biogen Idec, MA, United States

A new highly efficient velocity imaging technique is developed with simultaneous multi-slice EPI and zoomed spatial resolution which enables real-time measurement of CSF and brain velocity. Application of this 7D imaging technique to normal subjects revealed respiratory synchronous motion modulating cardiac waveforms in brain parenchyma and CSF. The brain pulsations may play a role in clearance of interstitial fluid in the brain.

Robert Trampel¹, Jochen Schmidt¹, Laurentius Huber¹, Andreas Schäfer¹, and Robert Turner^1
1Max Planck Institute for Human Cognitive and Brain Sciences, Leipzig, Germany

Although the contrast between gray and white matter obtained by spin-echo and turbo spin-echo techniques is usually referred to as “T2-weighted”, transverse relaxation mechanisms contribute hardly at all to the contrast between those brain tissue types. We found that gray-white contrast in spin-echo brain images at 7T arises mainly from proton density and T1 relaxation, depending on TR. Magnetization transfer has a somewhat smaller influence, while the influence of T2 variations in tissue is negligible.

Won-Joon Do¹, Paul Kyu Han¹, Seung Hong Choi², and Sung-Hong Park^1
1Department of Bio and Brain Engineering, Korean Advanced Institute of Science and Technology, Daejeon, Korea, ²Seoul National University Hospital, Seoul, Korea

In this study we implemented a new dual-echo sequence for simultaneous acquisition of T1 and T2* weighted 3D Anatomical images. An echo-specific K-space reordering scheme was used to separately satisfy T1 and T2* contrast for the two echoes, which was determined to be 30 and 10-20 for the first echo and second echo, respectively. The results showed that the proposed method enables us to acquire both 3D T1 and T2* weighted images with scan time of ~3 min and a reasonable spatial coverage and resolution. The technique may be helpful for accelerating routine clinical studies requiring T1 and T2* acquisitions.

Katharina Fuchs¹, Fabian Hezel¹, Sabrina Klix¹, Ralf Mekle², Jens Wuerfel^3,4, and Thoralf Niendorf^1,5
1Berlin Ultrahigh Field Facility (B.U.F.F.), Max-Delbrueck Center for Molecular Medicine, Berlin, Germany, ²Medical Metrology, Physikalisch Technische Bundesanstalt, Berlin, Germany, ³Institute of Neuroradiology, University Medicine Goettingen, Germany, ⁴NeuroCure Clinical Research Center, Charité - University Medicine Berlin, Germany,⁵Experimental and Clinical Research Center (ECRC), a joint cooperation between the Charité Medical Faculty and the Max-Delbrueck Center, Berlin, Germany

2in1-RARE is a RARE variant which is simultaneously sensitive to T₂^* and proton density contrast. This is achieved by strictly separating spin echo and stimulated echo magnetization within a RARE echo train. The performance of 2in1-RARE is elucidated using point spread function assessment. 2in1-RARE traits for T₂^* mapping are validated against conventional mulit-echo gradient echo acquisitions. The applicability of dual contrast weighted double echo 2in1-RARE is demonstrated for brain imaging using susceptibility weighted imaging, T₂^* mapping and proton density weighted imaging in healthy subjects and multiple sclerosis patients.

Melisa Okanovic^1,2, Martin Blaimer², Felix Breuer², and Peter Michael Jakob^2,3
1Comprehensive Heart Failure Center, University Hospital of Wuerzburg, Wuerzburg, Bavaria, Germany, ²Magnetic Resonance Bavaria (MRB), Wuerzburg, Bavaria, Germany, ³Department of Experimental Physics 5, University of Wuerzburg, Wuerzburg, Bavaria, Germany

For high resolution images of the human brain a hybrid sequence with a radial readout is presented. In this technique gradient refocused echoes replace refocusing radio-frequency pulses up to a certain level. This offers a significant SAR-reduced imaging compared to a turbo-spin-echo technique with the same echo-train-length. The radial k-space acquisition and a view-sharing technique (KWIC) for the reconstruction, allow several arbitrary T2-weighted images. High resolution in-vivo human brain images are presented.

Jakob Kreutner¹, Andreas J. Hopfgartner², Julian Boldt³, Kurt Rottner³, Ernst J. Richter³, Peter M. Jakob^1,2, and Daniel Haddad^1
1MRB-Research Center Magnetic-Resonance-Bavaria, Würzburg, Germany, ²Experimental Physics 5, University of Würzburg, Würzburg, Germany, ³Prosthodontics, Dental School, University of Würzburg, Würzburg, Germany

Recent studies have shown the accuracy of a VIBE sequence in visualizing the mandibular canal compared to X-ray based methods. In order to increase resolution without unnecessarily lengthen acquisition time a small field of view is needed. This can be achieved by using regional saturation bands in VIBE or local look technique in TSE sequences to avoid aliasing artifacts. Since TSE offers higher SNR and in combination with local look technique is more comfortable to set up, we compared both methods by calculating the surface difference of the mandibular canal.

Andrew L Alexander¹ and Steven R Kecskemeti^1
1Waisman Center, University of Wisconsin, Madison, WI, United States

An efficient 3D brain imaging pulse sequence was developed with a pair of inversion pulses that simultaneously acquire both single inversion recovery (SIR) and double inversion recovery (DIR) image data. Sampling was performed using a spoiled gradient echoes with a 3D radial k-space readout. A sliding window reconstruction was used to generate multiple image volumes with different single and double inversion times. The SIR volumes included a nulled gray matter frame and a nulled white matter frames and also more standard T1-weighted contrast. The DIR sampling yielded a gray matter specific map with nulling of both white matter and CSF.

Anup Singh¹, Ravi Prakash Nanga¹, Mohammad Haris^1,2, Kejia Cai^1,3, Felik Kogan¹, Hari Hariharan¹, and Ravinder Reddy^1
1Radiology, University of Pennsylvania, Philadelphia, PA, United States, ²Research Branch, Sidra Medical and Research Center, Doha, Qatar, ³Radiology, University of Illinois at Chicago, Chicago, IL, United States

Objective of current study was to develop a new method for mapping molecular integrity of cartilage based upon Transverse Relaxation Amplified by Chemical Exchange (TRACE). This simple and time efficient method provides a quantitative index of macromolecular content of cartilage and is less sensitive to fluid changes associated with pathological conditions of the tissue. The method was tested in chondroitin sulfate (CS) phantoms with different concentrations and pH as well as on human knee cartilage at 3T and 7T whole body MR scanners.

Samo Lasic¹, Savannah C. Partridge², Cheng-Liang Liu², Stina Oredsson³, Lao Saal⁴, Daniel Topgaard⁵, Markus Nilsson⁶, and Karin Bryskhe^1
1CR Development AB, Lund, Sweden, ²Dept. of Radiology, University of Washington, Seattle Cancer Care Alliance, Seattle, United States, ³Department of Biology, Lund University, Lund, Sweden, ⁴Department of Oncology, Lund University Hospital, Lund, Sweden, ⁵Center for Chemistry and Chemical Engineering, Lund University, Lund, Sweden, ⁶Lund University Bioimaging Center, Lund University, Lund, Sweden

A feasibility study of filter exchange imaging (FEXI) for detection of apparent exchange rate (AXR) was performed on breast cancer cell lines and on a breast tumour patient in vivo. Results suggest that different breast cancer types could be distinguished with FEXI based on their AXR values. The AXR could be determined for the tumour ROI, while in normal tissue the AXR was outside the experimental range. Low SNR did not allow voxel based data analysis. Further optimization of FEXI will be required before FEXI can be evaluated in a larger group of breast cancer patients.

Valentin Prevost¹, Olivier M. Girard¹, Gopal Varma², David C. Alsop², and Guillaume Duhamel^1
1CRMBM UMR 7339, CNRS/Aix-Marseille Université, Marseille, France, ²Radiology Departement, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, MA, United States

A previously reported new MT approach able to specifically image the inhomogenous component of the MT spectrum, and referred as inhomogeneous MT (ihMT), appeared to be sensitive to tissue with myelin. The present study proposed a qualitative investigation of ihMT specificity toward myelin by comparison with DTI and myelin stained images.

Hyunseok Seo¹, Yeji Han¹, and HyunWook Park^1
1Electrical Engineering, KAIST, Daejeon, Daejeon, Korea

In this abstract, a novel isotropic perfusion-weighting approach is proposed, where bi-polar gradients are used instead of the tagging pulses to obtain a high-resolution PW image with radial trajectory. By using a pair of spin-echo (SE) images acquired with single and double bi-polar gradients from multi-directions, anisotropic characteristics of the cerebral perfusion are also considered. Computer simulations were performed to evaluate the relation between the proposed method and the cerebral perfusion. MR experiment results from in-vivo brain imaging show that the proposed method produces a high-resolution PW image.

Liliana Caldeira¹, Seong Dae Yun¹, Nuno A da Silva¹, Christian Filss¹, and N Jon Shah^1,2
1Institute of Neuroscience and Medicine - 4, Research Centre Jülich GmbH, Jülich, Germany, ²Department of Neurology, RWTH Aachen University, Aachen, Germany

The arterial input function (AIF) is essential for quantification in MRI and PET imaging. The ground truth for AIF estimation is arterial cannulation. Alternatively, the AIF can be estimated using MRI and/or PET images, but a reasonable temporal resolution of dynamic image series is necessary (<2s). In PET imaging, high temporal resolution is limited (>5s). Here, we propose a method to acquire data for the AIF based on an EPI with keyhole (EPIK) sequence. The EPIK sequence combines both high temporal resolution and high spatial resolution. Furthermore, dual-contrast EPIK (DC-EPIK) can also be acquired to provide additional information.

Takashi Nishihara¹, Hiroyuki Itagaki¹, Kuniaki Harada¹, Masatomo Yokose¹, Oka Kuniharu¹, and Tetsuhiko Takahashi^1
1MRI System Division, Hitachi Medical Corporation, Kashiwa, Chiba, Japan

In order to visualize the blood flow in portal vein, we investigate use of a 2D beam excitation pre-saturationpulse (hereafter Beam Sat pulse) with the flow phantom and healthy volunteer.�@As a result, the Beam Sat pulse is able to saturate the portal vein selectively. When combined with unenhanced portal venography, the Beam Sat pulse seems to visualize additional information about the blood flow in portal vein.

Rasim Boyacioglu¹, Jenni Schulz¹, Peter Koopmans², Markus Barth^1,3, and David Norris^1,3
1Radboud University, Donders Institute, Nijmegen, Netherlands, ²FMRIB Centre, University of Oxford, Oxford, United Kingdom, ³Erwin L. Hahn Institute for Magnetic Resonance Imaging, University Duisburg-Essen, Essen, Germany

Recently we have implemented a multiband (MB) multi-echo (ME) sequence to investigate the potential improvement in sensitivity at 7T for resting state (RS) fMRI. In this study we investigated various approaches for cleaning ME and MB ME RS fMRI data to fully exploit the rich temporal information of MB ME data. With two different analysis strategies we have showed that MB acquisition improves functional connectivity compared a standard ME sequence after the removal of non-BOLD related artifactual signals.

Heiko Schmiedeskamp¹, Eric Peterson¹, Julian Maclaren¹, Rafael O'Halloran¹, Thomas Christen¹, Samantha J Holdsworth¹, Eric Aboussouan¹, William A Grissom², and Roland Bammer^1
1Department of Radiology, Stanford University, Stanford, CA, United States, ²Biomedical Engineering, Vanderbilt University, Nashville, TN, United States

The combined acquisition of gradient-echo and spin-echo fMRI signals facilitates the differentiation between simultaneously acquired BOLD signal changes with distinct contrast mechanisms at the expense of increased repetition time or decreased slice coverage. To facilitate combined gradient-echo and spin-echo fMRI while maintaining whole-brain coverage without prolonging TR, we propose utilizing multiband RF excitation in simultaneous multi-echo spin- and gradient-echo (SAGE) EPI acquisitions for fMRI, and we present preliminary results using a breath-hold task as proof-of-principle for stimulus-based multiband SAGE fMRI experiments.

Tom Hilbert^1,2, Tobias Kober^1,3, Tilman J. Sumpf⁴, Zhengguo Tan⁴, Jens Frahm⁴, Pavel Falkovskiy^1,3, Heiko Meyer⁵, Rolf Bendl^6,7, Jean-Philippe Thiran², Reto Meuli⁸, and Gunnar Krueger^1,8
1Advanced Clinical Imaging Technology, Siemens Healthcare IM BM PI, Lausanne, Switzerland, ²Signal Processing Laboratory (LTS5) École Polytechnique Fédérale de Lausanne, Lausanne, Switzerland, ³CIBM - AIT, École Polytechnique Fédérale de Lausanne, Lausanne, Switzerland, ⁴Biomedizinische NMR Forschungs GmbH am Max-Planck-Institut für biophysikalische Chemie, Göttingen, Germany, ⁵Siemens Healthcare, Erlangen, Germany, ⁶Division of Medical Physics in Radiation Oncology, DKFZ Heidelberg, Heidelberg, Germany, ⁷Department of Medical Informatics, Heilbronn University, Germany, ⁸Centre Hospitalier Universitaire Vaudoise and Univ. of Lausanne, Lausanne, Switzerland

In this work we investigate the combination of Model-based Accelerated RelaxomeTry by Iterative Nonlinear Inversion (MARTINI) with Generalized Autocalibrating Partially Parallel Acquisition (GRAPPA) to further accelerate and improve the reconstruction quality of T2 maps. GRAPPA is used to interpolate missing k-space lines of two-fold subsampled blocks of the MARTINI scheme prior to the MARTINI reconstruction. Images from an analytical phantom and in-vivo datasets are investigated. Resulting T2 maps of nominal 10-fold accelerated whole brain exams (1:40 minutes scans) are qualitatively and quantitatively compared to the reconstruction of the fully sampled and conventional 5-fold accelerated MARTINI datasets.

Jenni Schulz¹, Rasim Boyacioglu¹, and David G Norris^1,2
1Donders Institute for Brain, Cognition and Behaviour, Radboud University Nijmegen, Nijmegen, Netherlands, ²Erwin L. Hahn Institute for Magnetic Resonance Imaging, University Duisburg-Essen, Essen, Germany

3D-TOF-MRA is commonly used for imaging intracranial vessels. This NCE-MRA technique is based on inflow enhancement suppressing stationary tissue. Unfortunately, it suffers from time-inefficiency. This problem can be overcome by exciting multiple slabs simultaneously which can be reconstructed with a fast low resolution 3D-FLASH reference scan. Acquisition time will be decreased by a factor almost equal to the number of simultaneously excited slabs. The obtained results are comparable to the reference and CNR is maintained which leads to an increase in CNR efficiency. Furthermore, by reducing the thickness of the slabs in the multislab multiband acquisition, in-flow contrast is improved.

Benedikt A Poser^1,2, Dimo Ivanov¹, Stephan A Kannengiesser³, Kamil Uludag¹, and Markus Barth^2,4
1Faculty of Psychology and Neuroscience, Maastricht University, 6200MD Maastricht, Netherlands, ²Donders Institute for Brain, Cognition and Behaviour, Radboud University Nijmegen, Nijmegen, Netherlands, ³MR Applications Development, Siemens AG, Healthcare Sector, Erlangen, Germany, ⁴Erwin L Hahn Institute for Magnetic Resonance Imaging, University Duisburg-Essen, Essen, Germany

2D CAIPIRINHA greatly improves volumetric parallel imaging reconstructions by making optimal use of the available coil sensitivities and distributing the image alias over the reduced FOV. We demonstrate blipped-CAIPIRINHA in 3D EPI, where CAIPIRINHA’s ability to freely distribute the undersampling capability between the two phase-encoding directions creates the flexibility to: (a) use maximum through-plane acceleration to achieve very short TR, (b) use maximum in-plane acceleration to shorten EPI echo train for high-spatial resolution and minimal geometric distortion, or (c) any trade-off between the two. Each case is demonstrated with a total undersampling factor 16.

Karl Landheer¹ and Simon Graham^1
1Medical Biophysics, University of Toronto, Toronto, Ontario, Canada

A novel technique to rapidly measure fMRI parameters has been improved upon. Our technique uses a modified PRESS sequence to excite three arbitrarily positioned voxels. SENSE is then used to reconstruct each of the four voxel’s time-dependent signal. A minimum TR of 73 ms was achieved, and showed to provide good agreeance between a single voxel measurement. This technique provides access to T2, T2* and centre frequency, all which can be used to monitor functional activity through the BOLD response. This technique will be used to measure very fast brain activity for mental chronometry in the future.

Xiaopeng Zong¹, Juyoung Lee², Alexander Poplawsky³, Seong-Gi Kim^3,4, and Jong Chul Ye^2
1Biomedical Research Imaging Center, University of North Carolina at Chapel Hill, Chapel Hill, NC, United States, ²Korea Advanced Institute of Science & Technology, Daejon, Korea, ³University of Pittsburgh, PA, United States, ⁴Dept. of Biological Sciences, SKKU, Suwon, Korea

Compressed sensing (CS) may be useful for accelerating data acquisitions in high-resolution fMRI. Most of the existing CS-fMRI studies have been conducted with synthesized experiments where fully sampled k-space data were retrospectively down-sampled. However, it is difficult to determine pulse sequence-dependent artifacts as well as potential advantages of improved temporal resolutions using retrospective analyses. Here, we systematically investigated the properties of CS-fMRI using computer simulations and in vivo experiments of rat forepaw and odor stimulations with 2-dimensional gradient-recalled echo (GRE) and echo planar imaging sequences. Our results show that CS improves the statistical performance of fMR with negligible image artifacts.

Marc Rea¹, Xavier Boullier², Ian Young³, and Donald McRobbie^1
1Radiological Sciences Unit, Imperial College Healthcare NHS Trust, London, United Kingdom, ²Bioengineering, Imperial College, London, London, United Kingdom, ³Electrical Engineering, Imperial College London, London, United Kingdom

This work investigates the combination of Single Point Imaging (SPI) with Compressed Sensing reconstruction for the reduction of artefacts near passive metal implants, enabling the benefits of SPI to be utilised with much reduced acquisition times.

Cristoffer Cordes¹ and Matthias Günther^1,2
1Fraunhofer MEVIS, Bremen, Bremen, Germany, ²MR-Physics and Spectroscopy, Faculty 1, Universität Bremen, Bremen, Germany

To further investigate the potential of the Extended Phase Graph algorithm, the reconstruction chain has been applied to the signal components created by Extended Phase Graph simulations. The resulting images can then be used as convolution kernels to simulate the sequence response to the parameter map of an object or to extract information about parameter sensitivity and influence from a sequence. An algorithm has been developed and utilized to generate images from object parameter maps to compare simulation results to measured images and to extract contrast information from a pulse sequence.

Matthew Firth¹, Marco Mingarelli¹, Hugh Seton¹, and Dana Dawson^1
1University of Aberdeen, Aberdeen, United Kingdom

We describe a gradient echo based imaging technique for fast murine cardiac gated cine imaging. Using relatively low strength gradients at 4.7T we have obtained a multi-slice, multi-frame cine with full ventricular coverage in just over 5 mins. These images were analysed to assess ventricular function and the results compared with images using a slower technique which acquired a single line of k-space per R–wave. There was no significant difference between the calculated left ventricular volumes and the “Fast” technique led to a reduction in flow artefacts.

Noel D. Montgomery¹, Goldie R.E. Boone², and Jack L. Lancaster^2
1Human Effectiveness Directorate, Bioeffects Division, Radio Frequency Radiation Bioeffects Branch, Air Force Research Laboratory, 711 Human Performance Wing, JBSA Fort Sam Houston, Texas, United States, ²Research Imaging Institute, UT Health Science Center San Antonio, San Antonio, TX, United States

Spin Echo and Echo Planar imaging of resected nervous tissue presents a specific challenge due to susceptibility differences between tissue and air, and due to high signal from encapsulating media. The researchers developed an encapsulation method for brain tissue that minimized susceptibility artifacts in echo planar images and preserved contrast in spin echo images.

M. Ethan MacDonald^1,2, M. Louis Lauzon^2,3, and Richard Frayne^2,3
1Biomedical Engineering, University of Calgary, Caglary, AB, Canada, ²Seaman Family Magnetic Resonance Research Centre, Hotchkiss Brain Institute, Foothills Medical Centre, Calgary, AB, Canada, ³Radiology and Clinical Neurosciences, University of Calgary, Calgary, AB, Canada

In this work combined methodologies for rapidly acquiring parametric maps of the brain are described. Several parameters in the brain, including: T1, T2, T2*, magnetic susceptibility and proton density are calculated, in addition to several machine distortion parameters, including: B0 and B1 field inhomogeneity, and coil profiles. The protocol time used to produce these images was less than 26 minutes, and resulted in whole brain coverage with 1 mm^3 isotropic resolution. Collection of these key physiological and machine distortion parameters will allow for advanced simulation of the MR system.

