Zhitao Li¹, Benjamin Paul Berman², Jean-Philippe Galons³, Ali Bilgin^4,5, Maria I. Altbach³, and Diego R. Martin^3
1Electrical and Computer Engineering, The University of Arizona, Tucson, AZ, United States, ²Program in Applied Mathematics, the University of Arizona, Tucson, AZ, United States, ³Department of Medical Imaging, the University of Arizona, Tucson, AZ, United States, ⁴Electrical and Computer Engineering, the University of Arizona, Tucson, AZ, United States, ⁵Biomedical Engineering, the University of Arizona, Tucson, AZ, United States

A golden angle radial steady-state free-precession technique and a principle component based iterative algorithm are developed for the reconstruction of high resolution T1 maps from highly undersampled data. The total acquisition time is < 3 seconds per slice.

Eun Ji Lim^1,2 and Jaeseok Park^2
1Center for Neuroscience Imaging Research, Institute for Basic Science (IBS), Suwon, Korea, Republic of, ²Department of Biomedical Engineering, Sungkyunkwan University, Suwon, Korea, Republic of

Dynamic contrast-enhanced (DCE) 3D MRA has been widely used for diagnostic assessment of vascular morphology and hemodynamics in a clinical routine. It acquires a series of time-resolved images, revealing details on contrast dynamics. To extract angiograms while eliminating unwanted background tissues, subtraction between the reference (pre-contrast) and DCE images in each time frame is typically employed. However, in the presence of non-stationary background signal transition such as subject motion and time-varying magnetic field, subtraction results in incomplete background suppression and noise amplification. Due to the inherent, subtraction sparsity in either between the reference and each dynamic image or between neighboring time frames, compressed sensing (CS) is well suited to DCE MRA to enhance spatial and temporal resolution. Nevertheless, these approaches remain suboptimal due to the inherent limitation of subtraction. In this work, we propose a new, DCE MRA method called “concentration time-course Model-based Angiogram SEparation (MASE)”, in which DCE signals in the temporal direction are directly modeled and reconstructed with sparsity priors while background signals are attenuated.

Adrian Emmanuel Georg Huber¹, Christian Binter¹, Claudio Santelli¹, and Sebastian Kozerke^1
1Institute for Biomedical Engineering, University and ETH Zurich, Zurich, Switzerland

Entangled denoising (eSPARSE) was developed for highly undersampled 4D Flow MRI, combining both k-t SPARSE-SENSE and 3D-L1-Wavelets. Undersampling factors of 8 and 10 were studied in healthy volunteers. The proposed algorithm improves the reconstruction of fluid vector fields in the aorta compared to k-t SPARSE-SENSE. The relative RMSE velocity is reduced by up to 16% for an undersampling rate of R = 8 and by up to 19% for R = 10. Entangled denoising (eSPARSE) is a promising approach to accelerate 4D Flow MRI while preserving key features of the flow field.

Rebecca Ramb¹, Anthony G. Christodoulou^2,3, Irina Mader⁴, Maxim Zaitsev¹, Zhi-Pei Liang², and Jürgen Hennig^1
1University Medical Center Freiburg, Dept. of Radiology - Medical Physics, Freiburg, Germany, ²Beckman Institute, Dept. of Electrical and Computer Engineering, University of Illinois at Urbana-Champaign, Urbana, IL, United States, ³Heart Institute and Biomedical Imaging Research Institute, Cedars-Sinai Medical Center, Los Angeles, CA, United States, ⁴University Medical Center Freiburg, Dept. of Neuroradiology - Freiburg Brain Imaging, Freiburg, Germany

This work is to provide perfusion parameters such as cerebral blood volume at high spatial resolution, for detailed delineation of tumor borders and to guide stereotactic surgery biopsies in locating the most aggressive tumorous tissue. This is achieved with an k-t-undersampled EPI acquisition and PEAK-GRAPPA reconstruction. Additionally, an advanced iterative reconstruction method is developed incorporating partial separability constraint into parallel imaging with SPIRiT kernels (PS-SPIRiT). Two half-dose first-pass perfusion acquisitions in tumor patients allow direct comparison of the standard clinical protocol with the proposed acquisition and reconstruction schemes. 

Xiaozhi Cao¹, Congyu Liao¹, Zhixing Wang¹, Huihui Ye¹, Ying Chen¹, Hongjian He¹, Song Chen¹, Hui Liu², and Jianhui Zhong^1
1Center for Brain Imaging Science and Technology, Department of Biomedical Engineering, Zhejiang University, Hangzhou,Zhejiang, China, People's Republic of, ²MR Collaboration NE Asia, Siemens Healthcare, Shanghai, China, People's Republic of

The original MRF method uses one spiral interleaf acquired within each time point to reconstruct one image, leading to dramatic fluctuation in the signal evolution caused by undersampling error. In this work, a sliding window is utilized to select multiple interleaves to generate one full-sampled image for each step so that the fluctuation in signal evolution could be significantly alleviated. Therefore our proposed method can reach an expected result with reduced time points. The results demonstrate that the proposed method can reduce the acquisition time from about 11s to 5.6s or even less. 