Alfonso Mastropietro^1,2, Annette Tennstadt², Nadine Henn², Andreas Beyrau², Maria Grazia Bruzzone³, Giuseppe Baselli⁴, and Mathias Hoehn^2,5
1Scientific Direction Unit, Fondazione IRCCS Istituto Neurologico C. Besta, Milan, Italy, ²In-vivo NMR lab, Max Planck Institute for Neurological Research, Cologne, Germany,³Neuroradiology Unit, Fondazione IRCCS Istituto Neurologico C. Besta, Milan, Italy, ⁴Department of Electronics, Information and Bioengineering, Politecnico di Milano, Milan, Italy, ⁵Department of Radiology, Leiden Medical School, Leiden, Netherlands

Aim of this work is to perform a punctual optimization of RARE parameters for 19F MRI studies of labeled neuronal stem cells. This approach is based on relaxation times estimation and numerical simulations. The optimization approach was evaluated in different biological environments at 11.7T MRI scanner.A correct optimization strategy increases the sensitivity of 19F MRI technique.

Gage Redler¹, Boris Epel¹, and Howard J Halpern^1
1Radiation and Cellular Oncology, University of Chicago, Chicago, IL, United States

Uniform projection acquisition in electron paramagnetic resonance oxygen imaging reduces artifacts from changes during imaging, enables useful real-time reconstruction, and allows post-factum temporal resolution adjustment. A novel uniform acquisition method is presented as an alternative to the commonly used golden ratio (GR) method. The GR method allows acquisition of arbitrary numbers of projections, maintaining approximate uniformity throughout and for subsets of arbitrary size. In some cases, arrival at a pre-defined final projection set is necessary, which is impossible using the GR method. The maximally spaced projection sequencing method presented enables acquisition of arbitrary projection sets, maintaining approximate uniformity throughout acquisition.

Huihui Xu¹, Vahid Khalilzad-Shargi¹, Karin wartella¹, and Shadi F Othman^1
1University of Nebraska - Lincoln, Lincoln, Nebraska, United Kingdom

There is a crucial need to assess ex vivo constructs. MRI helps fulfill this role, but specimens allocated to a test tube for imaging cannot be returned to incubators, therefore, are wasted due to transfer incubation in a less than optimal growth environment. We present a standalone, miniature MRI-compatible incubator, termed the e-incubator, which uses a microcontroller to automatically sense and regulate physiological conditions for tissues and allow concurrent tissue culture and evaluation. The e-incubator offers an innovative scheme to study multiple applications, including underlying mechanisms related to the structural and functional changes of tissues due to growth and maturation.

Katharina Göbel¹, Jochen Leupold¹, Bibek Dhital^1,2, Pierre LeVan¹, Marco Reisert¹, Johannes Gerlach³, Robert Kamberger⁴, Carola Haas³, Jürgen Hennig¹, Dominik von Elverfeldt¹, and Jan G. Korvink^4
1Medical Physics, Dept. of Radiology, University Medical Center Freiburg, Freiburg, Germany, ²German Cancer Consortium (DKTK), Heidelberg, Germany, ³Dept. of Neurosurgery, Experimental Epilepsy Research, University Medical Center Freiburg, Freiburg, Germany, ⁴Dept. of Microsystems Engineering (IMTEK), Technical Faculty, University of Freiburg, Freiburg, Germany

Organotypic hippocampal slice cultures are a well established neuronal culture system that combines the advantages of cell culturing with a neuronal network tightly reflecting the in vivo state. They are frequently used to study morphological, molecular and electrophysiological changes associated with epilepsy. Our aim is to investigate these changes during epileptogenesis, particularly using high spatial resolution MR microscopy and DTI which allows continuous monitoring near the cellular level. High resolution MR images of fixed mouse hippocampi were obtained ex vivo and directly compared to histology. First trials in 2D DTI give notion of the structural composition of the hippocampus.

Jinil Park¹, Chanhee Lee¹, Soon Ho Yoon², Jin Mo Goo², and Jang-Yeon Park^1
1School of Biomedical Engineering, Konkuk University, Chung-ju, Chungcheongbuk-do, Korea, ²Department of Radiology, Seoul National University College of Medicine, Seoul, Korea

A new and simple strategy of improving the image quality of respiratory-gated PA data was proposed here when using a 3D spiral trajectory to fill in k-space. Based on the gradient-view generation algorithm we upgraded. show that increasing the number of interleaves can practically be a good and easy way to avoid streak artifacts as well as blurring due to the non-uniformity of k-space trajectory when filling in 3D k-space with a spiral trajectory. Considering the wide use of a 3D k-space spiral trajectory with interleaves, this method would be useful in practice.

R Reeve Ingle¹, Michael V McConnell^1,2, Juan M Santos³, William R Overall³, Bob S Hu^1,4, and Dwight G Nishimura^1
1Electrical Engineering, Stanford University, Stanford, California, United States, ²Cardiovascular Medicine, Stanford University, Stanford, California, United States,³HeartVista, Inc., Menlo Park, California, United States, ⁴Palo Alto Medical Foundation, Palo Alto, California, United States

A free-breathing, fat-suppressed cardiac cine pulse sequence is proposed, allowing multiple respiratory and cardiac phases to be simultaneously resolved. Fat suppression is achieved using an alternating repetition time balanced steady-state free precession sequence. By redesigning the slice-select rephasing gradients, a 1D navigator readout is acquired during the unused short TR interval and used to track respiratory motion. This technique allows data from a single free-breathing scan can to be displayed in a standard “cardiac-cine” format for a fixed respiratory phase, or in a “respiratory-cine” format for a fixed cardiac phase.

Manoj Kumar Sarma¹, Rajakumar Nagarajan¹, Paul Michael Macey², Ravi Aysola³, and M. Albert Thomas^1
1Radiological Sciences, UCLA School of Medicine, Los Angeles, CA, United States, ²School of Nursing, UCLA School of Medicine, Los angeles, CA, United States, ³Division of Pulmonary and Critical Care Medicine, UCLA School of Medicine, Los Angeles, CA, United States

Obstructive sleep apnea syndrome (OSAS) is a common sleep disturbance affecting the adult population leading to numerous health problems. Chronic intermittent hypoxic episodes, hypercapnia and transient blood pressure elevation in OSAS may damage neural structures and induce cerebral metabolic changes. Even though many structural imaging studies have shown brain tissue changes in OSAS, only a limited number of MRS based studies can be found so far. In this study, we investigated neurochemical changes in multiple brain regions of OSAS patients using compressed sensing (CS) based Echo-planar J-resolved spectroscopic imaging (EP-JRESI) and quantify the metabolites using prior knowledge fitting (ProFit) algorithm.

Brian L Burns¹, Neil Wilson², and M Albert Thomas^1,2
1Department of Biomedical Engineering, UCLA, Los Angeles, CA, United States, ²Department of Biomedical Physics, UCLA, Los Angeles, CA, United States

Scan times for the 4D EP-COSI pulse sequence can be 20-40 minutes depending on scan parameters. To reduce these scans times to clinically acceptable levels non-uniform under-sampling of the phase encoded dimensions coupled with non-linear reconstruction can be used to accelerate acquisition. This work compares the reconstruction results of 4X and 8X phantom data from Compressed Sensing, Total Variation, and Maximum Entropy with Group Sparse reconstruction, a variant of Compressed sensing that used a mixed 1-1,2 regularizer. Group Sparse reconstruction with overlapping groupings is shown to provide qualitatively and quantitatively superior results to other methods tested.

Subechhya Pradhan¹, Susanne Bonekamp¹, Joseph Gillen¹, Laura Rowland², S. Andrea Wijtenburg², Richard A.E. Edden¹, and Peter B Barker^1
1Russell H. Morgan Department of Radiology and Radiological Science, Johns Hopkins University School of Medicine, Baltimore, Maryland, United States, ²Psychiatry, University of Maryland School of Medicine, Baltimore, Maryland, United States

MRS is expected to benefit from moving to high field strengths such as 7T. However, there are also technical challenges associated with high fields, such as increased B0 and B1 field inhomogeneities. Therefore, it is important to compare 7T measurements to those at lower field. In this study, SNR, linewidths and Cramer-Rao lower bounds are compared between 3T and 7T MRS in 3 brain regions of 4 healthy subjects. Measurements made with near identical methodology at both field strengths, including 32-channel head coils and semi-LASER localization. Improvements at 7T were found in all metrics examined, and were consistent with expectations.

Sebastian C. Niesporek¹, Stefan H. Hoffmann¹, Moritz C. Berger¹, and Armin M. Nagel^1
1Medical Physics in Radiology, German Cancer Research Center (DKFZ), Heidelberg, Germany

The tissue sodium (²³Na) concentration is of great interest in biomedical research. However the ²³Na nucleus has a low MR sensitivity and fast relaxation times which requires larger voxel sizes in ²³Na MRI. Thus, partial volume effects reduce the accuracy of concentration measurements. In this work, a partial volume correction method was transferred to ²³Na MRI, optimized and tested in phantom measurements and simulations as wells as in ²³Na MRI of the human brain. The study showed good correction capability for phantom and in-vivo data.

Nicolas G.R. Behl¹, Christine Gnahm¹, Peter Bachert¹, and Armin M. Nagel^1
1Medical Physics in Radiology, German Cancer Research Center (DKFZ), Heidelberg, Germany

3D-dictionary-learning-CS is applied for the reconstruction of radial ²³Na-MRI-data. The dictionary used for the sparsifying transform consists of 3D-blocks learnt on the gridding-reconstruction of the data. A K-SVD algorithm is used to learn the dictionary and the corresponding representation, the self-consistency of the actual image and the raw-data is enforced through a conjugate gradient algorithm. The performance of the reconstruction algorithm is verified with simulated data (2mm isotropic), phantom ²³Na-data (1.5mm isotropic) and in-vivo ²³Na-data (2mm isotropic), showing significant noise reduction compared to the corresponding gridding reconstructions, as well as increased SSIM and reduced RMSE.

Christine Gnahm¹, Nicolas G.R. Behl¹, Armin Biller², Peter Bachert¹, and Armin M. Nagel^1
1Medical Physics in Radiology, German Cancer Research Center (DKFZ), Heidelberg, Germany, ²Department of Neuroradiology, University Hospital Heidelberg, Heidelberg, Germany

²³Na MRI is still hampered by low signal-to-noise ratio (SNR) and long acquisition times. We present an iterative reconstruction method using a priori information from high-SNR high-resolution ¹H MRI through anatomically weighted 2^nd order Total Variation regularization (AnaWeTV). By anatomical weighting, intensity variations in the ²³Na image are promoted at positions with high confidence of tissue boundaries. In simulated brain images, it is shown that the total sodium concentration in small lesions that are known a priori can be determined more precisely. Furthermore, we find a 2.2 fold SNR increase over gridding in ²³Na MRI of a MS patient.

Fabio Ginnari Satriani¹, Emiliano Surrentino¹, Alessandro Ricci¹, and Rossella Canese^1
1Cell Biology and Neurosciences Dept, Istituto Superiore di Sanità, Rome, Italy

31P MRS offers a powerful approach to non-invasively measure extracellular pH (pHe) and intracellular/extracellular pH gradient in vivo. 31P MRS techniques require long acquisition times.The exogenous cell-impermeant 31P reporter 3-aminopropyl phosphonate (3-APP) used for pHe evaluation should be retained within the tumor for the entire duration of the measurements. This condition is not always fulfilled in highly vascularised tumors. We here propose a new technique for fast 31P MRS in highly vascularised tumours by improving the localization of the surface coil (positioned on superficial tumours) with a saturation band which dephases the signal arising from the mouse body.

Denis Grenier¹, Anne-Laure Perrier¹, Hervé Saint-Jalmes², and Olivier Beuf^1
1CREATIS, CNRS UMR 5220, INSERM U1044, INSA-Lyon, Université de Lyon, Villeurbanne, France, ²PRISM, LTSI, INSERM U1099, Université Rennes 1, Rennes, France

In this work we investigate the feasibility of the acquisition of an NMR signal during a strong RF irradiation. By acquiring the signal during the RF excitation, dipolar interaction can be canceled, leading to a solid state spectroscopy technique suited for in vivo physiologically sound applications.

Kelvin J. Layton¹, Bahman Tahayori¹, James Korte¹, Iven M. Y. Mareels¹, Peter M. Farrell¹, and Leigh A. Johnston^1
1Department of Electrical and Electronic Engineering, The University of Melbourne, Melbourne, Victoria, Australia

The response of the magnetization to amplitude-modulated continuous wave RF excitation is investigated with theory and experiments. Proof-of-concept measurements of the steady-state magnetization waveform demonstrate substantial frequency components at harmonics of the modulation frequency, emphasizing the nonlinear nature of the spin system. Experiments are in excellent agreement with theoretical expressions derived from the Bloch equations. Furthermore, experiments show that the steady-state magnetization is relatively large when the amplitude of the field matches the modulation frequency, establishing a secondary resonance condition not previously observed in magnetic resonance systems.

Rajakumar Nagarajan¹, Neil Wilson¹, and M.Albert Thomas^1
1Radiological Sciences, University of California Los Angeles, LOS ANGELES, CA, United States

Conventional three-dimensional MRSI is time-consuming because it involves a large number of phase encodings. Echo-planar spectroscopic imaging (EPSI) approaches have been used to reduce the long acquisition time required for multiple spatial encoding steps, In our study, we have implemented non-uniform undersampling (NUS) with compressed sensing (CS) reconstruction to a three dimensional (3D) EPSI in order to accelerate the data acquisition. We propose that by 2X nonuniform undersampling along phase encodings, the 3D EPSI acquisition time could be significantly reduced without sacrificing the spectral quality with more metabolite detection in the prostate phantom.

Hao Chen¹, Zhiyong Zhang¹, Shuhui Cai¹, and Zhong Chen^1
1Department of Electronic Science, Xiamen University, Xiamen, Fujian, China

The inhomogeneity of magnetic fields always affects the quality of NMR spectra and hampers the spectral assignment. To circumvent this influence, a new NMR acquisition scheme based on spatial encoding technique and intermolecular zero-quantum coherence is proposed to achieve a high-resolution 3D NMR spectrum in a few minutes without the peaks broadening caused by field inhomogeneity. Both homonuclear correlation and J-resolved information can be obtained from this spectrum. The new scheme may be useful for the study of biological tissues.

David S Smith^1,2, Saikat Sengupta^1,2, and E. Brian Welch^1,2
1Radiology and Radiological Sciences, Vanderbilt University, Nashville, TN, United States, ²Institute of Imaging Science, Vanderbilt University, Nashville, TN, United States

We present results from a rapid, whole-body, free-breathing MR imaging protocol that uses a continuous moving bed coupled with a golden angle radial acquisition and a compressed sensing reconstruction to achieve high image quality and artifact reduction despite an equivalent undersampling factor of 7.0. Additionally, the use of a golden angle radial acquisition allows slice positions and thicknesses to be determined after data collection, allowing the reconstructed images to be tailored to the clinical application so as to minimize through-slice signal averaging effects in regions of interest or control the balance of SNR and spatial resolution.

Ganesh Adluru¹, Liyong Chen², and Edward V.R. DiBella^1
1Radiology, University of Utah, Salt lake city, Utah, United States, ²Advanced MRI Technologies, Sebastopol, CA, United States

Myocardial perfusion imaging is an invaluable tool to diagnose and study coronary artery disease. Complete coverage of the heart without sacrificing temporal or spatial resolution is desired. Many approaches aim to achieve the desired goal by undersampling k-space data for each time frame and using sophisticated reconstruction algorithms like compressed sensing. Another complementary approach to increase coverage without compromising spatio-temporal resolution is simultaneous multi-slice imaging in which multiple slices are simultaneously excited with phase modulation and acquired simultaneously. Here we combine CAIPI, a simultaneous multi-slice method, with compressed sensing for radial myocardial perfusion and present promising results.

Sagar Mandava¹, Zhitao Li¹, and Ali Bilgin^1,2
1Electrical and Computer Engineering, University of Arizona, Tucson, AZ, United States, ²Biomedical Engineering, University of Arizona, Tucson, AZ, United States

A novel view ordering scheme for radial multi-band RF encoded acquisitions is presented. This view ordering enables sliding window reconstructions for arbitrary window positions and lengths. This flexible view ordering scheme combined with sparsity regularized reconstructions paves the way for rapid time resolved simultaneous multi-slice MRI.

Martin Krämer¹, Karl-Heinz Herrmann¹, Judith Biermann¹, and Jürgen R Reichenbach^1
1Medical Physics Group, Institute of Diagnostic and Interventional Radiology I, Jena University Hospital - Friedrich Schiller University Jena, Jena, TH, Germany

To perform cardiac cine imaging a single slice is repeatedly acquired with radial readouts rotated by an angle based on the golden ratio. Performing correlation analysis between interspersed 1D navigator projections, time points corresponding to the same cardiac motion phases were automatically identified and used for retrospective combination of radial readouts from multiple data windows. Analysis of the 1D navigator data provided a detailed correlation function revealing cardiac motion over time. Imaging results were comparable to images reconstructed based on a temporally synchronized ECG showing low artifact level and good image quality in terms of CNR.

Joshua D. Trzasko¹, Armando Manduca¹, Yunhong Shu¹, John Huston III¹, and Matt A Bernstein^1
1Mayo Clinic, Rochester, MN, United States

Sparse reconstruction of non-Cartesian MRI data remains computationally challenging since multiple “gridding” operations must be executed at each iteration of the reconstruction. Recently, an efficient alternating-direction-method-of-multiplier (ADMM) strategy was proposed for sparse MRI reconstruction. For non-Cartesian MRI, the data fidelity sub-problem must also be solved iteratively. If off-resonance effects are accounted for, standard circulant preconditioners cannot be used to accelerate this task. In this work, we show that an algebraic reformulation of the ADMM scheme enables the use of simple but effective diagonal PCs for non-Toeplitz models, and demonstrate their practical benefit for undersampled SWIRLS 3D CE-MRA.

Philip J Beatty^1,2
1Physical Sciences, Sunnybrook Research Institute, Toronto, Ontario, Canada, ²Medical Biophysics, University of Toronto, Toronto, Ontario, Canada

An improved method of k-space channel combination for non-Cartesian acquisitions is described. With the proposed method, it is possible to simultaneously grid and perform channel combination in k-space using small width (4x4) convolution kernels for 2D non-Cartesian imaging. Enabling good image quality with small convolution kernels could lead to very fast non-Cartesian multi-channel image reconstruction.

Weiran Deng¹ and V. A. Stenger^1
1University of Hawaii, John A. Burns School of Medicine, Honolulu, HI, United States

This abstract presents a method that uses the vectorized parallelism in the modern CPUs to accelerate the reconstruction of images acquired using non-Cartesian trajectories such as spiral. Modern CPUs have SSE (Stream SIMD Extension) and AVX (Advanced Vectorization Extension) features, which are designed to improve the performance of the applications that have inherent parallel structures. We show that data acquired using spiral can be re-arranged and reconstructed using this approach. The reconstruction speed is four to five times faster than the conventional approach that uses multiple threads.

David S Smith^1,2 and E Brian Welch^1,2
1Radiology and Radiological Sciences, Vanderbilt University, Nashville, TN, United States, ²Institute of Imaging Science, Vanderbilt University, Nashville, TN, United States

We present a model for k-space trajectory shifts due to independent timing delays in gradient channels. We use this model to successfully detect and correct for gradient delays in 3D golden angle radial MRI data without the need for additional data collection, calibration, or hardware measurements. The method can also be used for 2D radial or other non-Cartesian trajectories with only slight modifications.

Hasan Ertan Cetingul¹, Peter Speier², Michaela Schmidt², Qiu Wang¹, and Mariappan S. Nadar^1
1Imaging and Computer Vision, Siemens Corporation, Corporate Technology, Princeton, NJ, United States, ²Siemens AG, Healthcare Sector, Erlangen, Germany

2D unsegmented real-time bSSFP cine MRI is used for non-invasive assessment of the cardiac function without breath-hold. bSSFP requires short inter-pulse distance repetition time (TR) for robustness against B0 field inhomogeneities and flow. In Cartesian bSSFP the TR is reduced by partial Fourier in the readout direction, i.e., the early part of the echo is omitted. We develop CS reconstructed radial bSSFP with asymmetric views for free-breathing cardiac cine MRI. This technique can speed up acquisition and stabilize bSSFP signals by reducing TR by about 12% with only minor degradation in image quality compared to the conventional full echo schemes.

Alan Chu¹ and Douglas C. Noll^1
1Biomedical Engineering, University of Michigan, Ann Arbor, MI, United States

A GRAPPA-based reconstruction method for non-Cartesian simultaneous multislice (SMS) imaging is presented. The method is non-iterative, decreasing image reconstruction time relative to iterative SENSE-based methods for non-Cartesian SMS. A concentric ring k-space trajectory is used, along with z-gradient blipping, to enhance the accuracy of GRAPPA weight calibration and subsequent slice separation. Preliminary results suggest good potential for use in fMRI or DTI experiments.