Andrii Y Petrov¹, Michael Herbst^1,2, and V Andrew Stenger^1
1Department of Medicine, University of Hawaii, Honolulu, HI, United States, ²Department of Radiology and Medical Physics, University Medical Center Freiburg, Freiburg, Germany

Recent advances in dynamic MRI propose low-rank plus sparse (L+S) matrix decomposition for image reconstruction from reduced data acquisition. The L+S method has been successfully applied to multiple applications including cardiac MRI, perfusion, angiography and recently to denoise resting-state fMRI data, suggesting that it might be promising for improving temporal resolution in task-based fMRI. We propose to use 3D spiral acquisition, undersampled in k_z-t domain, and L+S method for image reconstruction. Our results indicate that proposed approach allows 4x acceleration for the data acquisition and improved statistical significance of activation maps in the S component from an increased temporal resolution and eliminating physiological noise in the L component.

Chen Zhongzhou¹, Zhang Xiaoyong¹, Zheng Hairong¹, Liu Xin¹, Fan Zhaoyang², and Xie Guoxi^1
1Lauterbur Research Center for Biomedical Imaging, Shenzhen Institutes of Advanced Technology,CAS, Shenzhen, China, People's Republic of, ²Biomedical Imaging Research Institute, Cedars-Sinai Medical Center, Los Angeles, CA, United States

Coronary vessel wall MR imaging has great potential to detect coronary plaques which can be important for heart disease diagnosis. However, the imaging is always time-consuming because it needs a large amount of data for clear vessel wall depiction. In this work, a novel method based on compressed sensing (CS) with block-weighted total variation (BWTV) was proposed to accelerate coronary vessel wall imaging. Simulation and in vivo experiment results demonstrated that the proposed method can significantly decrease the amount of data for image reconstruction without compromising the depiction of the tiny coronary vessel wall.  

Joseph Y. Cheng¹, Mariya Doneva², Tao Zhang¹, John M. Pauly³, Shreyas S. Vasanawala¹, and Michael Lustig^2
1Radiology, Stanford University, Stanford, CA, United States, ²Electrical Engineering & Computer Sciences, University of California, Berkeley, CA, United States, ³Electrical Engineering, Stanford University, Stanford, CA, United States

A major obstacle in MRI is artifacts from patient motion. This is especially challenging for pediatric imaging where anesthesia is often required to obtain diagnostic image quality from uncooperative patients. Thus, we developed a PROPELLER-based method to correct for non-rigid motion. Simple localized motion is estimated for each spatial region by first applying a spatial window to the image data. Low-rank minimization is used to iteratively estimate the motion without the need of determining a reference motion state and to potentially enable higher-resolution navigation. The final image is constructed via autofocusing. This approach is demonstrated in free-breathing abdominal scans of healthy and patient volunteers. 

Yu Y. Li^1
1Radiology, Cincinnati Children's Hospital Medical Center, Cincinnati, OH, United States

In MRI data acquisition, gradient encoding introduces a non-uniform distribution of tissue contrast and boundary information in k-space. As a result, data correlation increases with tissue boundary sparsity from the center to the outer k-space. The presented work investigates a new approach to accelerating MRI by taking advantage of non-uniform k-space data correlation. In this approach, k-space data are collected and reconstructed in a region-by-region fashion using a previously developed high-speed imaging framework, "correlation imaging"^1,2. It is demonstrated that region-by-region correlation imaging can introduce a gain over parallel imaging in imaging acceleration by utilizing more information.

Xiaobo Qu¹, Yunsong Liu¹, Zhifan Zhan¹, Jian-Feng Cai², Di Guo³, and Zhong Chen^1
1Department of Electronic Science, Xiamen University, Xiamen, China, People's Republic of, ²Department of Mathematics, Hong Kong University of Science and Technology, Hong Kong SAR, Hong Kong,³School of Computer and Information Engineering, Xiamen University of Technology, Xiamen, China, People's Republic of

Redundant sparse representations can significantly improve the MRI image reconstruction with sparsity constraint. An appropriate sparse model is very important to improve image quality even with the same sparsifying transforms and undersampled data. We propose a new fast, stable, compatible and simple iterative thresholding algorithm to solve the analysis sparse models that can obviously improve the image reconstruction for tight-frame-based sparsifying transform in compressed sensing MRI.  We theoretically prove the convergence of the proposed projected fast iterative soft-thresholding algorithm (pFISTA). Numerical results show that pFISTA achieves better reconstruction than state-of-art FISTA for synthesis sparse model and more stable and compatible than the state-of-art SFISTA. 

Matthan W.A. Caan^1,2, Abdallah G. Motaal¹, Bram F. Coolen¹, Wouter V. Potters¹, Kerry Zhang¹, Pieter Buur², and Aart J. Nederveen^1
1Radiology, Academic Medical Center, Amsterdam, Netherlands, ²Spinoza Centre for Neuroimaging, Amsterdam, Netherlands

We aimed to increase spatial resolution while maintaining scanning time for 3D-T1 weighted imaging by on-scanner undersampling at 7T. A 3DT1-weighted structural scan with a resolution of 0.5 mm³was acquired with 4 times variable poisson disc compressed sensing (CS) undersampling. For comparison, scans with resolutions of 0.5 and 0.7 mm³ were acquired, with SENSE undersampling adapted to match scanning time.The CS-reconstructed scan showed  no streaking artifacts, was less hampered by noise in deep brain regions, had better contrast between gray matter and CSF and less partial voluming effects.

Mark Chiew¹, Nadine N Graedel¹, and Karla L Miller^1
1FMRIB Centre, University of Oxford, Oxford, United Kingdom

In this work we propose a simple method for increasing the spatio-temporal incoherence of a golden angle radial time-series acquisition by mildly perturbing the golden angles with a random variable. Despite its quasi-random distribution of golden angle samples, x-f point spread function analysis reveals strong coherence along the frequency domain. When the golden angles are slightly jittered using a normal random variable with small variance, the x-f point spread functions take on a more diffuse, noise-like appearance, making the acquisition scheme more appropriate for k-t reconstruction methods relying on incoherence, while maintaining its favourable spatial properties. 

Eric G Stinson¹, Stephen J. Riederer¹, and Joshua D. Trzasko^1
1Radiology, Mayo Clinic, Rochester, MN, United States

Multi-level (uniform + non-uniform) sampling for accelerated dynamic MRI either solves a computationally expensive full regression problem, or breaks the problem into two separate steps for each sampling operator.  Here, the latter approach is taken, with low-rank matrix completion used as a pre-processing step to complete the non-uniform sampling operator before SENSE reconstruction unfolds the effects of the uniform sampling operator.  It is shown that both spatial and temporal resolution are retained with LRMC + SENSE in comparison with more traditional pre-processing steps.

R. Marc Lebel^1,2,3, Csanad G Varallyay⁴, and Edward A Neuwelt^4
1GE Healthcare, Calgary, AB, Canada, ²Radiology, University of Calgary, Calgary, AB, Canada, ³Biomedical Engineering, University of Calgary, Calgary, AB, Canada, ⁴Neurology, Oregon Health and Science University, Portland, OR, United States

Quantitative or semi-quantitative mapping of cerebral blood volume typically involves complex modeling of a dynamic gadolinium-enhanced acquisition. Off-label use of ferumoxytol is being explored as a mechanism for high-resolution quantification of cerebral blood volume. Acquisition involves high-resolution pre- and post-contrast T₂*-weighted scans; quantification is straightforward and does not require fitting. We present a multi-echo acquisition and optimal quantification algorithm for improved detection of ferumoxytol-based blood volume measurements. Our approach provides high dynamic range and minimal noise amplification.

Eun Ju Cha¹, Kyong Hwan Jin¹, Dong-Wook Lee¹, Eung Yeop Kim², Seung-Hong Choi³, and Jong Chul Ye^1
1KAIST, Daegeon, Korea, Republic of, ²Gacheon Unversity Gil Medical Center, Incheon, Korea, Republic of, ³Seoul National University Hospital, Seoul, Korea, Republic of

In dynamic contrast enhanced (DCE) MRI, temporal and spatial resolution can be improved by time-resolved angiography with interleaved stochastic trajectories (TWIST). However, due to view sharing, the temporal resolution of TWIST is not a true one. To overcome this limitation, we employ recently proposed annihilating filter-based low rank Hankel matrix approach (ALOHA) that interpolates the missing k-space data by performing low-rank matrix completion of weighted Hankel matrix. In vivo results showed considerably better temporal resolution than standard TWIST reconstruction.

Jinhong Huang^1,2, Biaoshui Liu¹, Gaohang Yu^1,2, Yanqiu Feng¹, and Wufan Chen^1
1School of Biomedical Engineering and Guangdong Provincial Key Laboratory of Medical Image Processing, South Medical University, Guangzhou, China, People's Republic of, ²School of Mathematics and Computer Science, Gannan Normal University, Ganzhou, China, People's Republic of

Conventional CS methods treat a 2D/3D image to be reconstructed as a vector. However, many data types do not lend themselves to vector data representation, and this vectorization based model may lose the inherent spatial structure property of original data and suffer from curse of dimensionality that occurs when working with high-dimensional data. In this work, we introduce a novel tensor dictionary learning method for dynamic MRI reconstruction. Numerical experiments on synthetic data and in vivo data show approximately 2 dB improvement in PSNR presented by the proposed scheme over existing method with overcomplete dictionary learning.

Lipeng Ning^1,2, Kawin Setsompop^2,3, and Yogesh Rathi^1,2
1Brigham and Women's Hospital, Boston, MA, United States, ²Harvard Medical School, Boston, MA, United States, ³Massachusetts General Hospital, Boston, MA, United States

We introduce a new method for rapid acquisition of sub-millimeter whole-brain diffusion imaging. Our method combines the gSlider-SMS acquisition method and the compressed-sensing super-resolution reconstruction algorithm. We demonstrate that this proposed approach is able to increase the resolution to 860μm iso in an effective acquisition time of 12 min.

Louis W Mann¹, David M Higgins², Carl N Peters¹, Sophie Cassidy¹, Ken K Hodson¹, Anna Coombs¹, Roy Taylor¹, and Kieren Grant Hollingsworth^1
1Institute of Cellular Medicine, Newcastle University, Newcastle upon Tyne, United Kingdom, ²Philips Healthcare, Guildford, United Kingdom

We compared hepatic fat fractions quantified with accelerated magnetic resonance (MR) imaging reconstructed with combined compressed sensing and parallel imaging (CS-PI) with conventional acquisitions. Undersampled data at ratios of 2.6×, 2.9×, 3.8×, and 4.8× were prospectively acquired from eleven subjects with type 2 diabetes and a healthy control. Fat fraction maps were calculated using CS-PI and Bland-Altman analysis. Inter- and intrarater analysis was performed. The fat fractions from the accelerated acquisitions had tight 95% limits of with no bias. The fat fractions were acceptable up to a factor of 3.8×, shortening the breath-hold from 17.7s to 4.7s.