Thomas W Polley¹, YunHong Shu², Joshua D Trzasko¹, and Matt A Bernstein^2
1Physiology and Biomedical Engineering, Mayo Clinic, Rochester, MN, United States, ²Radiology, Mayo Clinic, Rochester, MN, United States

We demonstrate an automated design algorithm for a 3D non-Cartesian trajectory inspired by SWIRLS. In addition to being automated, the resulting trajectory maximizes the uniformity of sampling density through the k-space volume. Phantom and volunteer acquisitions demonstrate its feasibility and image quality improvement over the default SWIRLS trajectory for a fixed scan time. The algorithm takes as inputs the FOV, resolution, and hardware gradient limitations, and may be efficient enough to be done in real time. This would allow it to be more easily implemented in a variety of clinical and research situations.

Ya Li¹, Ran Yang¹, and Zhongping Zhang^2
1Sun Yat-sen University, Guangzhou, Guangdong, China, ²GE Healthcare, Guangzhou, Guangdong, China

In general, the CS-MRI process carries out two main operations: undersampling in the k-space, which meets the incoherence condition, and the reconstruction of the image, which meets the sparsity or compressibility condition. To meet the incoherence condition, the basic theory of compressed sensing requires acquisition of randomized set of measurements. For MRI scanner, however, random sampling would yield longer sampling trajectory because of the MR hardware constraints, and requires bigger changes in amplitudes and polarity of MR gradients those making it infeasible practically. Here we introduce a chaotic k-space trajectory for CS-MRI, which is a good candidate k-space trajectory of incoherent-like sampling scheme.

Li Zhao¹ and Craig H. Meyer^1,2
1Biomedical Engineering, University of Virginia, Charlottesville, VA, United States, ²Radiology, University of Virginia, Charlottesville, VA, United States

Compressed sensing has demonstrated notable acceleration in 3D and dynamic imaging, but it is limited by sample pattern design in 2D imaging. The flexibility of variable-density spiral trajectories gives them unique advantages for optimal sampling pattern design for 2D compressed sensing. This study demonstrates that these trajectories can be used to implement a theoretically optimal multilevel random sampling pattern. Simulations and experiments using this sampling pattern yield high acceleration rates with good fidelity with a single-channel 2D acquisition.

Guido Buonincontri¹, Carmen Methner², Thomas Krieg², T Adrian Carpenter¹, and Stephen J Sawiak^1
1Wolfson Brain Imaging Centre, Department of Clinical Neurosciences, University of Cambridge, Cambridge, United Kingdom, ²Department of Medicine, University of Cambridge, Cambridge, United Kingdom

Radial acquisitions can suffer from trajectory errors leading to reduced image quality. We present a new method of trajectory correction that uses all spokes of a radial acquisition, including B0 correction, and compare it to an existing method that uses a two-spoke pre-scan calibration, demonstrating this in the mouse heart. We also compared the quality of navigator signals obtained from the radial data with each technique. The typical shading artifacts seen in radial scans were significantly reduced with our method. Furthermore, modulation of navigator signals due to the acquisition angle was significantly reduced with the new technique.

Judith Biermann¹, Martin Krämer¹, and Jürgen R Reichenbach^1
1Medical Physics Group, Institute of Diagnostic and Interventional Radiology I, Jena University Hospital - Friedrich Schiller University Jena, Jena, Germany

When performing MR imaging with radial trajectories time delays of readout gradients cause severe artifacts in the reconstructed images. A correction algorithm was implemented using only the acquired data without additional calibration or template measurements. Radial data were shifted along their readout direction by applying a range of discrete readout shifts. By analyzing the reconstructed image for each shift, a corrected image was obtained fully automatically. The proposed method was tested in phantom and in vivo data and proved its reliability.

Nadine Gdaniec¹, Andrea J. Wiethoff^2,3, Peter Börnert⁴, Mariya Doneva⁴, Ivan Pedrosa^3,5, and Alfred Mertins^1
1Institute for Signal Processing, University of Luebeck, Luebeck, Luebeck, Germany, ²Philips Research North America, Briarcliff Manor, New York, United States, ³Advanced Imaging Research Center, UT Southwestern Medical Center, Dallas, Texas, United States, ⁴Philips Research Laboratories, Hamburg, Germany, ⁵UT Southwestern Medical Center, Dallas, Texas, United States

An abdominal dynamic contrast enhanced (DCE) examination typically consists of a series of images acquired with the same imaging sequence. The resulting images represent the same anatomy, but differ due to contrast agent arrival and wash-out. The contrast agent injection can make it more difficult for the patient to hold their breath properly, resulting in severe artifacts after contrast injection. Data are acquired with a modified adaptive sampling pattern, which implies higher undersampling for shorter breath-holds. The post-contrast images should, therefore, benefit from a joint reconstruction of pre- and post-contrast images, which is evaluated in this work.

Thomas Küstner^1,2, Sergios Gatidis¹, Christian Würslin¹, Nina Schwenzer¹, Bin Yang², and Holger Schmidt^3
1Department of Radiology, Universtity of Tübingen, Tübingen, Germany, ²Institute of Signal Processing and System Theory, University of Stuttgart, Stuttgart, Germany,³Department of Preclinical Imaging & Radiopharmacy, Universtity of Tübingen, Tübingen, Germany

For a clinical feasible Motion Correction setup in a PET/MR system, one should have accurate and sharp images which are acquired as fast as possible. Compressed Sensing promises high acquisition accelerations, whilst penalizing image quality with regard to sharpness. In order to sample the high frequencies denser, we propose a new subsampling scheme which reduces the sampled k-space region to a smaller subset. The k-space reduction has to be corrected for during the Compressed Sensing reconstruction process which uses a combined FOCUSS and POCS algorithm. The framework is called ESPReSSo (comprEssed Sensing PaRtial SubSampling).

Felipe Yanez^1,2 and Pablo Irarrazaval^1,2
1Department of Electrical Engineering, Pontificia Universidad Católica de Chile, Santiago, Chile, ²Biomedical Imaging Center, Pontificia Universidad Católica de Chile, Santiago, Chile

We propose a new reconstruction technique tailored for free-breathing dynamic 3D liver MRI by estimating the motion between frames to correct inconsistencies in k-space data. Our approach produced results that demonstrate it is feasible to achieve a 10x speedup in acquisition time and remove motion artifacts without diminishing image quality. The proposed method produced gains up to 3 dB with respect of traditional CS framework.

Qiu Wang¹, Michael Zenge², Hasan Ertan Cetingul¹, Edgar Mueller², and Mariappan S Nadar^1
1Imaging and Computer Vision, Siemens Corporation, Corporate Technology, Princeton, NJ, United States, ²MR Application & Workflow Development, Siemens AG, Healthcare Sector, Erlangen, Germany

Compressed sensing or sparsity based MR reconstruction takes advantage of the fact that the image is compressible in a specific transform domain, and enables reconstruction based on under-sampled k-space data thereby reducing the acquisition time. One requirement for the compressed sensing theory to work is the data acquisition in k-space to be incoherent. Although many random sampling schemes theoretically meet such requirements good enough, the MR physics or even the pathophysiology of a patient might impose additional constraints which have to be taken into account. This is considered the coherence barrier. In the current work, we formulate a sampling strategy that promises to achieve asymptotic incoherence, thus breaking the coherence barrier. Please notice that both the data acquisition and the reconstruction which have been used are investigational prototypes which experience continuous development. Nonetheless, experimental results in a phantom and a volunteer demonstrate a significant improvement of the spatial resolution with an increasing sub-sampling rate and a constant data acquisition time accordingly.

Frank Zijlstra¹, Jaco J.M. Zwanenburg¹, Max A. Viergever¹, and Peter R. Seevinck^1
1Image Sciences Institute, UMC Utrecht, Utrecht, Netherlands

We propose a novel, data-driven method for optimizing Cartesian undersampling patterns for Compressed Sensing. The method iteratively adds sampling points based on CS reconstructions of a training set. The performance of the proposed optimized sampling patterns are evaluated against the commonly used Variable Density undersampling methods. Our method shows improvements in both the Normalized Root Mean Square Error and the mean Structural Similarity index. The method generalizes to any reconstruction method that allows Cartesian undersampling in any number of dimensions and would enable optimization of patterns for a combination of CS and parallel imaging.

Angshul Majumdar¹ and Rabab Ward^2
1Indraprastha Institute of Information Technology, New Delhi, Delhi, India, ²Electrical and Computer Engineering, University of British Columbia, Vancouver, BC, Canada

This work proposes to employ elastic-net for MRI reconstruction. The elastic net favours grouping effect among the stree-structured wavelet coefficients and thereby yields better reconstruction than straighforward sparse reconstruction. We propose generalization of the standard elastic net formulation by introducing an analysis prior elastic net. The results from our generalization yields better results than state-of-the-art techniques in MRI reconstruction.

Zhaoyang Jin¹ and Qing-San 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, three interleaved datasets along PE direction were acquired to separate the overlapped ghosts by solving a least-square problem. This study reveals that overlapped ghosts can be separated by using only two interleaved k-space datasets since the ghost order index (n1, n2) can be readily calculated from central k-space data. Results from phantom and in vivo data demonstrated feasibility of the proposed method, leading to further acceleration of SPEED imaging.

Qing-San Xiang^1
1Radiology, University of British Columbia, Vancouver, British Columbia, Canada

Skipped Phase Encoding and Edge Deghosting (SPEED) can accelerate MRI with a single coil. Previously, SPEED sparsely sampled 3 interleaved datasets with skip size N and relative offsets along PE, and resolved aliasing ghosts by solving linear equations pixel-by-pixel. In this study, it is demonstrated that SPEED can be performed with only 1 dataset, similar to that used in GRAPPA except from only one receiver coil, achieving greater acceleration. The new algorithm imposes partial knowledge of the data in both reciprocal Fourier domains, and finds a global answer through straightforward iterations between Object and K-space (iOK).

Jieying Luo¹, Taehoon Shin², Tao Zhang¹, Joseph Y. Cheng¹, Bob S. Hu³, and Dwight G. Nishimura^1
1Electrical Engineering, Stanford University, Stanford, California, United States, ²University of Maryland, Baltimore, Maryland, United States, ³Palo Alto Medical Foundation, Palo Alto, California, United States

Perfusion imaging in the lower extremities remains challenging due to the requirements of large volumetric coverage and high temporal resolution. A low-rank matrix-completion reconstruction method has been proposed for highly accelerated dynamic contrast-enhanced perfusion imaging. In this work, an improved reconstruction method that combines low-rank matrix-completion reconstruction with image-based segmentation and parallel imaging is developed and tested in vivo. The proposed method can recover perfusion dynamics with less temporal blurring, and is promising for quantitative perfusion imaging in the lower extremities.

Yudong Zhu^1
1Zhu Consulting, Scarsdale, NY, United States

Compressed sensing offers a capacity for accelerating data acquisition while keeping aliasing and noise effects subdued. Theories and experiences however are yet to establish a more robust guidance on random sampling, sparse model and non-linear solver, to help manage the challenge of using the technology in diagnostic MR. In this work we took a new angle and investigated the feasibility of introducing self-validation into compressed sensing MR. The goal is to assist fidelity assessment and improvement in practice with validation tests that can be automatically performed on any specific imaging instance itself, without requiring additional data or comparison references.

Mohammad Kayvanrad^1,2, Amy Lin³, Rohit Joshi³, Jack Chiu³, and Terry Peters^1,2
1Biomedical Engineering, University of Western Ontario, London, Ontario, Canada, ²Robarts Research Institute, University of Western Ontario, London, Ontario, Canada,³Medical Imaging, University of Western Ontario, London, Ontario, Canada

While the quality of under-sampled MRI reconstructions is commonly evaluated based on quantitative metrics, such as the error with respect to the fully-sampled reference, such quantitative measures do not necessarily correlated with the subjective quality as perceived by radiologists and other expert end users. Therefore, unless accompanied by subjective measurements, quantitative quality metrics will be of limited clinical impact. This abstract presents the results of our experiments aimed towards subjective quality assessment of under-sampled MRI reconstructions for specific clinical applications. Particularly, it is shown that the advantages of different reconstructions greatly depend on the underlying clinical application.

Hao Li¹, Li Dong², Bo Li³, Bin Chen³, and JUE ZHANG^1,3
1Academy for Advanced Interdisciplinary Studies, Peking University, Beijing, Beijing, China, ²Radiology, Beijing Anzhen Hospital, Beijing, Beijing, China, ³College of Enigneering, Peking University, Beijing, Beijing, China

High-resolution 3D-MERGE technique based black-blood imaging has been reported to quantitatively measure carotid atherosclerotic plaque morphology and tissue composition, and receive more and more clincial concerns. However, it brings about relative long time consuming due to the use of motion-sensitized preparation, which can increase the probability of motion artifacts due to swallowing, respiration or neck movements. In our previous research, compressed sensing (CS) is used to improve temporal resolution by reconstructing images from a dramatically small number of data without introducing severe image artifacts[1]. The purposes of this study are: (1) to determine the reproducibility of the fast 3D-MERGE imaging using CS reconstruction. (2) to determine the consistency of this method when using different CS acceleration factors.

Amaresha Shridhar Konar¹, Divya Jain¹, Shamshia Tabassum¹, Rajagopalan Sundaresan², Ramesh Venkatesan², and Sairam Geethanath^1
1Medical Imaging Research Center, Dayananda Sagar Institutions, Bangalore, Karnataka, India, ²GE Healthcare, Bangalore, Karnataka, India

Compressed Sensing (CS) and Parallel Imaging (PI) are widely used to reduce MRI scan time and their combination yields better performance than used individually. The proposed method implements the combination of Region of Interest Compressed Sensing (ROICS) and SENSitivity Encoding (SENSE) which applies weighted CS to a particular ROI and the resulting output is then reconstructed using SENSE for arbitrary k-space. Proposed method performs better compared to PI and CS+PI for 6 channel brain data as validated qualitatively through images and quantitatively as determined by Peak Signal to Noise Ratio (PSNR) parameter by comparing with existing methods.

Leo K. Tam¹, Gigi Galiana¹, Haifeng Wang¹, Emre Kopanoglu¹, Andrew Dewdney², Dana C. Peters¹, and R. Todd Constable^1
1Diagnostic Radiology, Yale University, New Haven, CT, United States, ²Siemens Healthcare AG, Erlangen, Bavaria, Germany

Incoherence in compressed sensing is known to be important, but is there new understanding to be gained beyond the canonical method of selecting k-space coefficients in a psuedo-random manner. The incoherence parameter is studied, which dictates the largest subset of vectors in the sparse domain that may be exactly recovered via convex optimization with an L1 norm constraint. Incoherence parameter maps, showing the pairwise incoherence at each pixel are presented. The incoherence parameter is optimized using nonlinear encoding gradients, and experiments with a 3T Siemens Trio are presented that show an optimized incoherence parameter leads to reduced MSE.

Jose Caballero¹, Anthony N. Price^2,3, Daniel Rueckert¹, and Joseph V. Hajnal^2,3
1Department of Computing, Imperial College London, London, United Kingdom, ²Division of Imaging Sciences and Biomedical Engineering Department, King's College London, London, United Kingdom, ³Centre for the Developing Brain, King's College London, London, United Kingdom

Acceleration of Magnetic Resonance (MR) acquisitions through partially parallel imaging using array coils is limited by noise amplification. Compressed sensing regularization has de-noising properties that can mitigate this effect. Recent results on dictionary learning have shown that using overcomplete patch-based frames and adapting them to the object can have a notable impact on reconstruction by finding sparser representations and adjusting to the natural features of the object. However, these results have not yet been tested for parallel MR. Here we propose an algorithm to exploit overcomplete and adaptive frames for SPIRiT reconstruction and demonstrate its superiority to traditional wavelet regularization.

Alicia W Yang^1,2, Li Feng^1,2, Daniel K Sodickson¹, and Ricardo Otazo^1
1Bernard and Irene Schwartz Center for Biomedical Imaging, Department of Radiology, New York University School of Medicine, New York, NY, United States, ²Sackler Institute for Biomedical Sciences, New York University, New York, NY, United States

A patch-based sparse image reconstruction method based on local image correlations is proposed for compressed sensing. The method divides the image into non-overlapping blocks and sparsity is enforced in each block separately by thresholding the Principal Component Analysis (PCA) representation of a series of small patches within the block. The method exploits correlations directly in the image without the need of an analytical transform and it is reference-less and computationally efficient, removing the need to search for similar patches in the whole image. We tested the performance of the method to reconstruct undersampled 2D and 3D MSK images.

Ganesh Adluru¹, Yaniv Gur², Jeffrey Anderson¹, and Edward V.R. DiBella^1
1Radiology, University of Utah, Salt lake city, Utah, United States, ²SCI Institute, University of Utah, Salt lake city, Utah, United States

Standard diffusion imaging is limited when evaluating regions with crossing fibers. Higher order diffusion acquisitions with multiple b-values and multiple diffusion directions can reveal crossing fiber information; however it takes a long time to acquire such data. Acceleration techniques including compressed sensing can be applied to reduce the readout time by undersampling k-space and using constraints that exploit redundancies to remove the undersampling artifacts. Here we test a patch-based low rank constraint that exploits redundancies in terms of patches across diffusion directions for undersampled diffusion data. Promising results are presented in a multi-shell acquisition in a stroke patient.

Pavel Dvorak^1,2, Wolf Dieter Vogl³, and Vladimir Juras^4
1Institute of Scientific Instruments of the ASCR, v.v.i., Brno, Czech Republic, ²Department of Telecommunications, Brno University of Technology, Brno, Czech Republic,³Computational Image Analysis and Radiology Lab, Department of Radiology, Medical University of Vienna, Vienna, Austria, ⁴MR Centre of Excellence, Department of Radiology, Medical University of Vienna, Vienna, Austria

This work deals with hip joint cartilage segmentation, which is an important task in joint diseases diagnosis. Since the manual segmentation, commonly used nowadays, is a tedious and lengthy task, this work brings new idea into its automation and simplification of the medical expert work. Nevertheless, the proposed method still preserve the medical expert interaction to enable manual corrections.

Xiaoguang Lu¹, Peter Speier², Hasan Cetingul¹, Marie-Pierre Jolly¹, Michaela Schmidt², Mariappan Nadar¹, Frank Sauer¹, and Edgar Mueller^2
1Corporate Technology, Siemens Corporation, Princeton, New Jersey, United States, ²Siemens AG, Erlangen, Germany

We describe a general method for reconstructing a sequence of volumes representing 3D dynamics of a target anatomy, e.g., cardiac, tongue, vocal tract, or moving joint, based on densely acquired complementary 2D slices. Speech dynamics imaging is used as a case study for demonstration. Preliminary results have shown promise.

Aiqi Sun¹, Rui Li¹, Bida Zhang², Chun Yuan^1,3, and Feng Huang^4
1Department of Biomedical Engineering & Center for Biomedical Imaging Research, School of Medicine, Beijing, Beijing, China, ²Philips Research Asia Shanghai, Shanghai, China,³Department of radiology, University of Washington, Seattle, WA, United States, ⁴Philips Healthcare, Gainesville, FL, United States

The previous studies have shown phase-contrast cine MRI is a powerful tool for studying flow-related physiology and pathophysiology. However, the additional scans for encoding the flow velocity information increase the total scan time and limit the achievable spatial and temporal resolutions within a clinically acceptable duration. Among kinds of k-t reconstruction algorithms exploiting spatiotemporal correlation, k-t GRAPPA has the advantage of self-calibration, but degrades the temporal resolution when acceleration factor is high. Here we propose an artificial sparse scheme using static tissue elimination scheme to improve the temporal resolution preservation of k-t GRAPPA for blood flow measurements of thoracic aorta.

Feiyu Chen¹, Feng Huang², Dan Zhu¹, Haikun Qi¹, Kui Ying³, and Huijun Chen^1
1Center for Biomedical Imaging Research, Department of Biomedical Engineering, Tsinghua University, Beijing, China, ²Philips Healthcare, Gainesville, FL, United States,³Department of Engineering Physics, Tsinghua University, Beijing, China

In this research, we proposed a k-t GS-SPIRiT method to combine k-t group sparse and parallel imaging, which can result in images better than each individual method in cardiac cine image.

Chen Guang Peng¹, Dan Zhu², Fei Yu Chen², and Kui Ying^1
1Physics and Engineering, Tsinghua University, Beijing, Beijing, China, ²Biomedical Engineering,Tsinghua University, Beijing, Beijing, China

We proposed k-t ESPIRiT for dynamic imaging, and applied it to PRFS thermometry. Accurate sensitivity maps computed from k-t ESPIRiT enables a phase preservation in image reconstruction, which is the prerequisite for the widely used PRFS on thermometry. Experimentally, we compare the proposed method with k-t FOCUSS on phantom under high under-sampled factor. We demonstrated that phase map reconstructed from K-t ESPIRiT is more accurate than k-t FOCUSS, and that temperature estimation of k-t ESPIRiT is closer to the reference. Therefore, k-t ESPIRiT performs better than k-t FOCUSS on PRFS acceleration.