Merry Mani¹, Mathews Jacob², Douglas Kelley³, and Vincent Magnotta^4
1Dept. of Psychiatry, University of Iowa, Iowa City, IA, United States, ²Dept. of Electrical and Computer Engineering, University of Iowa, Iowa City, IA, United States, ³General Electric Healthcare Technologies, San Francisco, CA, United States, ⁴Dept. of Radiology, University of Iowa, Iowa City, IA, United States

Multi-shot diffusion imaging holds great potential for enabling high spatial resolution diffusion imaging as well as short echo time imaging to enhance studies at higher field strengths. However, the imaging throughput of multi-shot diffusion scheme is low. To increase the efficiency, under-sampled multi-shot acquisitions can be employed. However the conventional multi-shot diffusion-weighted imaging reconstructions that rely on motion-induced phase estimates are not appropriate for such acquisitions since the phase estimates will be highly corrupted due to the under-sampling. Here we propose a new total-variation regularized reconstruction for under-sampled multi-shot diffusion data using an annihilating filter bank formulation in a weighted k-space domain.  

Li Zhang^1,2, Jennifer Barry², Mihaela Pop^1,2, and Graham A Wright^1,2
1Department of Medical Biophysics, University of Toronto, Toronto, ON, Canada, ²Sunnybrook Health Sciences Centre, Toronto, ON, Canada

Multi-contrast late enhancement (MCLE)¹ images offer better visualization of myocardial infarction (MI) than conventional IR-FGRE. However, current MR images either with IR-FGRE or MCLE provide an inferior spatial resolution of 1.6-2.0mm in-plane with a slice thickness of 5-8mm in the clinical setting. Characterization of infarct heterogeneity requires high spatial resolution. We propose a novel method to reconstruct MCLE images at a high spatial resolution from a highly accelerated dataset acquired prospectively with three-dimensional (3D) MCLE. The method was validated in a preclinical model, producing an isotropic resolution of 1.5mm within a single breath-hold.

Emilie Mussard^1,2,3, Tom Hilbert^1,2,3, Reto Meuli², Jean-Philippe Thiran^2,3, and Tobias Kober^1,2,3
1Advanced Clinical Imaging Technology (HC CMEA SUI DI BM PI), Siemens Healthcare AG, Lausanne, Switzerland, ²Department of Radiology, University Hospital (CHUV), Lausanne, Switzerland, ³LTS5, École Polytechnique Fédérale de Lausanne, Lausanne, Switzerland

MP2RAGE is a T1 imaging method providing greatly reduced B1 bias as well as less T2* and PD-contributions. It requires, however, long acquisition time (standard protocol with GRAPPAx3: ~8min) which hampers its clinical application. This work proposes to use sparse iterative reconstruction techniques to shorten MP2RAGE acquisition times. Resulting images are benchmarked using contrast assessment, changes in obtained T1 values as well as evaluating the effect of undersampling on an automated brain morphometry algorithm. Acceptable penalty in image quality and morphometric outcome was achieved with up to 5-fold acceleration, yielding a measurement time of 3.8min compared to fully sampled 20min.

Yasmin Blunck¹, Sonal Josan², Brad A. Moffat³, Shawna Farquharson⁴, Roger J. Ordidge³, and Leigh A. Johnston^1
1Electrical & Electronic Engineering, University of Melbourne, Melbourne, Australia, ²Siemens Healthcare, Melbourne, Australia, ³Anatomy & Neuroscience, University of Melbourne, Melbourne, Australia, ⁴Florey Institute of Neuroscience and Mental Health, Melbourne, Australia

Sodium plays a vital role in various physiological processes and functions due to the delicate balance between intra- and extracellular sodium concentration. Disturbance to this electrochemical gradient is considered to be a sensitive, early indicator of cell breakdown and provide an insight into cellular integrity.  Iterative reconstruction of sparsely undersampled data has been shown to cause inaccuracies in the estimation of tissue sodium content. This work investigates the bias in sodium concentration estimates and presents a method for correction that recovers the quantitative property of sodium imaging for high undersampling factors.

Thomas Loughran¹, David M Higgins², Anna Coombs¹, Michelle McCallum³, Volker Straub³, and Kieren Grant Hollingsworth^1
1Institute of Cellular Medicine, Newcastle University, Newcastle upon Tyne, United Kingdom, ²Philips Healthcare, Guildford, United Kingdom, ³Institute of Genetic Medicine, Newcastle University, Newcastle upon Tyne, United Kingdom

Fat fraction measurement in muscular dystrophy is important for therapy trials. Accelerated acquisition reconstructed by combined compressed sensing and parallel imaging (CS-PI) can improve patient compliance. Eight patients with Becker muscular dystrophy were recruited and prospectively undersampled data at ratios of 3.65×, 4.94×, and 6.42× were acquired in addition to fully sampled data. The CS-PI reconstructions were of sufficient quality at 3.65× and 4.94× acceleration. Compared to fully sampled data, non-significant bias and 95% limits of agreement of 1.65%, 1.95% and 2.22% were found for the three CS-PI reconstructions, significantly outperforming conventional parallel imaging alone.