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

Fast magnetic resonance imaging (MRI) techniques [1-4], which lead to signal recovery from incomplete data, have been introduced in dynamic imaging to improve spatiotemporal resolution without apparent loss of image quality. In this respect, we propose a novel, highly accelerated dynamic parallel MRI reconstruction method exploiting a constrained state space model with low rank and sparsity while jointly estimating spatiotemporal kernels and missing signals in k-t space in an iterative fashion. Spatiotemporal kernels stacked across multiple time frames are estimated using the low rank constraint due to the nature of smoothly varying spatiotemporal correlation in k-t space during calibration, while the solution is projected onto the reconstructed k-t space with the sparsity constraint imposed on the estimated dynamic images in x-f space.

Sunrita Poddar¹, Sajan Goud Lingala², and Mathews Jacob^1
1Electrical and Computer Engineering, University of Iowa, Iowa City, Iowa, United States, ²Biomedical Engineering, University of Iowa, Iowa City, Iowa, United States

This work enables free-breathing un-gated cardiac MR imaging from highly under-sampled k-space data. It eliminates the need for multiple breath-holds to evaluate cardiac function, thus increasing patient comfort. We consider that the images live on a 2D manifold (parameterized by cardiac and respiratory phases) embedded in higher dimensional space. K-space data of images that lie close on the manifold are combined for reconstruction. The proposed approach does not require cardiac and respiratory gating or manual self-gating and can be automated for routine clinical use. This scheme can work for a range of existing k-space trajectories, including golden angle radial sequences.

Xiaoying Cai¹, Feng Huang², Kui Ying³, Chun Yuan⁴, and Huijun Chen^4
1Biomedical Engineering, University of Virginia, Charlottesville, Virginia, United States, ²Philips Healthcare, Gainesville, FL, United States, ³Department of Engineering Physics, Tsinghua University, Beijing, China, ⁴Center for Biomedical Imaging Research, Beijing, China

We proposed a new framework for reconstruction of accelerating perfusion imaging. By prediction intensity change of perfusion images using pre-contrast image and adaptive weights , the sparsity of dynamic images was strengthened, which benefits the following regular imaging reconstruction (k-t GRAPPA, k-t PCA or k-t SLR). The framework was tested with black blood vessel wall imaging data. The results showed that our new framework could result in better image reconstruction and improved kinetic parameter fitting for all tested reconstruction methods when acceleration factors were high.

Haikun Qi¹, Feng Huang², Xiaoying Cai¹, Dan Zhu¹, Feiyu Chen¹, Chun Yuan^1,3, and Huijun Chen^1
1Center for Biomedical Imaging Research, Tsinghua University, Beijing, Beijing, China, ²Philips Healthcare, Florida, United States, ³Department of radiology, University of Washington, Seattle, WA, United States

When multichannel coil is used for data acquisition, partially parallel imaging can be combined with k-t PCA to improve reconstruction accuracy. In this study, a sequential combination of partially parallel imaging and k-t PCA, which is called k-t PCA GROWL (GRAPPA Operator for Wider readout Line) is proposed. Compared with k-t PCA and k-t PCA/SNESE on artificially undersampled cardiac cine data, the proposed method resulted in the lowest error level. K-t PCA GROWL is a more efficient and more accurate scheme to combine k-t PCA and parallel imaging, and is promising in high spatiotemporal resolution dynamic MRI.

Gregory Duane^1,2, Yanwei Wang¹, Blake R. Walters¹, and Jae K. Kim^1
1Thunder Bay Regional Research Institute, Thunder Bay, Ontario, Canada, ²University of Colorado, Boulder, Colorado, United States

The impulse-propagator (matrix) method is extended to a multi-compartment idealized cell geometry describing nucleus, cytoplasm, and extracellular fluid, with semi-permeable membranes. The R matrices now represent propagation between compartments as well as intra-compartment propagation, with an approximate adjustment of the latter. Results are compared with those of Monte Carlo simulations. For biologically realistic nuclear membrane permeability, there are quantitative differences with the Monte Carlo results, but the contrast between signals for cases of large and small nuclei is qualitatively similar to the Monte Carlo contrast. The extended matrix method appears adequate to optimize q values for sensitivity to nucleus size.

Ketut Fundana¹, Oliver Bieri², and Philippe C. Cattin^1
1Medical Image Analysis Center, University of Basel, Basel, Switzerland, ²Radiological Physics, University Hospital of Basel, Basel, Switzerland

Imaging the structure and function of the human lungs is of importance for early detection of lung diseases. With the new development of steady state free precession (SSFP) imaging concepts in combination with dedicated image registration methods for fast functional and morphological MRI, it is expected that we are able to study the lung functions. We propose a novel method for image registration of the lung MRI sequences by using a convex optical flow model. The model is based on combined local and global optical flow method and regularized by an anisotropic total variation (TV) norm. The anisotropy derived from the structure tensor in order to take into account local variations at each point and to preserve the discontinuities of the motion fields. Qualitative and quantitative evaluations are done to show the robustness of the method.

Bridgette Webb¹, Andreas Petrovic¹, and Eva Scheurer^1,2
1Ludwig Boltzmann Institute for Clinical Forensic Imaging, Graz, Styria, Austria, ²Institute of Forensic Medicine, Medical University Graz, Graz, Styria, Austria

To improve the correlation of externally visible lesions in skin and subcutaneous fatty tissue with MRI findings, external strand-shaped markers were developed and tested. The visibility of the markers in photography and MRI, and quality of the final registration were investigated. Visualization of the markers and internal soft tissue damage was possible using a combination of sequences. Using manual feature detection followed by an affine transformation, MR images were registered to photographs documenting external and internal damage. Selected points in the registered MR images and photographs were used to evaluate the registration, giving an overall RMSE of 1.05 mm (95%CI[0.46;1.42]).

Ania Benítez^1,2, Gerardo Peláez-Brioso^1,2, Alexandra Borges³, Pilar López-Larrubia¹, Sebastián Cerdán¹, and Manuel Sánchez-Montañés^2
1Instituto de Investigaciones Biomédicas "Alberto Sols", Madrid, Madrid, Spain, ²Computer Science and Engineering, Escuela Politécnica Superior, Universidad Autónoma de Madrid, Madrid, Madrid, Spain, ³Instituto Português de Oncologia Centro de Lisboa, Lisboa, Portugal

MRI is presently one of the most important non-invasive methods to investigate and diagnose High Grade Gliomas (HGG) with the automatic classification of medical images into different pathological categories or grades playing an important role. A common problem to both approaches is many times, the small size of individual observations, while the data set from each individual is very large. We propose here an interesting protocol to predict therapy response in an animal HGG model, from the MRIs obtained during the first two days of anti-VGEF treatment. This approach in combination with LDA predicts therapy response outperformimg the classical approaches.

Darko Stern¹, Thomas Ebner¹, Eva Scheurer^2,3, and Martin Urschler^1,2
1Institute for Computer Graphics and Vision, Graz University of Technology, Graz, Styria, Austria, ²Ludwig Boltzmann Institute for Clinical Forensic Imaging, Graz, Styria, Austria, ³Medical University of Graz, Graz, Styria, Austria

In forensic practice determination of legal majority age is getting an increasing interest, e.g. for dealing with asylum seekers without proper identification. Traditional methods based on X-ray investigations of bone ossifications have the drawback of exposure to harmful ionizing radiation. MR based techniques overcome this limitation. Together with detection, automatic feature extraction and age estimation based on machine learning, MR may eventually become a powerful tool in this context. A first step in this direction is presented, by investigating age estimation based on the ossification of the wrist radius bone in an automatic fashion.

Babak A Ardekani¹, Sang Han A Lee¹, Donghyun Yu², Johan Lim², and Alvin H Bachman^1
1The Nathan S. Kline Institute for Psychiatric Research, Orangeburg, New York, United States, ²Statistics, Seoul National University, Seoul, Korea

We describe a multi-atlas-based method for corpus callosum segmentation and a fused Lasso logistic regression (FLLR) classifier that is able to differentiate patients with very mild/mild AD from normal controls (NC) using their CC thickness profile. We evaluated this technique using data from 196 individuals (98 AD and 98 NC) in the OASIS database. The FLLR classification accuracy was estimated to be 84% using cross-validation. Furthermore, the FLLR method highlights regions of the CC that are significantly thinner in AD relative to NC. The FLLR model presented can be extended to include other imaging or chemical biomarkers of AD.

Eelke Visser¹, Gwenaëlle Douaud¹, and Mark Jenkinson^1
1FMRIB Centre, Nuffield Department of Clinical Neurosciences, University of Oxford, Oxford, United Kingdom

Automated segmentation of subcortical structures is typically performed using T1-weighted volumes. However, not all of these structures' boundaries are clearly visible with T1 contrast. We describe a multimodal segmentation method that integrates information from different contrasts: in this case T1-weighted, T2-weighted and FA volumes. Since the images contain complementary information, the method does not need to rely heavily on prior shape knowledge obtained from training data. We show that, in specific areas, the method performs considerably better than FIRST, a segmentation method that only uses T1-weighted volumes.

Hui Xue¹, Peter Kellman¹, Souheil Inati², Thomas Sorensen³, and Michael Schacht Hansen^1
1Magnetic Resonance Technology Program, National Heart, Lung and Blood Institute, National Institutes of Health, Bethesda, Maryland, United States, ²National Mental Health Institute, National Institutes of Health, Bethesda, Maryland, United States, ³Department of Computer Science, Department of Clinical Medicine, Aarhus University, Aarhus, Denmark

Although non-linear reconstruction has shown potential for shortening acquisition times, its clinical usage is often hampered by the lengthy reconstruction time. To speedup non-linear reconstruction and enable its clinical usage, we have extended previously published Gadgetron framework to support cloud computing. With this extension (named GT-Plus), any number of Gadgetron processes can run cooperatively across multiple computers. GT-Plus framework was deployed on Amazon EC2 cloud and NIH’s Biowulf system. We demonstrate that with the GT-Plus cloud, a multi-slice free-breathing myocardial cine imaging with the whole ventricular coverage can be completed within 1min, including data acquisition and reconstruction.

Eric A. Borisch¹, Paul T. Weavers¹, and Stephen J. Riederer^1
1Mayo Clinic, Rochester, MN, United States

A comparison of the relative performance of multiple hardware options (CPU, GPU, MIC) available for accelerating reconstruction software performance is provided. The performance results were obtained during the process of implementing of a computationally intensive MR acceleration prescription-optimizing calculation across multiple hardware generations. Implications for purchasing and development activities where computational throughput as well as development time must be balanced are discussed.

Michael Oluwaseun Dada¹, Simona Baroni², and Bamidele Omotayo Awojoyogbe^1
1Department of Physics, Federal University of Technology, Minna, Niger State, Nigeria, ²Invento Laboratory, Molecular Biotechnology Center (MBC), Torino, Turin, Italy

Field-cycling magnetic resonance imaging (FC MRI) allows switching of the magnetic field during an imaging scan1. FC MRI has been very successful in relaxometry studies and there arises the need to offer more theoretical supports to the rich experimental results available in NMR laboratories. These theories are expected to offer new ways of interpreting the results for new discoveries. In view of this, we have developed a methodology based on the time – independent Bloch NMR flow equations for calculating the transverse magnetization in terms of the applied RF field. The results obtained in the study is useful for spectroscopic studies and 3D mapping of tissues

Qing He¹, Snehashis Roy¹, Amod Jog², and Dzung L Pham^1
1Henry M. Jackson Foundation for the Advancement of Military Medicine, Bethesda, Maryland, United States, ²Dept. of Computer Science, The Johns Hopkins University, Baltimore, Maryland, United States

An example based brain MRI synthesis method is proposed. No physical simulation of the image acquisition is involved and the synthesis is purely based on an example MR image. Patch based regression is used to predict the image based on anatomical models constructed from segmentation fusion. Results show that the synthetic images generated by our method are more realistic looking than those from the physics based methods.

Sudarshan Ragunathan¹, Dinghui Wang¹, Zhiqiang Li¹, and James G Pipe^1
1Neuroimaging Research, Barrow Neurological Institute, Phoenix, Arizona, United States

Current methods estimating noise content in an image require either multiple acquisitions or selecting an appropriate background region of the image. This work does not require either, and provides an estimate of the noise by eliminating signal content from a single dataset acquired by spiral or similar trajectories. The signal elimination was achieved by appropriate modification of the sampling density correction (SDC) weights. The modified weights was input to a gridding algorithm to obtain a noise image.

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

Magnitude reconstruction leads to biased signal at low signal-to-noise ratio (SNR). The Rician distribution describes the signal magnitude given an underlying value from signal amplitude. However, one typically wants the opposite: the underlying amplitude given the measured magnitude. We show how to derive this latter, inverse Rician, distribution and demonstrate some of its properties.

Christian G. Graff^1
1Division of Imaging and Applied Mathematics, U.S. Food and Drug Administration, Silver Spring, MD, United States

To facilitate dynamic and quantitative MR imaging there is significant interest in accelerated data acquisitions. Less data are acquired, which is often compensated for through iterative reconstruct techniques such as compressed sensing. These techniques involve non-linear regularization that complicates image analysis and challenges the validity of traditional image quality metrics such as pixel SNR. Here we develop the concept of Hoteling-SNR efficiency which quantifies the task-specific efficiency of compressed sensing reconstructions relative to a fully-sampled acquisition, measuring the tradeoff between image acquisition speed and clinical utility, overcoming inherent limitations of prior image quality metrics when analyzing compressed sensing techniques.

Florent Lalys¹, Sushmita Datta¹, Léorah Freeman¹, Stacey S. Cofield², Gary R. Cutter², Fred D. Lublin², Jerry S. Wolinsky¹, and Ponnada A. Narayana^1
1University of Texas Health Science Center at Houston, Houston, Texas, United States, ²University of Alabama at Birmingham, Birmingham, Alabama, United States

Intensity normalization (IN) is a critical step in image processing, and particularly in MR image segmentation. The IN technique described by Nyul et al. has been routinely used in numerous studies, but never critically evaluated on large cohorts or optimized for specific applications. In this study we significantly improved this IN method by identifying an optimal set of parameters, and verified it on a large cohort of multiple sclerosis patients. Our findings support implementing different parameters than those used in the majority of published studies.

Seiji Kumazawa¹, Takashi Yoshiura², Hiroshi Honda², and Fukai Toyofuku^1
1Department of Health Sciences, Faculty of Medical Sciences, Kyushu University, Fukuoka, Fukuoka, Japan, ²Department of Clinical Radiology, Graduate School of Medical Sciences, Kyushu University, Fukuoka, Fukuoka, Japan

EPI suffers from geometrical distortion due to magnetic field inhomogeneity. Our purpose was to develop an image-based method for estimating the magnetic field map based on the distorted EPI image and T1WI which requires no additional acquisitions. Our method synthesizes the distorted image to match the measured EPI image through the generating process of EPI image by changing the magnetic field. To evaluate the performance of our method, we applied it to simulation data. Our preliminary results demonstrate that the magnetic field inhomogeneity in EPI image can be estimated by our method without any additional acquisitions for estimation of the field map.

Xuan Vinh To¹, Xin Hong¹, Irvin Teh², Jian Rui Soh¹, and Kai-Hsiang Chuang^1
1Lab of Molucular Imaging, Singapore Bioimaging Consortium, Singapore, Singapore, Singapore, ²A*STAR - NUS Clinical Imaging Research Centre, Singapore, Singapore

Echo Planar Imaging (EPI) is an ultrafast acquisition technique widely used in fMRI, diffusion imaging and perfusion imaging but suffers from geometric and intensity distortions, especially in ultrahigh field strength used in small animal imaging. In this study we compared several distortion correction methods. We found using a pair of reference EPI images acquired with opposite phase encoding directions to calculate an unwrapping field provides the most accurate, robust and efficient distortion correction for the mouse brain.

Chia-Ling Chang¹, Tzu-Cheng Chao^1,2, Maria Alejandra Duràn-Mendicuti³, Ming-Ting Wu^4,5, and Bruno Madore^3
1Department of Computer Science and Information Engineering, National Cheng-Kung University, Tainan, Taiwan, ²Institute of Medical Informatics, National Cheng-Kung University, Tainan, Taiwan, ³Department of Radiology, Brigham and Women's Hospital, Harvard Medical School, Boston, Massachusetts, United States, ⁴Department of Radiology, Kaohsiung Veteran General Hospital, Kaohsiung, Taiwan, ⁵School of Medicine, National Yang-Ming University, Taipei, Taiwan

In an emergency room context, there is a need for a reliable non-contrast enhanced pulmonary MR angiography method to help handle the large number of patients with suspicion for pulmonary embolism and contra-indication for CT angiography. A promising method was previously introduced whose main weakness might have been the presence of discontinuities at the junction between the several slabs used to cover the chest volume. The present work aimed at alleviating this problem, thus making the overall approach more generally applicable. Through a specially-tailored registration algorithm as well as intensity-correction strategies, such discontinuities have been considerably suppressed here.

Bogdan Dzyubak¹, Armando Manduca², Kevin J. Glaser³, and Richard L. Ehman^3
1Physiology and Biomedical Engineering, Mayo Clinic, Rochester, MN, United States, ²Physiology and Biomedical Engineering, Mayo Clinic, Rochester, Minnesota, United States, ³Radiology, Mayo Clinic, Rochester, Minnesota, United States

Our clinical implementation of liver magnetic resonance elastography (MRE), used for noninvasively staging hepatic fibrosis, has recently been extended to analyze wave propagation in 3D using a fast EPI-based acquisition. This produces images with severe intensity inhomogeneity, low edge contrast, and fat suppression, which make fully automated processing, such as that developed for 2D GRE-based MRE, extremely challenging. A new method that does not depend on global intensity values and is able to find large liver areas to initialize liver segmentation despite severe intensity inhomogeneity has been developed and evaluated in clinical images produced by the standard 3D MRE sequence.

Tae-joon Eo¹ and Dosik Hwang^1
1School of Electrical and Electronic Engineering, Yonsei University, Seoul, Korea

This study demonstrates that the proposed denoising method can reduce noise effectively on both the phase and magnitude data. The magnitude and phase data were simultaneously denoised by neighboring complex decaying signals using the tissue relaxation property. Consequently, the resulting SW venographies showed substantial improvement.

Felipe Yanez^1,2 and Pablo Irarrazaval^1
1Pontificia Universidad Católica de Chile, Santiago, Chile, ²École Normale Supérieure, Paris, France

To improve the traditional Compressed Sensing (CS) framework for image reconstruction, we propose a CS technique with variable regularization parameter, which penalizes the pixels of the recovered image according to their magnitude. Herein, we present quantitative susceptibility map (QSM) reconstructions in in-vivo data, where the Sorted Compressed Sensing (SCS) technique produced results that demonstrate it is feasible to reconstruct high quality images. The proposed method produced gains up to 5-6 dB with respect of traditional CS.

Chen Cui¹, Xiaodong Wu^1,2, John D. Newell³, and Mathews Jacob^1
1Electrical and Computer Engineering, University of Iowa, Iowa City, IA, United States, ²Radiation Oncology, University of Iowa, Iowa City, IA, United States, ³Radiology, University of Iowa, Iowa City, IA, United States

We introduce a robust algorithm to resolve the ambiguities in fat-water decomposition by utilizing the smoothness of the field-map in three spatial dimensions. Many current methods are still sensitive to local minima effects. We had recently introduced a novel graph cut algorithm termed as GlObally Optimal Surface Estimation (GOOSE) that is guaranteed to provide the global minimum, which was observed to considerably improve the performance on challenging datasets. However, GOOSE was restricted to two dimensional and hence not capable of exploiting the field smoothness between slices. This work is to extend GOOSE to 3-D therefore to further improve the robustness.

Michael Anthony Mendoza¹, Raul Villalpando¹, Danny Park¹, Daniel Gardner¹, Kevin Perkins¹, and Neal Bangerter^1
1Electrical Engineering, Brigham Young University, Provo, Utah, United States

Motivated by the need for reliable and uniform water fat separation, which is used to improve medical diagnosis, we present a novel technique for water fat decomposition. This technique combines the advantages of balanced steady-state free precession (bSSFP) MRI with Dixon reconstruction to produce water fat decomposition with high SNR in a short imaging time, while simultaneously reducing banding artifacts that traditionally degrade image quality. This algorithm utilizes four phase-cycled bSSFP acquisitions at specific echo times to generate “in-phase” and “out-of-phase” images. Linear combinations of the strongest signals from these images are used to produce separate water and fat images.

Christopher D J Sinclair^1,2, Jasper M Morrow¹, Arne Fischmann³, Michael G Hanna¹, Mary M Reilly¹, Tarek A Yousry^1,2, and John S Thornton^1,2
1MRC Centre for Neuromuscular Diseases, UCL Institute of Neurology, London, United Kingdom, ²Department of Brain Repair and Rehabilitation, UCL Institute of Neurology, London, United Kingdom, ³University of Basel Hospital, Basel, Switzerland

Fat-water quantification is a promising method for sensitively measuring changes in muscle in upcoming clinical trials in neuromuscular diseases. In this work we applied texture analysis to 3-point Dixon fat fraction maps to assess fat-infiltration in the hamstring muscles in 17 patients with the representative condition inclusion body myositis, measured twice at a 1 year interval. Mean fat-fraction, contrast, homogeneity run length image texture measures all changed significantly over 1 year. There were no equivalent texture parameter changes in a healthy volunteer group over this period, demonstrating the promise of muscle fat-fraction texture analysis for measuring longitudinal fat-infiltration in neuromuscular conditions.

Sunil K. Valaparla^1,2, Qi Peng³, Oscar S. E. Nateras¹, Feng Gao¹, Timothy Q. Duong¹, and Geoffrey D. Clarke^1,2
1Research Imaging Institute, University of Texas Health Science Center at San Antonio, San Antonio, Texas, United States, ²Radiology, University of Texas Health Science Center at San Antonio, San Antonio, Texas, United States, ³Department of Radiology, Albert Einstein College of Medicine, Bronx, NY, United States

Excessive body fat, particularly visceral adipose tissue (VAT) correlates with diabetes and cardiovascular diseases. This study compared 3T breath-hold MRI T1W water-saturated (WS) and 2pt Dixon fat (DF) MRI with body and phase array coils in the application of subcutaneous (SAT), (VAT), inter-muscular (IMAT) adipose tissue volumetric assessment using semi–automated fuzzy c-means (FCM) clustering algorithm. Results indicate strong concordance and correlation between the SAT, VAT and IMAT calculated from different acquisitions. These methods can be used interchangeably along with FCM algorithm to effectively assess body fat for large-scale clinical imaging studies.

José P. Marques¹, Daniel Gallichan², and Rolf Gruetter^2,3
1CIBM, Department of Radiology, University of Lausanne, Lausanne, Vaud, Switzerland, ²LIFMET, École Polytechnique Fédérale de Lausanne, Vaud, Switzerland, ³CIBM, Department of Radiology, University of Lausanne and Geneva, Vaud, Switzerland

In this work we evaluate the possibility to perform motion tracking of a subjects head by using small MR markers containing a liquid which is off-resonance in respect to water. TrackDOTS is able to determine the position of various MR markers with 3 orthogonal k-space readouts by constructing receive coil modes that limit the sensitivity to regions of the various markers. The methodology was demonstrated in vivo showing high correlation between the motion parameters obtained using conventional MR image co-registration and the proposed methodology.

Paul Wighton¹, M. Dylan Tisdall¹, Erez Nevo², Kawin Setsompop¹, Stephen F Cauley¹, Himanshu Bhat³, Thomas Benner⁴, Dara S Manoach⁵, and André van der Kouwe^1
1Radiology, MGH, Charlestown, MA, United States, ²Robin Medical, Baltimore, MD, United States, ³Healthcare Sector, Siemens, Charlestown, MA, United States, ⁴Healthcare Sector, Siemens AG, Erlangen, Germany, ⁵Psychiatry, MGH, Charlestown, MA, United States

Recent research has shown that head motion can cause spurious group differences in functional as well as structural connectivity analyses. We examine slice-by-slice prospective hardware motion correction in EPI sequences (BOLD and 3x simultaneous multislice BOLD) in order to mitigate such differences. Our method incurs a 7.8% and 15.9% penalty to tSNR due motion in the BOLD and 3xSMS-BOLD sequences respectively, compared to 38.5% and 23.8% tSNR penalty respectively in the next best methods. Additionally, we observe that susceptibility due to motion has as large an effect as the motion itself.

Brian Keating¹, Aditya Singh¹, Benjamin Zahneisen¹, Linda Chang¹, and Thomas Ernst^1
1Department of Medicine, University of Hawaii, Honolulu, HI, United States

System latencies can reduce the efficacy of prospective motion correction (PMC) with external optical tracking, especially during fast movements (50mm/s or 50°/s range). We integrated a Kalman filter into a prospectively corrected gradient echo (GRE) sequence in order to estimate the velocity and acceleration of the head from lagged optical tracking data. The latency was accounted for by extrapolating forward in time before each prospective update. In addition, conjugate gradient-based retrospective motion correction was performed in Matlab to correct for residual tracking errors. GRE images show reduced motion artifacts when predictive filtering is used as compared to standard PMC.

Nick Todd¹, Oliver Josephs¹, Martina Callaghan¹, and Nikolaus Weiskopf^1
1Wellcome Trust Centre for Neuroimaging, London, United Kingdom

An approach to using the XPACE camera system for prospective motion correction of MRI data was implemented for a 3D echo-planar imaging pulse sequence. The method uses a high frame rate camera to track the six degrees-of-freedom movement of a marker that is attached to the patient. This information is fed to the scanner host computer to dynamically update the imaging gradients necessary for rigid body realignment of the data to be acquired. The PMC method was tested for the application of acquiring 3D EPI data for functional MRI studies where high SNR and temporal stability are of paramount importance.

Erik Beall¹ and Mark Lowe^1
1Imaging Institute, Cleveland Clinic, Cleveland, OH, United States

Censoring methods have gained a lot of attention recently, but, given the importance of head motion artifact in fMRI and fcMRI, more evaluation is necessary. Accurate identification of corrupted volumes, which we evaluate here, is a critical dependency. First we distinguish between intravolumetric (realistic) and volumetric (unrealistic) motion and point out that metrics reported to-date assume volumetric . We simulate intravolume and volumetric motion with a novel acquisition (SimPACE), and compare the true and false positive rates of several motion metrics. We report that volumetric metrics generally perform poorly, with negative implications for censoring and global signal regression.

Shireen Elhabian¹, Yaniv Gur¹, Joseph Piven², Martin Styner^2,3, Ilana Leppert⁴, G. Bruce Pike^4,5, and Guido Gerig^1
1Scientific Computing and Imaging Institute, University of Utah, Salt Lake City, Utah, United States, ²Psychiatry, University of North Carolina, North Carolina, United States,³Computer Science, University of North Carolina, North Carolina, North Carolina, United States, ⁴Neurology and Neurosurgery, Montréal Neurological Institute, Montréal, Quebec, Canada, ⁵Radiology, University of Calgary, Calgary, Canada

Diffusion weighted imaging (DWI) is known to be sensitive to motion originating from vibration, cardiac pulsation, breathing and subject movement, creating artifacts which require post-imaging correction. Users often do not fully understand the consequences of different choices for post-correction schemes for HARDI such as elimination versus alignment of affected DWIs with inherent choices of image interpolation, and how correction would affect ODF estimation, ability to resolve crossing fibers and final quantitative measures. We report about an experimental synthetic data platform and comparison with real data to systematically explore motion correction schemes under different scenarios to provide recommendations for best choices.

David Andrew Porter¹ and Stefan Huwer^1
1Healthcare Sector, Siemens AG, Erlangen, Germany

This paper introduces a new motion-correction technique for high-angular-resolution diffusion imaging studies. The method uses an optimized set of diffusion gradient vectors, which allows intermediate images to be calculated with a contrast that approximates to that of a trace-weighted image. The low contrast variation makes these intermediate images useful for motion tracking during the time series. Studies in healthy subjects suggest that the method delivers motion parameter estimates that are similar to those provided by interleaved low-b-value images. The temporal resolution can be optimized by using a sliding-window method, in which the intermediate image is updated after each diffusion-weighted volume.

Matthias Schloegl¹, Clemens Diwoky¹, and Rudolf Stollberger^1
1Institute of Medical Engineering, Graz University of Technology, Graz, Austria

A novel method for intrinsic qualitative motion information detection for 3D radial balanced SSFP sequences is presented. Validation was performed for potential applications of correction for rigid head motion and self-gating cardiac CINE imaging. The comparison was carried out with low-resolution navigators and in the second case by comparison to the matched ECG signal. Results show that the method effectively captures motion variations or can serve as an ECG surrogate not requiring any additional external devices or internal navigators.

Maryna Babayeva*^1,2, Alexander Loktyushin*³, Tobias Kober^2,4, Cristina Granziera⁵, Hannes Nickisch³, Rolf Gruetter^1,6, and Gunnar Krueger^2,4
1CIBM-AIT, École Polytechnique Fédérale de Lausanne and University of Lausanne, Lausanne, Switzerland, ²Advanced Clinical Imaging Technology, Siemens Healthcare IM BM PI, Lausanne, Switzerland, ³Max Planck Institute for Intelligent Systems, Tübingen, Germany, ⁴CIBM-AIT, École Polytechnique Fédérale de Lausanne, Lausanne, Switzerland,⁵Departments of Clinical Neurosciences, University Hospital Center (CHUV) and University of Lausanne (UNIL), Lausanne, Switzerland, ⁶Departments of Radiology, Universities of Lausanne and Geneva, Switzerland

This work investigates the possibility of using FID navigator signals to improve the performance of a recently proposed autofocusing-based retrospective motion correction technique. FID navigators were incorporated into an MPRAGE sequence and 3 subjects were scanned at 3T while performing head movements. The acquired data was retrospectively corrected for motion by exploiting the FID signals to constrain the unknown motion parameters. The results were compared against the reconstructions obtained from a non-FID-guided version of the algorithm, demonstrating that the use of FID navigators for retrospective motion correction leads to improvement in both image quality and computation time.

Yilong Liu¹, Zhe Liu², Zhe Zhang¹, Huailing Zhang³, and Hua Guo^1
1Center for Biomedical Imaging Research, Department of Biomedical Engineering, School of Medicine, Tsinghua University, Beijing, Beijing, China, ²Department of Biomedical Engineering, Cornell University, New York, United States, ³School of Information Engineering, Guangdong Medical College, Guangdong, China

Proper reference for motion estimation is critical for accurate and robust motion correction in PROPELLER MRI. In this work, single blade reference (SBR), combined blade reference (CBR), grouped blade reference (GBR) and Pipe et al¡¯s latest revised method in which no blade reference is required (NBR) are implemented and compared. Simulation results show that all methods above is accurate enough to generate a good image. For some special cases of in vivo imaging, SBR, CBR and GBR may fail to converge, while NBR still maintains relatively good performance, which indicates NBR is more robust.

Jalal B Andre¹, Mahmud Mossa-Bosha¹, Michael N. Hoff¹, C. Patrick Smith², and Wendy Cohen^1
1Radiology, University of Washington, Seattle, WA, United States, ²Radiology, Harborview Medical Center, Seattle, WA, United States

Patient motion frequently degrades clinical MR examinations, but its prevalence remains unknown. We sought to assess the prevalence of significant patient motion in MR examinations of the neuroaxis by manually evaluating images sent to PACS. 55 of the 175 total neuroaxis MRI examinations completed in one calendar week were sent to PACS with at least some motion degradation (31.4%). We conservatively estimate that motion-related imaging costs approach $296,092.00/year at our institution. Since patient motion may be common in the global MR community, we suggest that attention be directed toward providing more practical solutions to this common problem.

Ulrike Nöth¹, Steffen Volz¹, Elke Hattingen², and Ralf Deichmann^1
1Brain Imaging Center (BIC), Goethe University Frankfurt/Main, Frankfurt/Main, Germany, ²Institute of Neuroradiology, University Hospital Frankfurt/Main, Frankfurt/Main, Germany

A method for reducing motion artefacts in quantitative T2* maps is presented, based on repeated acquisitions with reduced spatial resolution. Raw data are averaged with weighting factors individually chosen for each k-space line, so the influence of corrupted lines is strongly reduced. A study performed on healthy subjects performing pre-trained motion shows that corrupted lines are reliably detected and suppressed. In a study on tumour patients, pathological details visible in a T2-weighted reference image are replicated in the motion corrected T2* map, whereas the uncorrected T2* map yields artefacts that can easily be misinterpreted as additional lesions.

Mathias Engström^1,2, Magnus Mårtensson^1,3, Ola Norbeck², Enrico Avventi¹, Axel Hartwig², and Stefan Skare^1,2
1Clinical Neuroscience, Karolinska Institutet, Stockholm, Stockholm, Sweden, ²Neuroradiology, Karolinska University Hospital, Stockholm, Stockholm, Sweden, ³EMEA Research and Collaboration, GE Applied Science Laboratory, GE Healthcare, Stockholm, Sweden

The chemical saturation pre-pulse is investigated as a source for rigid-body motion correction, utilizing three orthogonal, highly accelerated, EPI readouts.

Dallas C Turley¹, Michael Schär², and James G Pipe^1
1Neuroimaging Research, Barrow Neurological Institute, Phoenix, AZ, United States, ²Philips Healthcare, Cleveland, OH, United States

Abdominal MRI presents many imaging challenges due to non rigid-body motion of respiration. Continuous data acquisition during free breathing is fastest and most comfortable for patients but respiratory motion must be corrected in order to obtain high quality images. In a T2-weighted free-breathing PROPELLER exam, instructing patients to pause 2-3 seconds after exhalation improved image sharpness without adding substantial patient discomfort. Respiratory motion was captured with bellows and the waveform was used as additional data to weight PROPELLER reconstruction in favor of blades collected during end-exhalation, reducing the effects of respiratory motion.

Joseph Y. Cheng^1,2, Tao Zhang¹, Xinwei Shi¹, Martin Uecker³, Marcus T. Alley², John M. Pauly¹, Michael Lustig³, and Shreyas S. Vasanawala^2
1Electrical Engineering, Stanford University, Stanford, California, United States, ²Radiology, Stanford University, Stanford, California, United States, ³Electrical Engineering & Computer Sciences, University of California, Berkeley, California, United States

Soft-gating (or motion-based weighting) reconstruction can reduce motion artifacts in free-breathing MRI. However, the performance of this method depends on the accuracy of the motion-based weights. For respiratory motion, the degree of motion varies throughout space. For 3D Cartesian imaging, 1D spatial localization can be achieved by transforming the data into the hybrid (x,ky,kz)-space. Taking advantage of this, x-slice specific weights are derived and applied to correct for the varying non-rigid motion on a slice-by-slice basis. The proposed technique is applied to free-breathing abdominal imaging of pediatric patients. Improvement over no soft-gating and soft-gating with a single set of weights are shown.

Freddy Odille^1,2, Bailiang Chen^1,2, Anne Menini^1,2, Pauline Ferry^1,2, Marine Beaumont^2,3, Jacques Felblinger^1,2, and Laurent Bonnemains^1,2
1U947, Inserm, Nancy, France, ²IADI, Université de Lorraine, Nancy, France, ³CIC-IT 801, Inserm, Nancy, France

Cardiac MRI conventionally relies on a large number of breath holds. This is relatively inefficient as it results in a high scanner idle time and motion inconsistencies (misalignment between sequences). In this work we evaluate the feasibility and potential benefits of performing a whole cardiac examination during free-breathing using GRICS non-rigid motion correction. The approach was validated in 5 Duchenne muscular dystrophy patients with a cardiac protocol including free-breathing function (2D short-axis SSFP with whole heart coverage) and fibrosis assessment (post-contrast cine and 3D late enhancement). These preliminary results indicate diagnostic quality can be achieved along with good motion consistency.

Daniel V Litwiller¹, James H Holmes², Manojkumar Saranathan³, Andreas M Loening³, James F Glockner⁴, Shreyas S Vasanawala³, and Ersin Bayram^5
1Global MR Applications and Workflow, GE Healthcare, Rochester, MN, United States, ²Global MR Applications and Workflow, GE Healthcare, Madison, WI, United States,³Department of Radiology, Stanford University, Palo Alto, CA, United States, ⁴Department of Radiology, Mayo Clinic, Rochester, MN, United States, ⁵Global MR Applications and Workflow, GE Healthcare, Houston, TX, United States

Modulated refocusing flip angle single-shot fast spin echo (SSFSE) is of particular interest for pediatric and breath-held abdominal imaging. When low minimum refocusing flip angles are utilized with this technique in the presence of motion, however, as in the liver, images may suffer from signal loss due to cardiac motion. In this work, we characterize the empirical sensitivity of this modulated SSFSE technique to impulsive cardiac-like motion and validate our findings in vivo.

Takao Goto¹, Shiro Ozaki², Koji Uchida³, Hajime Kitagaki³, and Hiroyuki Kabasawa^1
1Global MR Applications and Workflow, GE Healthcare, Hino-shi, Tokyo, Japan, ²Shimane University Hospital, Izumo-shi, Shimane, Japan, ³Department of Radiology, Shimane University Faculty of Medicine, Izumo-shi, Shimane, Japan

We present a robust method for automated positioning of a Navigator tracker for MRI liver scans. Our method detects the dome peak of the liver and uses it for Navigator tracker positioning. We used AdaBoost to extract liver dome shape following detection of the edges close to the liver dome and identify the dome peak. During application of our method to 118 volunteers and 53 patients, the tracker did not show any significant mistakes and performed slice tracking successfully. We expect our practical approach to assist the scanning operator and improve workflow.

Junmin Liu¹, David W Holdsworth^1,2, and Maria Drangova^1,2
1Imaging Research Laboratories, Robarts Research Institute, Schulich School of Medicine & Dentistry, University of Western Ontario, London, Ontario, Canada,²Department of Medical Biophysics, Schulich School of Medicine & Dentistry, University of Western Ontario, London, Ontario, Canada

Phase unwrapping techniques are conventionally used for B0 mapping. However, perfect unwrapped phase images are still very difficult to obtain. We present a novel non-iterative phase-unwrapping-based field mapping method. The method uses a raw fat-fraction map estimated from multi-echo magnitude images followed by histogram analysis of phase images to generate an accurate B0 map; both global and local phase errors are removed. The method is tested with a large number of data sets, including all cases from 2012 ISMRM Challenge.

Olivier Reynaud¹, Daniel Gallichan¹, Benoit Schaller¹, and Rolf Gruetter^1
1CIBM, Lausanne, Switzerland

At 7T, conventional static field (B0) projection mapping techniques (FASTMAP, FASTESTMAP) suffer from elevated SAR and force long acquisition times (TA). Here the series of adiabatic pulses needed for pencil selection is replaced by a single 2D-RF pulse in the small tip angle regime. After in-vivo characterization of the selection profile, results (N=7, 6 VOIs) show no significant difference between the water spectral linewidths obtained with the conventional adiabatic (TA = 4 min) and the optimized 2D-RF FASTMAP sequence (TA = 42 s). In addition, SAR is reduced by two orders of magnitude without impact on shimming or spectrum quality.

Vincent O. Boer¹, Mariska P. Luttje², Peter R. Luijten¹, and Dennis W.J. Klomp^1
1radiology, UMC Utrecht, Utrecht, Utrecht, Netherlands, ²radiotherapy, UMC Utrecht, Utrecht, Utrecht, Netherlands

Bilateral shimming of the breast puts higher restraints on shimming capabilities as compared to unilateral shimming. Where unilateral shimming can be performed well with 2nd order spherical harmonics, we show that up to fourth order spherical harmonics do not reach the same field homogeneity for bilateral shimming. A midplane shimcoil was designed and used to generate a locally varying field, which was able to compensate the fields in bilateral shimming to a degree beyond that possible with fourth order shimming.

Tijl A. van der Velden¹, Dennis W.J. Klomp¹, Peter Luijten¹, and Vincent O Boer^1
1Radiology, University Medical Center Utrecht, Utrecht, Utrecht, Netherlands

Respiration causes motion of boundaries with large susceptibilities differences, such as the shoulders, the heart and the diaphragm. These motions can cause artefacts in a variety of MR scans. With field probes the field fluctuations caused by this motion can be observed. However, these probes measure the field outside the body. In this simulation study we suggest to fit a dipole to field probe measurements to translate field fluctuations from outside the body to field fluctuations inside the breast.

Paul W. de Bruin¹, Maarten J. Versluis¹, Sebastian A. Aussenhofer¹, Peter Börnert^1,2, and Andrew G. Webb^1
1Radiology Department, Leiden University Medical Center, Leiden, Netherlands, ²Innovative Technologies Research Laboratories, Philips Technologie GmbH, Hamburg, Germany

Off-resonance correction of spiral 23Na trajectories using a 1H B0-map is used to achieve a completely free modest increase in SNR which is beneficial for quantitative 23Na imaging.

Ryan K Robison¹, Zhiqiang Li¹, Michael Schär^1,2, and James G Pipe^1
1Neuroimaging Research, Barrow Neurological Institute, Phoenix, Arizona, United States, ²Philips Healthcare, Cleveland, Ohio, United States

Spiral-in/spiral-out acquisitions can be affected by artifacts that manifest differently between the spiral-in and spiral-out portions of the acquisition. If uncorrected these artifacts will be accentuated in the final combined image. Time dependent phase errors from B0 eddy currents are one source of artifacts that manifest differently between spiral-in and spiral-out acquisitions. These phase errors were measured and corrected. The resulting images demonstrate improvements in apparent resolution.