Marcel Gratz^1,2, Marc Schlamann^3,4, Sophia Göricke⁴, Stefan Maderwald², and Harald H. Quick^1,2
1High Field and Hybrid MR Imaging, University Hospital Essen, Essen, Germany, ²Erwin L. Hahn Institute for Magnetic Resonance Imaging, University Duisburg-Essen, Essen, Germany, ³Neuroradiology, University Hospital Giessen and Marburg GmbH, Giessen, Germany, ⁴Department of Diagnostic and Interventional Radiology and Neuroradiology, University Hospital Essen, Essen, Germany

Highly undersampled contrast-enhanced intracranial MR angiography (SPARSE CE-MRA) was evaluated regarding the vascular visibility and image quality in a clinical setting on a 3T MRI system for 23 patients with various pathologies. The overall performance is comparable to TOF MR angiography, yet with much shorter acquisition times and a high-resolution whole-head coverage in both arterial and venous phase. However, a strong dependence of the obtained MRA quality on the bolus timing of the contrast agent was found and needs to be properly addressed by the operators to minimize the arterio-venous overlap in the imagery for best diagnostic quality.

Yawen Huang¹, Leandro Beltrachini¹, Ling Shao², and Alejandro Frangi^1
1Department of Electronic and Electrical Engineering, The University of Sheffield, Sheffield, United Kingdom, ² Department of Computer Science and Digital Technologies, Northumbria University, Newcastle, United Kingdom

Multi-modality MRI protocols are becoming standard in the everyday clinical practise. The advantages of such acquisitions were shown to be fundamental in a wide range of applications, such as medical diagnosis and image segmentation. However, the implementation of these protocols tends to be time-consuming, consisting in one key limitation. In this paper we address this problem by presenting a novel method for synthesising any MRI modality from a single acquired image. This is done using machine learning techniques for dictionary learning. Results show that our approach can lead to significant performance over the state-of-the-art methods.

Julia Velikina¹ and Alexey Samsonov^2
1Medical Physics, University of Wisconsin, Madison, WI, United States, ²Radiology, University of Wisconsin, Madison, WI, United States

We describe next generation of Model Consistency COndition (MOCCO) reconstruction which makes use of new Weighted low-rank models and attains high image quality through Iterative model adaptation to complexity of the local temporal dynamics (WI-MOCCO). 

Georg Spinner¹, Johannes Frieder Matthias Schmidt¹, and Sebastian Kozerke^1
1Institute for Biomedical Engineering, ETH Zurich, Zurich, Switzerland

In vivo Intravoxel Incoherent Motion (IVIM) parameter mapping in the brain is particularly challenging because of inherent noise amplification of parallel imaging. To address this limitation, correlations in space and in the b-value dimension may be jointly exploited. To this end, an iterative approach of k-t PCA was adapted to allow image reconstruction from undersampled IVIM data. Reconstruction and parameter estimation errors of the proposed k-b PCA approach relative to parallel imaging were assessed. Mean absolute parameter errors of k-b PCA were lower compared to CG-SENSE (R=4): 1.3±1.9·10^-4/1.8±2.0·10-4 mm²/s (D) 0.068±0.083/0.085±0.101 (f) and 6.4±16.9·10^-2/7.6±18.3·10^-2 mm²/s (D*). It is concluded that k-b PCA is a promising alternative to parallel imaging to reduce scan times while maintaining the quality of diffusion and perfusion parameter maps in brain exams.

Xiaozhi Cao¹, Congyu Liao¹, Zhixing Wang¹, Huihui Ye¹, Ying Chen¹, Hongjian He¹, Song Chen¹, Hui Liu², and Jianhui Zhong^1
1Center for Brain Imaging Science and Technology, Department of Biomedical Engineering, Zhejiang University, Hangzhou,Zhejiang, China, People's Republic of, ²MR Collaboration NE Asia, Siemens Healthcare, Shanghai, China, People's Republic of

MR fingerprinting has been used for estimating the intra-voxel tissue component fractions by resolving the equation S_voxel=Dw. In this study, potential fractions of interested components are used for building dictionary instead of T1 and T2, changing the solution of S_voxel =Dw into an optimization problem with regularization terms. The results demonstrate that the proposed method could provide more robust quantification of tissue composition and estimation of partial volume effects.

Brian C Allen¹, Felix Lugauer², Dominik Nickel², Lubna Bhatti¹, Randa A Dafalla¹, Brian M Dale³, Tracy A Jaffe¹, and Mustafa R Bashir^1,4
1Radiology, Duke University Medical Center, Durham, NC, United States, ²MR Application Predevelopment, Siemens Healthcare GmbH, Erlangen, Germany, ³MR R&D Collaborations, Siemens Healthcare, Cary, NC, United States, ⁴Center for Advanced Magnetic Resonance Development, Duke University Medical Center, Durham, NC, United States

Applying the described low-rank denoising algorithm to a liver fat/iron quantification technique reduces image noise in PDFF and R2* maps without adversely affecting mean values of the quantitative measures or reader assessment of edge sharpness.

Thomas Amthor¹, Peter Koken¹, Karsten Sommer¹, Mariya Doneva¹, and Peter Börnert^1
1Philips Research Europe, Hamburg, Germany

We propose an acceleration technique for MR Fingerprinting measurements based on Cartesian or otherwise segmented sampling. We show that quick repetition of MR Fingerprinting sequences leads to stationary signal responses. Instead of waiting for the spin system to relax completely before the start of the next fingerprint sequence, we take the evolution of the spin system into account and calculate a dictionary of steady-state fingerprints. We present a Cartesian fingerprint measurement with a SENSE factor of four. Using our method, additional acceleration by more than a factor of two could be achieved, without reducing the match accuracy.

Alessandro Sbrizzi¹, Annette van der Toorn¹, Hans Hoogduin¹, Peter R Luijten¹, and Cornelis AT van den Berg^1
1UMC Utrecht, Utrecht, Netherlands

Magnetic Resonance Spin TomogrAphy in Time-domain (MR-STAT) aims at reconstructing all physical quantities (T1,T2, PD, etc) directly from the data in the time-domain. By solving a large scale nonlinear inversion problem, the parameters can be inferred. The spatial encoding is entangled implicitly in the time response of the system. A very quick acquisition can be performed and processed to derive all system parameters, including electromagnetic field maps like B0 and B1.  We illustrate MR-STAT at high resolution for a realistic numerical phantom and we show high precision reconstructions of quantitative maps.

Congyu Liao¹, Xiaozhi Cao¹, Huihui Ye¹, Ying Chen¹, Hongjian He¹, Song Chen¹, Qiuping Ding¹, Hui Liu², and Jianhui Zhong^1
1Center for Brain Imaging Science and Technology, Zhejiang University, Hangzhou, China, People's Republic of, ²MR Collaboration NE Asia, Siemens Healthcare, Shanghai, China, People's Republic of

In this study, a low rank and sparsity based MRF reconstruction scheme (L&S MRF) is proposed for reducing the artifacts of each time point with a fraction of acquisition times.

Hye-Young Heo^1,2, Yi Zhang¹, Dong-Hoon Lee¹, Shanshan Jiang¹, Xuna Zhao¹, and Jinyuan Zhou^1,2
1The Russell H. Morgan Department of Radiology and Radiological Science, Johns Hopkins University School of Medicine, Baltimore, MD, United States, ²F.M. Kirby Research Center for Functional Brain Imaging, Kennedy Krieger Institute, Baltimore, MD, United States

Chemical exchange saturation transfer (CEST) imaging is a unique contrast enhancement technique that enables the indirect measurement of endogenous metabolites with water-exchangeable protons. The measurement of a complete Z-spectrum using standard imaging acquisition scheme is time consuming because a large number of the RF saturation frequency offset followed by MR signal readout is inevitable to obtain the full Z-spectrum. In this study, the feasibility of accelerated CEST imaging using combined CS and SENSE technique (CS-SENSE) was demonstrated in healthy volunteers and glioma patients at 3 T.

Merry Mani¹, Mathews Jacob², Douglas Kelley³, and Vincent Magnotta^4
1Dept. of Psychiatry, University of Iowa, Iowa City, IA, United States, ²Dept. of Electrical and Computer Engineering, University of Iowa, Iowa City, IA, United States, ³General Electric Healthcare Technologies, San Francisco, CA, United States, ⁴Dept of Radiology, University of Iowa, Iowa City, IA, United States

Multi-shot diffusion imaging holds great potential for enabling high spatial resolution diffusion imaging as well as short echo time imaging to enhance studies at higher field strengths. Conventional reconstructions rely on motion-induced phase estimates to recover the diffusion-weighted images from multi-shot acquisitions. Since a good estimate of phase cannot be obtained in many situations either due to noisy data or high under-sampling, these methods are unreliable in such situations. Here we propose a new reconstruction for multi-shot diffusion imaging that recovers the missing k-space data of the multiple shots by formulating the recovery as a structured low-rank matrix completion problem.

Peng Wu¹ and Hua Guo^1
1Center for Biomedical Imaging Research, Department of Biomedical Engineering, School of Medicine, Tsinghua University, Beijing, China, People's Republic of

Adaptive reconstruction (AR) can be used to combine multichannel images without acquiring coil sensitivity information. It can improve the SNRs of combined magnitude and phase images. But the reconstructed phase images may suffer from open-ended fringe artifacts when coil-sensitivity maps vary a lot. We propose a two-step adaptive reconstruction method to combine multichannel images. This method is shown to be more robust than the traditional AR method and can still maintain the high SNR.

Eric A. Borisch¹, Joshua D. Trzasko¹, and Stephen J. Riederer^1
1Mayo Clinic, Rochester, MN, United States

By adding an age-of-sample dependent weighting to the data fidelity penalty of an Alternating Direction Method of Multipliers sparse reconstruction of a view-shared accelerated acquisition, improved temporal fidelity is observed in a time-resolved motion-controlled phantom study.

Xi Peng¹, Shanshan Wang¹, Qingyong Zhu¹, and Dong Liang^1
1Shenzhen Institutes of Advanced Technology, Shenzhen, China, People's Republic of

In various MR applications, a pre-scan is usually practicable that extra morphology information can be extracted from the reference image. However, with a reference image obtained from a different contrast mechanism, the signal variation in the reference may differ from that in the target image. In this work, we propose to exploit gradient orientation information, which is closely related to the anatomical structures but less dependent on the image contrast, to enable superior CS-based reconstruction. The proposed method was validated using multi-scan experiment data and is shown to provide high speed and high quality imaging.