M. Dylan Tisdall^1,2 and André J. W. van der Kouwe^1,2
1A. A. Martinos Center for Biomedical Imaging, Massachusetts General Hospital, Charlestown, Massachusetts, United States, ²Radiology, Harvard Medical School, Boston, Massachusetts, United States

The phase-correction navigators of EPI sequences have previously be shown to enable real-time estimation and prospective correction of resonance frequency drift. In the present work we describe a novel algorithm that provides more accurate estimates than previous methods. We also demonstrate how a small modification to our EPI-based volumetric navigator (vNavs), allows us to use it not just for prospective motion correction, but also prospective frequency correction. Results from a vNav-enabled MPRAGE sequence are shown.

Ulrich Koellisch^1,2, Rolf F. Schulte², Markus Durst^1,2, Axel Haase¹, and Florian Wiesinger^2
1IMETUM, Technical University München, Munich, Germany, ²GE Global Research, Munich, Germany

An algorithm for IDEAL spiral CSI reconstruction of hyperpolarized 13C metabolites is presented. It calculates field maps or CS frequencies and metabolite images with a joint estimation approach. The field maps gets estimated by a set of polynomials. The estimation of the CS-frequencies decreases the number of necessary excitations, because no extra spectra have to be recorded. The application of the off-resonance correction on in-vivo datasets has shown, that this approach increases the image quality due to a reduced amount of blurring in the metabolite maps.

Marta Bianciardi¹, Jonathan R Polimeni¹, Kawin Setsompop¹, Cornelius Eichner¹, Berkin Bilgic¹, and Lawrence L Wald^1
1Department of Radiology, A.A. Martinos Center for Biomedical Imaging, MGH, Harvard Medical School, Boston, MA, United States

Chest motion due to respiration produces off-resonance (OR) effects in the brain, resulting in signal-instability in both structural and functional MRI. Previous work performed dynamic OR-correction using a second-order spatial model, and a univariate temporal respiratory model. Aim of this work was to evaluate dynamic OR correction for increased degrees-of-freedom in both spatial and temporal domain. Our results demonstrate the benefits of higher order (up to the fifth) spatial models combined with the use of bivariate temporal modeling of respiratory effects. These findings show the expected dynamic capabilities of high-order shim-arrays at high magnetic field and on a whole-brain basis.

Anjali Datta¹, Reeve Ingle¹, Bob Hu^1,2, and Dwight Nishimura^1
1Electrical Engineering, Stanford University, Stanford, California, United States, ²Cardiology, Palo Alto Medical Foundation, Palo Alto, California, United States

Respiration-induced B₀ variations are of interest because they may lead to off-resonance artifacts in free-breathing acquisitions and may contribute to variable image quality across patients. Using the XCAT1 4D computational phantom to generate susceptibility models, we simulate the main field map over the heart in several respiratory frames and in different anatomies to determine if B₀ variations across the breathing cycle and between individuals may be significant. This work suggests that respiration induces spatially-variant B₀ shifts in the heart and that the magnitude and distribution depend on the left ventricular long-axis orientation.

Saikat Sengupta^1,2, David Smith^1,2, and E. Brian Welch^1,2
1Radiology and Radiological Sciences, Vanderbilt University, Nashville, TN, United States, ²Vanderbilt University Institute of Imaging Science, Vanderbilt University, Nashville, TN, United States

Continuous moving bed MRI (COMBI) is a high throughput imaging technique for rapid whole-body examination. A primary limitation of COMBI is that the frequency and B0 shims are typically optimized for only one body location. As a result, image quality and parameter quantification suffers from field variations along the extent of the body. In this abstract, we present slice-specific zeroth and first order dynamic B0 shimming in COMBI. Slice-optimized shimming is demonstrated on a 1.500 meter field of view phantom setup with significant field homogeneity gains throughout the full field of view over a static single station shim.

Simon Gross¹, Christoph Barmet^1,2, Thomas Schmid¹, and Klaas Paul Prüssmann^1
1Institute for Biomedical Engineering, ETH and University Zurich, Zurich, Switzerland, ²Skope Magnetic Resonance Technologies, Zurich, Switzerland

By doping epoxide-resins with paramagnetic substances, solid materials with widely adjustable magnetic susceptibility can be produced. With this method, we built an MR phantom whose susceptibility is matched to air (including the liquid filling). It has the unique properties of not affecting the magnetic field homogeneity, neither internally nor externally, turning it into a versatile tool for sequence development and quality assurance in MR imaging and spectroscopy. Furthermore, the excellent mechanical properties of this material allow the construction of NMR equipment and consumables that are invisible within the magnetic field.

Michael N Hoff¹, Greg J Wilson¹, Qing-San Xiang^2,3, and Jalal B Andre^1
1Department of Radiology, University of Washington, Seattle, WA, United States, ²Department of Radiology, University of British Columbia, Vancouver, BC, Canada,³Department of Physics, University of British Columbia, Vancouver, BC, Canada

The geometric solution (GS) is clinically applied to correct artifacts in balanced steady state free precession (bSSFP) images of the temporal bone at 3T. Four phase cycled bSSFP brain images are acquired from a patient, and a pixel-by-pixel GS is computed in the complex plane and compared with a complex average of the images. The solution not only eliminates bSSFP banding and dependence on signal off-resonance, it also shows an insensitivity to motion. Further testing is planned in order to determine the GS’ capacity for correcting motion artifacts.

Eric G. Stinson¹, Joshua D. Trzasko¹, and Stephen J. Riederer^1
1Mayo Clinic, Rochester, Minnesota, United States

Dixon-based methods avoid errors by accounting for B₀ inhomogeneities during the fat-water separation, but can extend scan time. Single-echo Dixon techniques avoid this limitation, but have yet to use partial Fourier sampling. Homodyne reconstruction for partial Fourier single-echo Dixon imaging is derived and experimentally demonstrated. The processing is performed within the framework of a phase constrained reconstruction and accomplishes fat-water separation and homodyne phase correction in a single step. Partial Fourier sampling with homodyne processing can reduce scan time by a factor of almost 2 and shows promise for time-resolved Dixon imaging for dynamic applications such as CE-MRA.

Martin Uecker¹ and Michael Lustig^1
1Electrical Engineering and Computer Sciences, University of California, Berkeley, California, United States

ESPIRiT is a recently developed algorithm for auto-calibrated parallel MRI. It is based on an eigenvalue analysis of the calibration data. The present work describes an extension using virtual conjugate coils, which allows the estimation of sensitivity maps which include image phase. Such maps can be used in a phase-constrained SENSE reconstruction for acceleration with partial Fourier sampling. In contrast to other methods, high-frequency phase is not simply discarded but taken into account by a second eigenvector map in affected image regions. This map can be used in an extended soft-SENSE reconstruction which is robust against errors from high-frequency phase.

Guobin Li¹, Maxim Zaitsev¹, Esther Meyer², Dominik Paul², Jan Korvink^3,4, and Jürgen Hennig^1
1University Medical Center Freiburg, Freiburg, Baden-Württemberg, Germany, ²Siemens Healthcare, Germany, ³Department of Microsystems Engineering — IMTEK, University of Freiburg, Baden-Württemberg, Germany, ⁴Freiburg Institute of Advanced Studies (FRIAS), University of Freiburg, Baden-Württemberg, Germany

Half Fourier acquisition has widely been used for a long time to shorten the measurement time in turbo spin echo sequences.The combination of half Fourier acquisition with compressed sensing has also been investigated. One drawback of typical Homodyne and POCS based methods is that they brutally replace the phase of the resulting images by an estimate. We introduce a relaxed phase constraint term in the regularized reconstruction, to encourage the phase consistency between the result and the estimation, instead of an overall phase replacement in the compressed sensing reconstruction with half Fourier acquisition.

Weitian Chen¹, Peng Lai¹, and Yuval Zur^2
1Global MR Applications & Workflow, GE Healthcare, Menlo Park, CA - California, United States, ²GE Healthcare, Haifa, Haifa, Israel

Fast spin echo imaging plays a central role in clinical imaging. Eddy current and other factors, however, can introduce phase error in FSE sequences, which causes the violation to CPMG condition and result in image artifacts. The existing phase correction method works well for 2D FSE, which assumes a linear phase error along readout direction. However, to extend this approach to 3DFSE, we may need assume a linear phase error along both readout and slice direction, which may not always be the case. Data acquisition with phase cycling can be used to address this problem at the cost of doubled scan time. In this work, we investigated compressed sensing acceleration of data acquisition for 3DFSE phase correction using phase cycling.

A. Alhamud¹, Paul A. Taylor¹, André J.W. van der Kouwe², and Ernesta M. Meintjes^1
1Human Biology,MRC/UCT Medical Imaging Research Unit, University of Cape Town, Cape Town, Western Cape, South Africa, ²Massachusetts General Hospital, Charlestown, Massachusetts, United States

Single- or multi-shot Echo Planar Imaging (EPI) is the first choice for several imaging modalities such as diffusion tensor imaging (DTI) and functional MRI (fMR) etc, due to its ability to acquire images very rapidly. EPI is affected by several artifacts such as distortion from magnetic field inhomogeneity. Although several techniques have been reported for correcting the B0 homogeneity, these methods have some limitations and drawbacks. In this work we introduce a sophisticated method using the 3D EPI phase maps to correct TR-by-TR in real time the distortion in B0 for any imaging modality that implement EPI.

Myung-Ho In¹, Oleg Posnansky¹, Erik B. Beall², Mark J. Lowe², and Oliver Speck^1,3
1Biomedical Magnetic Resonance, Otto-von-Guericke University, Magdeburg, Germany, Magdeburg, Germany, ²Radiology, Imaging Institute, Cleveland Clinic, Cleveland, OH, USA, OH, United States, ³Leibniz Institute for Neurobiology, Magdeburg, Germany, Magdeburg, Germany

In echo-planar imaging (EPI), compressed distortion is a more difficult challenge than local stretching as spatial information can be lost in strongly compressed areas. To resolve this problem, two EPIs with opposite phase-encoding (PE) polarity were acquired and combined after distortion correction. A modified point spread function (PSF) mapping and distortion correction method was developed. A single PSF reference acquisition was extrapolated to reverse PE (extended) for reverse PE correction and an improved weighted combination of the two distortion-corrected images that properly accounts for the differential loss of information in forward and reverse PE images.

Rita G. Nunes^1,2 and Joseph V. Hajnal^2,3
1Institute of Biophysics and Biomedical Engineering, Faculty of Sciences, University of Lisbon, Lisbon, Portugal, ²Division of Imaging Sciences and Biomedical Engineering, King's College London, London, United Kingdom, ³Centre for the Developing Brain, King's College London, London, United Kingdom

With the drive to higher B0 magnetic fields, geometric distortion correction of Echo Planar Images (EPI) becomes essential. Point Spread Function (PSF) mapping has been shown to be more robust than B0 field mapping, however previous attempts to accelerate acquisition, still required a minimum of 10 EPI repeats. We demonstrate that by using a pattern search approach it is possible to estimate the peak of the PSF at each voxel from as little as 2 EPI repeats, one of which is the EPI image itself. This would enable estimation of distortion maps to be incorporated in standard EPI preparation phases.

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

In regions with magnetic field inhomogeneities, e.g. close to significant susceptibility differences as in the vicinity of major air cavities, T2*-weighted single-shot EPI with GRAPPA is susceptible to ghosting artifacts. To some extend, these artifacts seem to be related to the offset of the resonance frequency that is not handled appropriately by the GRAPPA reconstruction algorithm. Here, it is shown that with slice-specific frequencies for the data acquisition, i.e. the analog-to-digital converter’s demodulation frequency and phase settings, these artifacts can be reduced or even avoided.

Steen Moeller¹, An T Vu¹, Edward Auerbach¹, Kamil Ugurbil¹, and Essa Yacoub^1
1Center for Magnetic Resonance Research, University of Minnesota, Minneapolis, Minnesota, United States

The use of a slice specific navigator correction for both the SENSE/GRAPPA algorithm and the slice-GRAPPA algorithm is demonstrated and compared with high quality data from the Human Connectome project.

Eric Peterson¹, Samantha Holdsworth¹, Rafael O'Halloran¹, Julian Maclaren¹, Eric Aboussouan¹, William Grissom², and Roland Bammer^1
1Radiology, Stanford University, Stanford, CA, United States, ²Biomedical Engineering, Vanderbilt, Nashville, TN, United States

When using slice-accelerated echo planar imaging (EPI), slice-wise Nyquist ghost correction requires extra scan time because it necessitates an additional pre-scan (reference scan) in addition to parallel imaging calibration. This work presents a method to perform slice-wise Nyquist ghost correction using the parallel imaging pre-scan. This obviates the Nyquist ghost correction pre-scan and also allows for Nyquist ghost correction before the parallel imaging is performed, which allows for standard parallel imaging reconstructions. More importantly, by not requiring an additional pre-scan scan, it allows for a faster, more streamlined scan session.

W Scott Hoge^1,2, Huan Tan³, Robert A Kraft⁴, and Jonathan R Polimeni^2,5
1Brigham and Women's Hospital, Boston, Massachusetts, United States, ²Harvard Medical School, Boston, MA, United States, ³Department of Surgery (Neurosurgery), University of Chicago, Chicago, Illinois, United States, ⁴Virginia-Tech Wake Forest School of Biomedical Engineering, Winston-Salem, NC, United States, ⁵A.A. Martinos Center for Biomedical Imaging, MGH, Charlestown, MA, United States

This works seeks to further improve the correction of Nyquist ghosting in multi-shot (or segmented) echo planar imaging. EPI is vulnerable to static local field inhomogeneity and eddy current effects induced by the EPI readout, which results in Nyquist ghosting. Multi-shot EPI is further complicated by phase inconsistencies that often occur between the acquisition of each segment. PLACE and GESTE have been shown previously to provide superior ghost correction in single-shot EPI images. We demonstrate their further effectiveness in segmented EPI with in-vivo brain data acquired at 7T using a 32-channel head coil.

Wanyong Shin¹, Sehong Oh¹, and Mark J Lowe^1
1Radiology, Cleveland Clinic, Cleveland, Ohio, United States

To correct Nyquist ghost artifacts in single-shot EPI, 2 or 3 lines of readout acquisition without a phase encoding gradient is commonly applied. While the phase correction information is updated for each measurement for timeseries of EPI, the phase shift could be corrected either by using single phase correction scan (called static here) or each phase correction scan (dynamic). In the abstract, we evaluated the performance of the dynamic and static phase correction in timeseries EPI. We found the static phase correction provides higher tSNR and less spatial variation of tSNR than the dynamic phase correction with parallel imaging technique.

Victor B. Xie^1,2, Adrian Tsang^1,2, and Ed X. Wu^1,2
1Laboratory of Biomedical Imaging and Signal Processing, The University of Hong Kong, Hong Kong, Hong Kong SAR, China, ²Department of Electrical and Electronic Engineering, The University of Hong Kong, Hong Kong, Hong Kong SAR, China

Echo planar imaging (EPI) has been used in many applications such as fMRI and DWI. However, EPI images are prone to Nyquist ghost and geometric distortion.In this study, we proposed a new scheme based on reversed readout strategy to correct both Nyquist ghost and geometric distortion artifacts on EPI images simultaneously. We have demonstrated the proposed method can effectively remove Nyquist ghost and correct geometric distortion both in phantom and rat brain. This method may be paticularly suited for dynamic EPI protocols such as fMRI and dynamic contrast-enhanced MRI applications.

Amir Seginer¹, Rita Schmidt¹, Eddy Solomon¹, Avigdor Leftin¹, and Lucio Frydman^1
1Chemical Physics Department, Weizmann Institute of Science, Rehovot, Israel

Single-shot sequences utilizing spatiotemporal encoding (SPEN) provide robust alternatives to single-shot EPI with similar acquisition durations and comparable resolution, provided that specialized super-resolved reconstruction algorithms are used in the image retrieval. Acquisition imperfections known to cause ghosting artifacts in EPI, can also give rise to artifacts in processed Hybrid-SPEN images. We demonstrate that these experiments, unlike EPI, do not require a reference scan to correct for these imperfections. A self-referencing algorithm based on the fact that under-sampling along the SPEN direction does not generate aliasing, but rather lower resolution images, is developed. The referenceless algorithm and sample results are presented.

Franck Mauconduit¹, Aurélien Massire², Nicolas Boulant², Alexis Amadon², and Alexandre Vignaud^2
1Siemens Heathcare, Saint Denis, France, ²CEA, DSV, I2BM, Neurospin, LRMN, Gif-Sur-Yvette, Ile de france, France

UHF suffers from significant inhomogeneity artifacts originating from transmission RF and reception sensitivity coils. In vivo measurement approaches have been proposed earlier. Based on these approaches, we suggest a very short TR GRE sequence to reduce image unihomogeneity. On a Siemens 7T Magnetom scanner, a B1 field map (AFI) and a GRE sequence with minimal tissue contrast are acquired. Then the GRE sequence is low pass filtered and a post processing corrects reception sensitivities on an MPRAGE acquisition. Our result shows a significant improvement using a GRE sequence with TR=9ms and 10 sec acquisition therefore providing a fast correction method.

weiwei zhang¹, Bing Wu¹, and Yongchuan Lai^1
1GE Healthcare, Beijing, Beijing, China

In this abstract, we proposed an easy-to-implement method to eliminate cusp artifacts, or called annefact, by optimizing the pulse sequence design in FSE. The gradient polarities for excitation and refocusing RF slice selection are the same. However, the amplitudes of excitation and refocusing gradients are optimized individually so that location shift of excited spins by the two RF is minimized, whereas the annefact regions correspond to the two RF are separated. The former ensures no signal drop takes place. The latter ensures no cusp artifact is formed. This method requires no additional post-processing and hardware modification.

Jianfeng Bao^1,2, Congbo Cai², Zhong Chen², and Jianhui Zhong^1
1Department of Imaging Sciences, University of Rochester, Rochester, NY, United States, ²Department of Electronic Science, Xiamen University, Xiamen, Fujian, China

Spatiotemporally encoded (SPEN) imaging is an emerging ultrafast MRI method that can be used to obtain 2D image in a single shot, with much less sensitivity to B0 inhomogeneity compared with EPI. However, a super-resolved reconstruction method is needed to improve spatial resolution of the SPEN image. Recently, we have demonstrated that the de-convolution reconstruction method is superior to other existing methods in simulations and studies on 7 T small animal scanner. Here, we show that the de-convolution can obtain artifact-free and high resolution images on human brain at 3.0T. This emerging ultrafast MRI method may be applied to reduce artifacts caused by B0 inhomogeneity in fMRI and DTI images.

Chemseddine Fatnassi^1,2, Gunnar Krueger^2,3, Reto Meuli¹, and Kieran O'Brien^2,4
1Université de Lausanne/Centre Hospitalier Universitaire de Lausanne, Lausanne, Switzerland, ²CIBM - AIT, École Polytechnique Fédérale de Lausanne, Lausanne, Switzerland, ³Advanced Clinical Imaging Technology, Siemens Healthcare IM BM PI, Lausanne, Switzerland, ⁴Hôpitaux Universitaires de Genève, Genève, Switzerland

In 3D GRE, B0macro field gradients due to air/tissue interfaces lead to an apparent increase in the intravoxel dephasing and leads to large signal loss or inaccurate R2* estimation. If these B0macro are measurable, their influence can be removed. The algorithms normally assume the phase evolves linearly with time; however, in the presence of a large B0macro, this assumption is broken, Furthermore, the central difference approximation used to estimate gradient leads to edge artifacts at the brain's edges. To overcome these problems, we hypothesize a non-linear phase evolution including a fast computation of the through-plane gradient.

Angus Z. Lau^1,2, Elizabeth M. Tunnicliffe¹, Damian J. Tyler^1,2, and Matthew D. Robson^1
1Department of Cardiovascular Medicine, University of Oxford, Oxford, United Kingdom, ²Department of Physiology, Anatomy, and Genetics, University of Oxford, Oxford, United Kingdom

Blipped CAIPI acquisitions reduce g-factor noise amplification in simultaneous multi-slice (SMS) experiments by introducing inter-slice image shifts using through-plane phase encoding gradients during the image readout. For closely-spaced slices, large Gz blips result in N-fold image ghosts (where N is the SMS factor), which overlap with the desired multlband aliasing pattern. In this abstract, we introduce a simple deconvolution model and novel image entropy-based approach to correcting phase errors in a SMS scan. The method does not require an additional reference scan, and we anticipate that this development will enable SMS accelerated scans of the heart with closely spaced slices.

Min-Oh Kim¹, Semin Kwak¹, and Dong-Hyun Kim^1
1Electrical and Electronic Engineering, Yonsei University, Seoul, Seoul, Korea

View angle tilting (VAT) technique was proposed to correct chemical shift artifact and in-plane field inhomogeneity artifact. In spite of its advantages, VAT suffers from blurring artifact. In this study, post-processing methods that can alleviate VAT blurring based on VAT signal equation are presented.