Sajan Goud Lingala¹, Sampada Bhave², Yinghua Zhu¹, Krishna Nayak¹, and Mathews Jacob^2
1Electrical Engineering, University of Southern California, Los Angeles, CA, United States, ²Electrical and Computer Engineering, University of Iowa, Iowa city, IA, United States

A number of dynamic MRI applications have seen the adaptation of data-driven models for efficient de-noising and reconstruction from under-sampled data. In this work, we develop a novel temporal point spread function interpretation of two data-driven models: low rank, and dictionary-learning. Through this interpretation, we show (a) the low rank model to perform spatially invariant non-local view-sharing, and (b) the dictionary-learning model to perform spatially varying non-local view-sharing. Both the models can be viewed as efficient data-driven retrospective binning techniques. We provide demonstrations using the application of de-noising real-time MRI speech data. 

Enhao Gong¹, Tao Zhang¹, Joseph Cheng¹, Shreyas Vasanawala², and John Pauly^1
1Electrical Engineering, Stanford University, Stanford, CA, United States, ²Radiology, Stanford University, Stanford, CA, United States

Low-Rank methods are widely applied to improve reconstruction for Dynamic Contrast Enhanced (DCE) MRI by imposing linear spatial-temporal correlation in global, local or multiple scales. This assumption does not fully capture the highly nonlinear spatial-temporal dynamics of DCE signals. We proposed a generalized Kernelized-Low-Rank model, assumed Low-Rank property after nonlinear transform and solved it by Regularizing singular-values with Adaptive Nonlinear Kernels. The proposed method captures the spatial-temporal dynamics as a sparser representation and achieves more accurate reconstruction results. Kernelized-Low-Rank model can be easily integrated to provide significant improvements to Global Low-Rank, Locally Low-Rank, LR+S and Multi-scale LR models.

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

Multi-flip, multi-echo 3D FLASH is routinely used to estimate T1. We present a novel, efficient algorithm for finding both the ML estimate and a confidence interval for T1, given full, multi-channel data from such an experiment. Finding the ML estimate is based on minimizing a cost function. The ends of the CI are a threshold value for the same cost function, where the threshold is determined via Monte Carlo experiment based on the protocol and desired confidence level. Results from a phantom study are shown, demonstrating the value of CIs in interpreting the validity of the ML estimates.

Volkert Roeloffs¹, Xiaoqing Wang¹, and Jens Frahm^1
1Biomedizinische NMR Forschungs GmbH, Max Planck Institute for Biophysical Chemistry, Goettingen, Germany

Fast and accurate determination of T1 values can be accomplished by Look-Locker type MRI sequences. Here, we formulate T1 mapping as a nonlinear optimization problem which we subsequently solve by the iteratively regularized Gauss Newton (IRGN) method. Our choice of the model parameterization allows to exploit smoothness of the spatial flip angle distribution as additional prior knownledge. This model-based reconstruction allows accurate and precise reconstruction of high resolution T1 maps from radial, highly undersampled data as validated in phantom studies and demonstrated in the human brain.

Hyunyeol Lee¹ and Jaeseok Park^2
1Center for Neuroscience Imaging Research, Institute for Basic Science, Suwon, Korea, Republic of, ²Department of Biomedical Engineering, Sungkyunkwan University, Suwon, Korea, Republic of

In this work we propose a novel, GM-selective single-slab 3D dual echo TSE with relaxation modulation (no long IR), in which white matter signals remain nulled before signal encoding while CSF signals are modulated to be cancelled during residual reconstruction between echoes with sparsity prior. We demonstrate the effectiveness of the proposed method over conventional DIR in that the former can achieve 1mm-isotropic whole-brain GM acquisition in 5-6 min. without apparent artifacts.   

Hyunkyung Maeng^1,2 and Jaeseok Park^2
1Center for Neuroscience Imaging Research, Institute for Basic Science (IBS), Suwon, Korea, Republic of, ²Department of Biomedical Engineering, Sungkyunkwan University, Suwon, Korea, Republic of

It is important to assess liver stiffness and monitor the progression of fibrosis in patients with liver diseases using non-invasive imaging. Exploiting the fact that the motion of the heart during cardiac cycle is an intrinsic driving source to deform the liver, dynamic tagged MRI can be employed to measure the cardiac motion induced displacements using harmonic phase images and thereby evaluate the corresponding strain maps in the liver. In this work, we propose a novel framework of compressed sensing (CS) for dynamic tagged liver MRI, in which: 1) data is acquired using tag-constrained, incoherent undersampling in k-t space, 2) a time series of morphologies is decomposed into transient tag-only images and stationary tag-free liver images, 3) both morphological components are then reconstructed directly from the tag-constrained, undersampled k-t space, and 4) the transient tag-only images are employed to estimate time-varying displacements and the corresponding strain maps while the stationary tag-free liver images are used as a structural roadmap.

Suhyung Park¹ and Jaeseok Park^2
1Center for Neuroscience Imaging Research, Institute for Basic Science (IBS), Suwon, Korea, Republic of, ²Department of Biomedical Engineering, Sungkyunkwan University, Suwon, Korea, Republic of

Dynamic contrast-enhanced magnetic resonance angiography(DCE-MRA) has been widely used for diagnostic assessment in clinical practices. To enhance the conspicuity of arteries relative to unwanted background tissues, subtraction between the pre-contrast and the post-contrast images was typically performed displaying maximum intensity projection images(MIP). Nevertheless, if there exists non-stationary signal transition due to time-drifting field inhomogeneity, and subject motion etc, the subtraction leads to incomplete background suppression, impairing the detectability of arteries as well as small vessel particularly at high reduction factors. In this work, we propose a novel DCE-MRA method exploiting motion subspace learning and sparsity priors for robust angiogram separation, in which the motion subspace is learned using partial Casorati matrix without any motion information while image reconstruction with sparsity priors is performed to jointly estimate motion-induced artifacts and DCE angiograms of interest under the framework of the decomposition. Simulation and experimental studies show that the proposed method is highly competitive with the competing methods including subtraction and fast reconstruction techniques.