Yuval Zur¹ and Weitian Chen^2
1GE Healthcare, Tirat Carmel, Israel, ²GE Healthcare, Menlo Park, California, United States

3D Fast Spin Echo (3DFSE) with flip angle modulation is used for high resolution T2 weighted imaging due to high T2 contrast and the ability to reformat the data in any desirable plane. Unavoidable violations of the CPMG condition due to system imperfections generate artifacts. In this work we present a technique to overcome this problem using a post processing method applied in conjunction with a two excitations approach. The drawback of the method is that two excitations are required. However, in many applications this method is necessary in order to acquire reliable high quality images.

Yoojin Lee¹, Jang Yeon Park², Yeji Han³, ChangHyun Oh³, and HyunWook Park^3
1Department of Radiology, University of Pittsburgh, Pittsburgh, Pennsylvania, United States, ²School of Biomedical Engineering, Konkuk University, Chungju, Korea,³Department of Electrical Engineering, Korea Advanced Institute of Science and Technology (KAIST), Daejeon, Korea

In this study, a new fat-suppressed spin-echo imaging technique using hyperbolic secant (HS) RF pulses for π/2 excitation and π refocusing in a conventional SE sequence is proposed. This technique utilizes the fact that non-linear phase profile across slice created by HS pulses is not compensated in the regions where the frequency offset exists. Phantom and in vivo experiments were performed at 3T to demonstrate this technique.

Rainer Schneider^1,2, Jens Haueisen², and Josef Pfeuffer^1
1MR Application Development, Siemens Healthcare, Erlangen, Bavaria, Germany, ²Institute of Biomedical Engineering and Informatics, TU Ilmenau, Ilmenau, Thuringia, Germany

Shaped fat saturation was realized for the first time on the basis of multi-frequency optimization of multidimensional spatially selective RF pulses in parallel transmission. Shaped fat saturation pulses were optimized based on a target-driven variable-density 2D spiral trajectory and evaluated in phantom and human in-vivo experiments and compared to the commonly used Gaussian FATSAT pulse. The fat saturation performance of the proposed RF pulses was found to be similar to the FATSAT pulse, but allowed also for the simultaneous saturation of other frequency bands. Furthermore, less interference with the water band was observed.

Haifeng Wang¹, Yihang Zhou², Yuchou Chang³, and Dong Liang^4
1Department of Diagnostic Radiology, Yale University, New Haven, CT, United States, ²Department of Electrical Engineering, University at Buffalo, The State University of New York, Buffalo, NY, United States, ³Barrow Neurological Institute, Phoenix, AZ, United States, ⁴Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen, Guangdong, China

Many methods of fat suppression are applied, but the cost is more computational time or sequence durations. In this abstract, a novel fat suppression method, named as Hybrid, is proposed to suppress the fat signal and enhance the image contrast. The method exploits non-Fourier random encoding pulse sequences to generate inhomogeneous B1 filed, thus the signal contrast of fat and other tissues has been changed comparing the conventional Fourier encoding method. The actual in vivo human knee experiments illustrate the proposed method augment the signal contrast of fat and other tissues, suppress more the fat signal than the conventional Fourier method, and has no growth of imaging time and decoding computational complexity.

Feng Zhao¹, Jeffrey A Fessler², Jon-Fredrik Nielsen¹, and Douglas C Noll^1
1Biomedical Engineering, The University of Michigan, Ann Arbor, MI, United States, ²EECS, The University of Michigan, Ann Arbor, MI, United States

Combining fat saturation and magnetization transfer (MT) preparation is beneficial in many clinical applications. The use of both fat sat and MT with steady-state incoherent (SSI) sequences may be limited by long minimal TR. Also, the conventional fat sat is sensitive to field inhomogeneity. We investigated a multi-dimensional spectral-spatial fat sat pulse for SSI sequences, i.e., small-tip fast recovery imaging (STFR) and spoiled GRE, to produce B0 insensitive fat suppression and MT preparation simultaneously. The methods were applied to cartilage imaging or MR angiography in brain at 3T.

Emeline Julie Ribot¹, Jean-Michel Franconi¹, and Sylvain Miraux^1
1CNRS/University Bordeaux, RMSB, Bordeaux, France

Musculo-skeletal MRI with fat signal suppression techniques are usually performed in 2D to shorten acquisition time and obtain sufficient signal and induce restrictions in echo time or flip angle values. Due to high SNR acquired in short acquisition time in 3D, bSSFP sequence was combined to a frequency-selective binomial pulse at 3T. Centering the frequency of this pulse to the resonance frequency of water or fat protons generated fat-free or water-free images, respectively. To remove banding artefacts, four images acquired at different resonance frequencies were summed using « Sum-Of-Square ». This new sequence allowed to obtain 3D high-resolution knee images without fat signal and banding artefacts.

Emeline Julie Ribot¹, Didier Wecker², Jean-Michel Franconi¹, and Sylvain Miraux^1
1CNRS/University Bordeaux, RMSB, Bordeaux, France, ²Bruker Biospin, France

To obtain fat-free MR images on small animals, several techniques have been developed. Low SNR, susceptibility arfetacts and echo time value restrictions limit their application at high magnetic field. 3D-bSSFP MR sequence, generating high signal in short acquisition time, was performed after a frequency-selective binomial pulse at 7T. Water-selective or fat-selective mouse whole-body 3D-bSSFP images were obtained with no banding artefacts due to the combination with « Sum-Of-Square » technique. Chemical shift artifact was removed allowing for better delineation and quantification of lymph node volumes. In addition, tumors growing in the renal sup-capsule were easily identified from abdominal fat.

Jong Bum Son¹, John Hazle¹, and Jingfei Ma^1
1Imaging Physics, The University of Texas MD Anderson Cancer Center, Houston, TX, United States

Fast spin echo triple-echo Dixon (fTED) acquires one in-phase (IP) and two out-of-phase (OP) images in a single acquisition without interleave. A two-point Dixon processing algorithm is then used to independently process two pairs of IP/OP images. A potential drawback is that any processing failure in each processing will lead to incomplete local or global water and fat separation. In this work, we proposed and developed a jointly-processed region growing based two-point Dixon phase correction algorithm. The proposed approach was capable of reconstructing uniformly separated water and fat images even when anatomic regions are separated by large signal-void.

Holger Eggers¹, Michael Schär^2,3, and James G. Pipe^3
1Philips Research, Hamburg, Germany, ²Philips Healthcare, Cleveland, OH, United States, ³Neuroimaging Research, Barrow Neurological Institute, Phoenix, AZ, United States

Combinations of PROPELLER and Dixon methods have recently been proposed to achieve motion insensitive water-fat imaging. They perform the water-fat separation after the PROPELLER reconstruction so far. However, applying the water-fat separation to single blades instead potentially offers several advantages. In this work, the feasibility of this approach is demonstrated in abdominal T2-weighted TSE imaging. By exploiting the Cartesian k-space sampling in single blades, an off-resonance correction for image quality enhancement is shown to become simpler in this way.

Holger Eggers^1
1Philips Research, Hamburg, Germany

The echo shifts commonly applied in turbo spin-echo Dixon imaging for chemical shift encoding affect the signal-to-noise ratio in the resulting in-phase and water images not only directly through the noise propagation in the water-fat separation, but also indirectly through the turbo spin-echo sequence. In this work, an optimization of the signal-to-noise ratio is proposed under the constraint of a fixed spacing between successive refocusing pulses. On the example of dual-echo Dixon imaging, it is shown to lead to shorter optimal echo shifts and to suggest a variable readout gradient strength, bandwidth and sampling window length per echo shift, which permits enhancing the signal-to-noise ratio.

Riad Ababneh¹, Thomas Benkert², and Felix Breuer^2
1Physics Department, Yarmouk University, Irbid, Jordan, ²Research Center Magnetic Resonance Bavaria, Würzburg, Bavaria, Germany

We used a stack-of-stars 3D radial acquisition with non-uniform fast Fourier transform (NUFFT) gridding to separate fat and water signals of different respiratory phases in free-breathing.

Dinghui Wang¹, Nicholas R. Zwart¹, Zhiqiang Li¹, Michael Schär^1,2, and James G. Pipe^1
1Neuroimaging Research, Barrow Neurological Institute, Phoenix, AZ, United States, ²Philips Healthcare, Cleveland, OH, United States

Spiral water-fat imaging suffers from blurring caused by B0 inhomogeneity and chemical shift of fat. Long spiral readout is efficient but often results in blurred and erroneous B0 field map at the interface of water and fat, and/or in regions of rapidly varying B0. We propose two approaches to iteratively correct and refine the initial field map obtained from Dixon water-fat imaging. Both methods employ a previously presented approach that simultaneously separates and deblurs water and fat. In vivo experiment results have demonstrated the feasibility of both approaches.

Nicholas Ryan Zwart¹, Dinghui Wang¹, and James Grant Pipe^1
1Neuroimaging Research, Barrow Neurological Institute, Phoenix, Arizona, United States

Blurring in spiral images can be caused by both field inhomogeneity and chemical shift. The method presented in this work addresses both by simultaneously separating fat-water signal and deblurring. A multi-peak fat model has been added to the previously presented algorithm, making the algorithm more robust while maintaining the same reconstruction time as the single-peak model.

Dinghui Wang¹, Zhiqiang Li¹, Ryan K. Robison¹, Nicholas R. Zwart¹, Michael Schär^1,2, and James G. Pipe^1
1Neuroimaging Research, Barrow Neurological Institute, Phoenix, AZ, United States, ²Philips Healthcare, Cleveland, OH, United States

Spiral in-out is an efficient sampling scheme, especially for spin echo and T2* weighted sequences. Two images can be reconstructed from the spiral-in and spiral-out parts respectively. Based on a previous conjugate gradient method, we propose an iterative approach to simultaneously separate and deblur water and fat using these two images with known B0 inhomogeneity. Ringing artifacts can arise at sharp boudaries since different frequences at the k-space converge at various speeds. In-vivo data have confirmed that the ringing artifacts can be eliminated by the assumption that water and fat are in phase at time 0.

Dinghui Wang¹ and James G. Pipe^1
1Neuroimaging Research, Barrow Neurological Institute, Phoenix, AZ, United States

We propose an analytical approach of three-point water-fat Dixon imaging with a known multi-peak fat model. This approach works with flexible even and uneven TE increments. By selecting appropriate TE points, spatial phase unwrapping can be avoided. This method solves the quadratic equations of water and fat. If the TE points are selected such that the magnitude of fat fluctuates significantly, a substantial portion of pure water and pure fat voxels can be identified according to the asymmetry between water and fat. The efficiency and robustness of the subsequent post-processing will thus be enhanced.

Jingfei Ma¹, Jong Bum Son¹, and John Hazle^1
1Imaging Physics, The University of Texas MD Anderson Cancer Center, Houston, TX, United States

We present a parallelizable region growing algorithm for phase correction in MRI. An image is divided into sub-images in which region growing is independently initiated from seed pixels and temporarily halted at a pre-defined quality threshold. Results from different sub-images are compared and then merged to form consistent pixel islands. Finally, region growing is resumed from the pixel islands by gradually relaxing the quality threshold used for controlling the region growing process and merging the different pixel islands whenever possible. The algorithm is demonstrated for generating in vivo two-point Dixon images.

Brady Quist^1,2, Evan G. Levine^1,2, Bruce L. Daniel¹, Brian A. Hargreaves¹, and Manojkumar Saranathan^1
1Department of Radiology, Stanford University, Stanford, California, United States, ²Department of Electrical Engineering, Stanford University, Stanford, California, United States

The 2-point Dixon and Homodyne method each rely on certain assumptions of the underlying phase in the image to work correctly. Here we analyze a published method of combining both reconstruction techniques into one, effectively performing fat/water separation while regaining lost resolution from a partial k-space acquisition. New echo times with improved phase characteristics are proposed which are shown to reduce the artifact associated with the method. The Homodyne Dixon method along with the new echo time choice can enable improved temporal or spatial resolution in time-sensitive scans such as DCE or breath hold imaging.

Sjoerd Crijns¹, Bjorn Stemkens¹, Alessandro Sbrizzi^1,2, Jan Lagendijk¹, Peter Luijten², Nico van den Berg¹, and Anna Andreychenko^1
1Radiotherapy, UMC Utrecht, Utrecht, Utrecht, Netherlands, ²Radiology, UMC Utrecht, Utrecht, Utrecht, Netherlands

Dixon water/fat separation sequences usually need at least two images acquired at different TEs constrained to specific values, causing prolonged repetition and acquisition times and complicating application in dynamic imaging (e.g. DCE-MRI). We propose a single point water/fat separation method based on alternation of two RF pulses that introduces a spatial shift of fat with respect to water in the reconstructed image. We demonstrate the feasibility of this method and obtain in-phase water/fat separated images at an arbitrary echo time in a single acquisition.

Lin Chen¹, Congbo Cai¹, Shuhui Cai¹, Jing Li¹, Miao Zhang¹, Ting Zhang¹, and Zhong Chen^1
1Department of Electronic Science, Xiamen University, Xiamen, Fujian, China

Separation of fat and water signals in MRI is very important for many clinical applications. Usual methods, such as Dixon and IDEAL, will lead to a long scan time. Spatiotemporally encoded (SPEN) single-shot MRI is an alternative method to separate fat and water in subsecond. For SPEN approach, super-resolved reconstruction is indispensable. However, existing algorithms will result in artifacts. In this abstract, compressed sensing is applied to the reconstruction to reduce the artifacts and improve the image quality. This reconstruction algorithm would benefit the application of SPEN single-shot MRI to fat-water separation.

Ying Chen¹, Congbo Cai¹, Jianhui Zhong², and Zhong Chen^1
1Department of Electronics Science,Xiamen University, Xiamen, Fujian, China, ²Department of Imaging Sciences, University of Rochester, Rochester, New York, United States

This abstract proposed a high-resolution water fat separation algorithm using single spatiotemporally encoded echo. The overall field inhomogeneity in each spatiotemporally encoded signal line was evaluated; then spatial smoothness regularization was imposed to the signal equation set to estimate water and fat profiles in these lines; by evaluating the phase linearity of the output, the regions where intermingling may exist can be identified and were constrained with adaptive filtering regularization in the second estimation for better output. Experiment results show this technique can deliver more efficient water fat separation in many cases that is challenging or intractable for conventional methods.

Su-Chin Chiu^1,2, Hsiao-Wen Chung², Martin Buechert¹, and Michael Bock^1
1Radiology - Medical Physics, University Medical Center Freiburg, Freiburg, Germany, ²Graduate Institute of Biomedical Electronics and Bioinformatics, National Taiwan University, Taipei, Taiwan

The MR fingerprinting (MRF) has recently been introduced to generate various quantitative parameter maps in a single image acquisition. In this study we extend MRF to apply it to fat/water separation. In addition to acquiring parameter maps of T1 and T2, fat and water fraction maps were created with a dedicated MRF pulse sequence and an orthogonal matching pursuit algorithm with 3 iterations. In a phantom experiment the fat/water separations shows a good agreement with data from an MR spectroscopic method.

Julien Picaud^1,2, Guylaine Collewet¹, Giulio Gambarota^3,4, and Jerome Idier^2
1UR TERE, IRSTEA, Rennes, France, ²IRCCyN, CNRS, Nantes, France, ³UMR 1099, INSERM, Rennes, France, ⁴LTSI, Universite de Rennes 1, Rennes, France

The purpose of this study was to compare fat quantification on fish using MRI water/fat separation with localized spectroscopy (MRS) and with chemical analysis. 36 samples inside a wild salmon were used for MRI vs. MRS and 15 salmon cutlets for MRI vs. chemical analysis. MRI images and MRS spectra were acquired at 1.5T. We used GRE with 6 echoes and the VarPro reconstruction method proposed by Hernando (MRM 2010, 63-1). High correlations were found in both cases. Very good agreement was found between MRI and MRS while and overestimation of fat content was observed for MRI versus chemical analysis.

Diego Hernando¹, Samir D. Sharma¹, Harald Kramer^1,2, and Scott B. Reeder^1,3
1Radiology, University of Wisconsin-Madison, Madison, WI, United States, ²Ludwig-Maximilians-University Hospital Munich, Munich, Germany, ³Medicine, University of Wisconsin-Madison, Madison, WI, United States

The proton resonance frequency (PRF) of water depends on temperature, whereas the PRF of triglycerides is temperature independent (aside from bulk susceptibility effects). This leads to a temperature dependence of the frequency shift between fat and water resonances, which may introduce errors in chemical shift-encoded (CSE) fat quantification methods that assume a known relative shift between the PRF of water and fat. In this work, we characterize the confounding effect of temperature on CSE fat quantification. Further, we demonstrate that a temperature-corrected spectral model of fat can be used to avoid these errors.

Axel Joachim Krafft¹, Ralf B. Loeffler¹, Xiao Bian^1,2, Ruitian Song¹, Beth M. McCarville¹, Jane S. Hankins³, and Claudia M. Hillenbrand^1
1Radiological Sciences, St. Jude Children's Research Hospital, Memphis, TN, United States, ²Rhodes College, Memphis, TN, United States, ³Hematology, St. Jude Children's Research Hospital, Memphis, TN, United States

Fat is a confounder in R2* based liver iron assessment as fat-water modulations in multi gradient echo imaging eventually affect the T2* evaluation. A simple solution could be the inclusion of fat suppression (FS) via chemically selective saturation (CHESS). However, decreasing T2* times as observed with increasing liver iron correspond to spectral profiles which overlap with the CHESS frequency band. This partial saturation potentially alters R2* leading to a biased iron evaluation. Here, the effect of CHESS FS on R2* is measured in 65 transfusional iron overload patients. A phenomenological model is presented to explain and correct the observed changes.

Mario A. Bacher^1,2, Xiaodong Zhong³, Brian M. Dale⁴, Marcel D. Nickel², Berthold Kiefer², Mustafa R. Bashir⁵, Rudolf Stollberger¹, and Stephan A.R. Kannengiesser^2
1Institute of Medical Engineering, Technical University Graz, Graz, Austria, ²MR Applications Development, Siemens AG, Healthcare Sector, Erlangen, Germany, ³MR R&D Collaborations, Siemens Healthcare, Atlanta, GA, United States, ⁴MR R&D Collaborations, Siemens Healthcare, Cary, NC, United States, ⁵Department of Radiology, Duke University Medical Center, Durham, NC, United States

Multi gradient echo quantification of liver proton density fat fraction (PDFF) and transverse relaxation rate (R2*) was analyzed using model consistency metrics. Different acquisition protocols and fat spectral models as reconstruction parameters were compared. Data from twenty healthy volunteers showed no significant differences in PDFF for different protocols, but significant differences in PDFF and R2* for different protocols and spectral models. With respect to fitting error, the protocols were significantly different, whereas the spectral models were not significantly different from each other. Model consistency analysis is a useful tool for evaluating multi gradient echo imaging with advanced quantification

Abraam S Soliman^1,2, Curtis Wiens³, Trevor Wade^2,4, Ann Shimakawa⁵, Terry M Peters^1,2, and Charles A McKenzie^1,4
1Biomedical Engineering, Western University, London, Ontario, Canada, ²Imaging Research Laboratories, Robarts Research Institute, London, Ontario, Canada, ³Radiology, University of Wisconsin, Madison, Wisconsin, United States, ⁴Medical Biophysics, Western University, London, Ontario, Canada, ⁵Global MR Applied Science Laboratory, GE Healthcare, Menlo Park, California, United States

Chemical-shift based multi gradient echo sequences have been widely used for water/fat separation. Typically, 6 unipolar readout gradients are applied over multiple shots in order to achieve optimal echo-spacing. Although single shot bipolar readout acquisition can offer optimal echo-spacing with shorter scan time, phase errors can significantly corrupt water/fat separation. To overcome this problem, a new interleaved bipolar acquisition is proposed. Accurate fat quantification is demonstrated in phantom and in-vivo experiments compared to the well-established unipolar sequence. The proposed bipolar acquisition scheme offers accurate fat fraction maps with shorter acquisitions and higher SNR efficiency compared to unipolar sequences.

Alexander Brunner¹, Daniela Strzoda², Karel D. Klika³, Mathies Breithaupt¹, Vanessa Stahl¹, Stephan Herzig², and Armin M. Nagel^1
1Medical Physics in Radiology, German Cancer Research Center (DKFZ), Heidelberg, Germany, ²Molecular Metabolic Control, German Cancer Research Center (DKFZ), Heidelberg, Germany, ³Molecular Structure Analysis, German Cancer Research Center (DKFZ), Heidelberg, Germany

Based on its unique thermogenic capacity, brown adipose tissue (BAT) shows very high potential to serve as a therapeutic node in the treatment of metabolic disorders, e.g., obesity. Recently, the BOLD effect T2*-weighted MRI during cold stimulation was used to detect cold-activated BAT in human individuals [1]. In this work, we compared 14 T NMR spectra and, for the first time, water fat fraction (WFF) [2], T1, T2 values measured in vitro by 7 T MRI between BAT, inguinal white adipose tissue (iWAT) and abdominal white adipose tissue (aWAT) in cold-stimulated mice and normal mice.

Theresa Bachschmidt^1,2, Peter Jakob², Markus Vester¹, Jürgen Nistler¹, and Mathias Nittka^1
1Siemens Healthcare, Erlangen, Germany, ²Experimental Physics 5, University of Würzburg, Würzburg, Germany

Susceptibility-induced artifacts in MR imaging of metal hip implants can be addressed by methods like SEMAC. Hence, B1 effects like shading and banding become more prominent at 3T. This work systematically analyzes B1 modulations and investigates a new approach to reduce those by means of B1 polarization, using state-of-the-art multi-channel transmit MRI systems. An analytical model is verified numerically and in phantom measurements. It is also used to predict and optimize signal intensity patterns for spin-echo based sequences and verified in vivo. This model helps to avoid signal loss or sharp transitions between hypo- and hyper-intense signal close to hip implants.

Bragi Sveinsson¹, Valentina Taviani¹, Kevin Koch², Garry Gold¹, and Brian Hargreaves^1
1Radiology, Stanford University, Stanford, CA, United States, ²Applied Sciences Laboratory, General Electric, Waukesha, WI, United States

MAVRIC-SL is routinely used for clinical imaging close to metallic orthopedic devices. MAVRIC-SL collects multiple off-resonant 3D volumes acquired under a constant slab selection gradient. The collection of multiple 3D volumes presents challenges in maintaining short scan times. Previous work has demonstrated how hexagonal sampling can substantially reduce scan time when employing 2×1 parallel imaging. In this work, we demonstrate a variation of this method that allows hexagonal sampling in conjunction with 3×2 parallel imaging for more rapid imaging.

Kevin Koch¹, Adriana Kanwischer¹, and Robert Peters^1
1GE Healthcare, Milwaukee, WI, United States

An artifact quantification study is presented on metal-on-metal and metal-on-poly total hip replacements at 1.5T and 3T. Computational models and known theoretical performance limitations are used to quantify artifact volumes in 2D-FSE and 3D-MSI imaging approaches. Analysis of these results demonstrates the expected performance differences across different pulse sequences, joint replacement constructions, and field strengths. Simulated images are presented using an anatomic hip model to demonstrate practical clinical impact of the predicted artifacts.

Matthew R. Smith¹, Nathan S. Artz¹, and Scott B. Reeder^1,2
1Radiology, University of Wisconsin, Madison, Wisconsin, United States, ²Biomedical Engineering, University of Wisconsin, Madison, WI, United States

Metallic implants induce extremely large B0 field perturbations that cause severe signal distortion. The purpose of this work is to examine the theoretically induced field map perturbation using the digital representation of commercially available metallic joint prostheses. Simulations presented here demonstrate that both RF excitation and frequency encoding is highly problematic for these implants using current 3D-MSI methods at both field strengths. Fully phase encoded methods may help with the frequency-encoding distortion but not with RF excitation limitations.

Kevin Koch¹ and Graeme McKinnon^1
1GE Healthcare, Milwaukee, WI, United States

It has been established that all frequency encoded imaging acquisitions have encoding limitations that are reached near commonly encountered implants at 3T. Here, we present a study whereby the known limitations on clinical MR scanners are modeled in the context of broadband single-point imaging. The resolution, scan time, and T2* signal loss of such a clinically viable single-point-imaging implementation are discussed. It is shown that such an implementation may provide signal directly near metal implants that cannot be acquired via any conventional readout-driven sequence

Jetse S. van Gorp¹, Chris J.G. Bakker^1,2, Frank Zijlstra¹, Jouke Smink³, Job G. Bouwman¹, and Peter R. Seevinck^1
1Image Sciences Institute, University Medical Center Utrecht, Utrecht, Netherlands, ²Department of Radiology, University Medical Center Utrecht, Utrecht, Netherlands,³Philips Healthcare, Best, Netherlands

In this work 3D imaging was performed in the presence of a titanium hip implant using conventional frequency encoded and purely phase encoded spin-echo sequences. Phase encoded images were shown to avert geometric distortion and to improve image quality in the proximity of the implant compared to conventional images. The total acquisition time could be successfully decreased a factor six by undersampling k-space in all three phase encoded dimensions followed by compressed sensing reconstruction.

Jetse S. van Gorp¹, Chris J.G. Bakker¹, and Peter R. Seevinck^1
1Image Sciences Institute, University Medical Center Utrecht, Utrecht, Netherlands

In this work the feasibility of a new hybrid multi-spectral method is investigated using two orthopedic implants. The method combines on-resonance frequency encoded with off-resonance reduced field of view fully phase encoded images, to exploit the efficiency of frequency encoding and the geometrical accuracy of phase encoding. Hybrid images were compared to MAVRIC-type images, showing that the hybrid method was able to improve the image quality adjacent to metal implants. Large excitation bandwidths and a reduced field-of-view could be used for the phase encoded images, decreasing the number of necessary off-resonance acquisitions and reducing the acquisition time compared to solely phase encoded imaging.

Toru Yamamoto¹, Yanhui Gao², and Kazuhiro Muramatsu^2
1Faculty of Health Sciences, Hokkaido University, Sapporo, Hokkaido, Japan, ²Department of Electrical and Electronic Engineering, Saga University, Saga, Japan

MR artifacts caused by implants are hindering proper diagnosis. Lowering the susceptibility of the material has been being developed, but even titanium is not enough for several diagnoses. To reduce these metal artifacts to a small area to enable diagnosis, we propose a new structural design of implants with hybrid of paramagnetic outer shell and diamagnetic inner core. The hybrid structure of paramagnetic and diamagnetic materials reduces the metal artifact of the implant dramatically. This structural design would be applied to various orthopedic implants.

JaeJin Cho¹, Yeji Han¹, and HyunWook Park^1
1Korea Advanced Institute of Science and Technology, Daejeon, Korea

This paper proposes the method that reduces the artifact caused by the electrode used in deep brain stimulation or electroencephalogram

Florian Dittmann¹, Sebastian Hirsch¹, Jing Guo¹, Jürgen Braun², and Ingolf Sack^1
1Institute of Radiology, Charité - Universitätsmedizin Berlin, Berlin, Germany, ²Institute of Medical Informatics, Charité - Universitätsmedizin Berlin, Berlin, Germany

We propose a single shot MRE sequence synchronized to continuous mechanical vibrations by adaptive measurement block shifting. As a result, measurement time is significantly reduced compared to former single shot MRE sequences without compromising SNR and data quality. The new sequence is demonstrated for high resolution elastography of the brain by applying 15 frequencies in the range from 25 to 60 Hz and multifrequency wave field inversion. For 7 slices, 8 wave dynamics and three field components data acquisition for 15 frequencies was accomplished within 9:40 min which is to our knowledge the fastest MRE sequence currently available.

Dieter Klatt¹, Curtis L. Johnson², Temel K. Yasar³, Joseph L. Holtrop^2,4, Bradley P. Sutton^2,4, Thomas J. Royston¹, and Richard L. Magin^1
1The Richard and Loan Hill Department of Bioengineering, The University of Illinois at Chicago, Chicago, Illinois, United States, ²Beckman Institute, University of Illinois at Urbana-Champaign, Urbana, Illinois, United States, ³Mechanical and Industrial Engineering, The University of Illinois at Chicago, Chicago, Illinois, United States,⁴Bioengineering, University of Illinois at Urbana-Champaign, Urbana, Illinois, United States

SampLe Interval Modulation Magnetic Resonance Elastography (SLIM-MRE) enables the simultaneous acquisition of the 3D displacement vector through direction-dependent shifting of motion encoding gradients relative to the applied vibration. In this study, we have embedded a SLIM-MRE acquisition scheme into a multishot, variable-density spiral imaging sequence and have determined the cerebral mechanical properties of three volunteers. Our findings suggest that SLIM reduces MRE scan time and allows immediate co-registration of the three displacement components without compromising inversion results.

Tomokazu Numano¹, Kazuyuki Mizuhara², Yoshihiko Kawabata³, Toshikatsu Washio⁴, and Kazuhiro Homma^4
1Radiological Sciences, Tokyo Metropolitan University, Arakawa, Tokyo, Japan, ²Mechanical Engineering, TOKYO DENKI UNIVERSITY, Adachi, Tokyo, Japan, ³Takashima Seisakusho Co.,Ltd, Hino, Tokyo, Japan, ⁴National Institute of Advanced Industrial Science and Technology, Tsukuba, Ibaraki, Japan

In this work a new MR Elastography (MRE) technique which can be performed on conventional MRI was developed. A wireless TR synchronization system consists of high-frequency radio receiver and a plain dipole antenna tuned to the RF excitation frequency was developed. The leak RF signal received via the dipole antenna, in the magnet room was available as the TR synchronization trigger then any electrical wiring from the MRI electronics is required. The fusion of the simple MRE sequence and the wireless synchronization pneumatic vibration system make it possible to construct the MRE system in any conventional MRI system.

Aaron T Anderson¹, Curtis L Johnson², Joseph L Holtrop^2,3, Elijah EW Van Houten^4,5, Matthew DJ McGarry⁵, Keith D Paulsen^5,6, Bradley P Sutton^2,3, and John G Georgiadis^1,2
1Mechanical Science & Engineering, University of Illinois at Urbana-Champaign, Urbana, IL, United States, ²Beckman Institute for Advanced Science, University of Illinois at Urbana-Champaign, Urbana, IL, United States, ³Bioengineering, University of Illinois at Urbana-Champaign, Urbana, IL, United States, ⁴Département de Génie Mécanique, Université de Sherbrooke, Sherbrooke, QC, Canada, ⁵Thayer School of Engineering, Dartmouth College, Hanover, NH, United States, ⁶Dartmouth-Hitchcock Medical Center, Lebanon, NH, United States

Magnetic resonance elastography (MRE) has seen many advances in shear wave drivers, imaging techniques, and material property reconstruction but continues to have issue with specificity of properties within anisotropic microstructures due to the isotropic assumption. Sections within the brain with highly ordered structure behave very differently depending on the direction of applied shear, longitudinal compared to transverse. Adding multiple shaking directions, within the existing isotropic framework, shows promise of increasing the fidelity of all reconstructed material properties and throughout the brain. The increase in fidelity will help improve diagnosis of diseases affecting the microstructure of the brain.

Selcan Ipek-Ugay¹, Michael Ledwig², Toni Drießle², Jing Guo³, Ingolf Sack³, and Jürgen Braun^4
1Radiology, Charité-Universitätsmedizin Berlin, Berlin, Berlin, Germany, ²Pure Devices GmbH, Würzburg, Germany, ³Radiology, Charité-Universitätsmedizin Berlin, Berlin, Germany, ⁴Medical Informatics, Charité-Universitätsmedizin Berlin, Berlin, Germany

A tabletop magnetic resonance elastography (MRE) system was developed based on a 0.5T permanent magnet for the analysis of viscoelastic properties of tissue samples. The system allowed us the measurement of viscoelastic parameters in soft tissue samples in a frequency range of 500-1000 Hz. The data are in good agreement to published data acquired in a 7T-highfield superconducting magnet. Acquiring the same viscoelasticity information by low costs with little requirements for space and maintenance might support MRE developments towards mechanics based histopathology capable to links viscoelastic constants with the etiology and pathogenesis of diseases.

Anna-Lisa Kofahl¹, Jakob Bindl¹, Deniz Ulucay¹, Sebastian Theilenberg¹, Judith Wild¹, Sylvia Napiletzki¹, Alexandra Vohlen¹, Jürgen Finsterbusch², Bernd Weber³, Carsten Urbach¹, and Karl Maier^1
1HISKP, University of Bonn, Bonn, Germany, ²University Medical Center Hamburg-Eppendorf, Hamburg, Germany, ³Life & Brain GmbH, Bonn, Germany

The knowledge of the viscoelastic properties of the human brain tissue may aid in the diagnosis of diseases like Alzheimer’s disease, brain cancer or multiple sclerosis. A novel method to image the viscoelastic properties of the brain in vivo and non-invasively with a good spatial resolution is Magnetic Resonance Rheology (MR-R). MR-R uses an acceleration and creep experiment inside an MRI, where a motion encoding EPI sequence is used to measure the relaxation movement of the substance under investigation. To prove the feasibility of this novel method and to estimate its potential agar-phantoms with and without inclusions are investigated.

Sebastian Theilenberg¹, Jakob Bindl¹, Anna-Lisa Kofahl¹, Deniz Ulucay¹, Judith Wild¹, Alexandra Vohlen¹, Sylvia Napiletzki¹, Jürgen Finsterbusch², Bernd Weber³, Carsten Urbach¹, and Karl Maier^1
1HISKP, University of Bonn, Bonn, Germany, ²University Medical Center Hamburg-Eppendorf, Hamburg, Germany, ³Life & Brain GmbH, Bonn, Germany

Magnetic Resonance Rheology is a novel method to image the viscoelastic properties of tissue in vivo using MRI phase images. By introducing a free fall over a small height the steady state of the brain tissue is disturbed, causing the tissue to move relative to the cranial bone. The exact trajectory of this motion is dependent on the local viscoelastic properties. Measuring these using a motion sensitive single-shot EPI sequence synchronized to the free fall creates a phase contrast. By varying the point in time of measurement the whole trajectory can be investigated.

Anthony Romano¹, Varsha Viswanath², Jing Guo³, Michael Scheel³, Sebastian Hirsch³, Jürgen Braun⁴, and Ingolf Sack^3
1Physical Acoustics, The Naval Research Laboratory, Washington, DC, United States, ²Department of Biomedical Engineering, University of California at Davis, Davis, CA, United States, ³Department of Radiology, Charité-Universitätsmedizin, Berlin, Germany, ⁴Institute of Medical Informatics, Charité-Universitätsmedizin, Berlin, Germany

Previously, we introduced a method called Waveguide Elastography and implemented this to analyze the anisotropic stiffnesses of the Corticospinal Tracts of both healthy volunteers and patients suffering from ALS. In these previous studies, we inverted the Orthotropic equations of motion along the local tangent vectors of the fiber tracts. Here, we invert along the principal direction of fiber tracts while including the effects of curvature. It was observed that the effects of curvature in the calculation of the Laplacians bias the “inherent” stiffness in a similar fashion as wave velocities are altered by an index of refraction in ray theory.

Elijah EW Van Houten¹, Curtis L Johnson², Aaron T Anderson², Joseph L Holtrop², Bradley P Sutton², John G Georgiadis², Matthew D McGarry³, John B Weaver³, and Keith D Paulsen^3
1Univ. de Sherbrooke, Sherbrooke, QC, Canada, ²Univ. Illinois at Urbana-Champaign, IL, United States, ³Dartmouth College, NH, United States

A multi-frequency magnetic resonance elastography image reconstruction method is presented with power-law frequency dependency for the elastic properties. The method is based on a nonlinear inversion framework and uses a generalized Rayleigh damping model for the soft tissue elastic energy absorption. Comparative reconstructions in a healthy brain, based on three frequencies of excitation, show improved fidelity and definition within the power-law multi-frequency image. Structures such as the ventricles and the falx cerebri are clearly defined and possess material property values in line with expectations.

Andreas Fehlner¹, Sebastian Papazoglou¹, Jing Guo¹, Kaspar-Josche Streitberger¹, Sebastian Hirsch¹, Jürgen Braun², and Ingolf Sack^1
1Department of Radiology, Charité - Universitätsmedizin Berlin, Berlin, Germany, ²Institute of Medical Informatics, Charité - Universitätsmedizin Berlin, Berlin, Germany

A new driver for brain MRE is introduced which avoids the direct application of vibrations to the head without compromising sufficiently high extrinsic wave amplitudes needed for MRE. Waves are introduced into the cranial cavity by a chest cradle mounted to a piezo-based actuator which was placed at the end of the patient table. The new driver is particularly suited for low drive frequencies (25–40 Hz) which are capable to resolve anatomical details and provide viscoelastic parameters in agreement to previous work. The new method is demonstrated in ten healthy volunteers for 8 driving frequencies and two 3D views.

Simon A. Lambert^1,2, Peter Nashölm³, Lauriane Juge¹, Lynne Bilston⁴, Bojan Guzina⁵, Sverre Holm³, and Ralph Sinkus^2
1INSERM U1149, 1- Center for research on inflammation, Université Paris 7, Clichy, Ile de france, France, ²Division of Imaging Sciences and Biomedical Engineering, KCL, BHF Centre of Excellence, london, London, United Kingdom, ³Informatics Department, University of Oslo, oslo, Norway, ⁴University of New South Wales, Neuroscience Research Australia, Randwick, Australia, ⁵University of Minnesota, Minneapolis, United States

Recently in vivo experiments using Multifrequency MRE (MMRE) have shown that the exponent of the power law derived from MMRE data fitting with a power law could be more sensitive to specific pathologies such as fibrosis, steatosis or even inflammation. However these works lack fundamental understanding of the relation existing between the tissue microstructure with its macroscopic nature. In this study we develop a full theoretical model of shear wave propagation at the microscopic scale in phantoms containing wave obstacles and demonstrate that shear waves are able to reveal at the macroscopic scale the hidden micro-architectural properties of the material.

Roger C Grimm¹, Joshua D Trzasko¹, Armando Manduca¹, and Richard L Ehman^1
1Mayo Clinic, Rochester, MN, United States

Clinical 2D GRE Elastography is acquired and processed as 2D data sets. Commonly, four slices of stiffness estimates can be obtained in four breath-holds. A true 3D wave field sample with 3D processing would provide a more accurate estimate of the tissue stiffness. The proposed 3D GRE sequence can provide similar slice throughput with four to six 3D processed images obtained in four breath-holds. Due to the nature of the acquisition, the GRE scans provide fewer artifacts compared to similar EPI scans. The CLEAR reconstruction algorithm is used to provide superior ghost reduction compared to ASSET while providing additional acceleration.

Jing Guo¹, Sebastian Hirsch¹, Sebastian Papazoglou¹, Kaspar-Josche Streitberger¹, Andreas Fehlner¹, Juergen Braun², and Ingolf Sack^1
1Department of Radiology, Charite - Universitaetsmedizin Berlin, Berlin, Berlin, Germany, ²Department of Medical Informatics, Charite - Universitaetsmedizin Berlin, Berlin, Germany

MRE can differentiate small lesions based on their stiffness or softness compared to surrounding healthy tissue. However, in tissue with fluid filled cavities the obtained elastograms are biased due to effects of the lesion's geometry to the refracted wave field, resulting an overestimation of stiffness values. We analyze this effect by numerical simulations and MRE data of phantoms and in vivo cysts in the abdomen and brain. Our results indicate severe overestimation of viscoelasticity in cysts unlike other fluid filled parts of the tissue. We show that this effect is a result of resonances in oscillating cavities with regular interfaces.

Marion Tardieu¹, Marie Poirier-Quinot¹, Ralph Sinkus², Luc Darrasse¹, and Xavier Maître^1
1IR4M (UMR8081), CNRS, Univ Paris-Sud, Orsay, France, ²Imaging Sciences & Biomedical Engineering Division, King's College, London, United Kingdom

MR-elastography aims at characterizing the mechanical properties of living tissues by probing wave propagation therein. Displacement fields are recorded over a mechanical cycle by encoding the inferred motion along the three spatial directions. Thus the complex shear viscoelastic moduli can be computed after inversion of the wave equation. Patients' motion during the MR-acquisition usually results in unrestrained spatial transformations of the targeted organ. It may also yield unwanted mismatch of the components of the acquired displacement fields. Spatial normalization of the phase image along the magnitude image tackles the correcting linear or non-linear transformations but, as numerically showed recently, displacement field normalization is required to fully recover the phase information in MR-elastography and improve the parametric reconstruction. Here, we experimentally validate the approach by applying these corrections on a breast phantom after MR-elastography exams for arbitrary three-dimensional rotations. This double normalization scheme was advantageously applied on a brain MR-elastography data set where the subject had involuntary moved during the acquisition.

Ingolf Sack¹, Barbara Steiner², Charlotte Klein², Elisabeth Hain², Kerstin Riek³, Jing Guo³, and Jürgen Braun^4
1Radiology, Charité-Universitätsmedizin Berlin, Berlin, Berlin, Germany, ²Neurology, Charité-Universitätsmedizin Berlin, Berlin, Germany, ³Radiology, Charité-Universitätsmedizin Berlin, Berlin, Germany, ⁴Medical Informatics, Charité-Universitätsmedizin Berlin, Berlin, Germany

Magnetic resonance elastography (MRE) was applied to a murine model of neurotoxin-induced Parkinson's disease and compared to histological findings. For the first time, a significant increase of cerebral elasticity was observed in response to disease. We found an increase in the storage modulus of the brain at day 6 after treatment with the neurotoxin MPTP which parallels the proliferation of new neurons predominantly in the hippocampal region. The results provide insight into the crucial role of neurons for the constitution of the viscoelastic matrix and motivate further applications of MRE in patients as a new biomarker for neurodegenerative diseases.