Lyu Li¹, Sheng Fang², Pascal Spincemaille³, Bida Zhang⁴, Yi Wang^3,5, and Hua Guo^1
1Center for Biomedical Imaging Research, Department of Biomedical Enginnering, School of Medicine, Tsinghua University, Beijing, China, People's Republic of, ²Institute of nuclear and new energy technology, Tsinghua University, Beijing, China, People's Republic of, ³Radiology, Weill Cornell Medical College, New York, NY, United States, ⁴Healthcare Department, Philips Research China, Shanghai, China, People's Republic of, ⁵Biomedical Engineering, Cornell University, New York, NY, United States

High spatiotemporal resolution 3D imaging is a desired technique for detecting dynamic information of detailed anatomy and getting accurate modeling parameters. In this abstract, we developed a method using golden angle spiral and compressed sensing with L0 homotopic minimization for high resolution 3D imaging. In this method, a high undersampling rate and high motion insensitivity can be achieved. In clinical applications such as dynamic contrast enhanced MRI, this new method is very promising to achieve high spatiotemporal resolution without breath-hold.

Brady Quist^1,2, Xinwei Shi^1,2, Hans Weber¹, and Brian A Hargreaves^1,2
1Department of Radiology, Stanford University School of Medicine, Stanford, CA, United States, ²Department of Electrical Engineering, Stanford University, Stanford, CA, United States

In order to image in the large B₀ field inhomogeneity near metallic implants, multi-spectral imaging sequences, such as MAVRIC-SL, acquire images at multiple spectral bins to reduce distortion while imaging the entire anatomy. However, blurring is introduced when these overlapping bins are combined. While a deblurring method, which uses a field-map estimated from the bin images, does exist, the resulting images are subject to increased noise and distortion. Here we propose both a better field-map estimation method along with an improved deblurring algorithm, both of which are less sensitive to noise and help provide excellent deblurring without distortion.

Mehdi Hedjazi Moghari^1,2, Jonathan I Tamir ³, John Axerio-Cilies³, Tal Geva^1,4, and Andrew J Powell^1,4
1Pediatrics, Harvard Medical School, Boston, MA, United States, ²Cardiology, Boston Children's Hospital, Boston, MA, United States, ³Arterys, San Francisco, CA, United States, ⁴Cardiology, Boston Children’s Hospital, Boston, MA, United States

We developed and evaluated a variable density Poisson disc undersampling technique and compressed sensing image reconstruction algorithm for free-breathing 3D cine steady-state free precession whole-heart imaging. In 10 patients, we found good agreement between 3D cine and conventional breath-hold 2D cine imaging measurements of ventricular volumes. 

T. Correia¹, G. Cruz¹, R. M. Botnar¹, and C. Prieto^1
1Division of Imaging Sciences and Biomedical Engineering, King's College London, London, United Kingdom

Accelerated 100% scan efficiency whole-heart 3D coronary MR angiography (CMRA) is achieved by combining undersampling and non-rigid motion correction in a unified regularized reconstruction. 3D undersampled CMRA is performed using a golden-step spiral-like Cartesian trajectory. Motion correction is achieved in two steps: beat-to-beat 2D translational correction with motion estimated from interleaved image navigators, and bin-to-bin 3D non-rigid correction with motion estimated from the data itself. A generalized matrix formalism with total variation regularization is used to perform the non-rigid correction directly in the reconstruction. This approach produces good quality images, comparable to those of a navigator-gated approach, but in a ~5x shorter scan time.

Tess E. Wallace¹, Andrew J. Patterson², Roie Manavaki¹, Martin J. Graves¹, and Fiona J. Gilbert^1
1Department of Radiology, University of Cambridge, Cambridge, United Kingdom, ²Cambridge University Hospitals NHS Foundation Trust, Cambridge, United Kingdom

Physiological fluctuations and motion artifacts are expected to be dominant sources of noise in BOLD fMRI experiments to assess tumor oxygenation and angiogenesis. This work assesses the impact of a non-rigid registration algorithm and retrospective image correction (RETROICOR) on the detection of activation signals in the breast, both at resting state and in response to a modulated respiratory stimulus paradigm. Our results suggest that correction for motion artifacts is associated with a reduction in false-positive activation effects, which can be further improved by the addition of RETROICOR, confirming the importance of these physiological corrections in functional parameter estimation.

Mahamadou Diakite¹, Steve Roys², Taehoon Shin², Jiachen Zhuo², Jaydev P. Desai³, and Rao P. Gullapalli^2
1Radiology, Center for Metabolic Imaging and Therapeutics, University of Maryland, School of Medicine, Parkville, MD, United States, ²Radiology, Center for Metabolic Imaging and Therapeutics, University of Maryland, School of Medicine, Baltimore, MD, United States, ³Department of Mechanical Engineering, University of Maryland, College Park, Maryland, United States, Baltimore, MD, United States

We present a prospective motion correction (PMC) technique using a miniature passive magnetic sensor. The motion sensor can virtually work with any imaging technique that is sensitive to motion. Especially techniques such as fMRI, DTI and spectroscopy sequences are likely to benefit greatly when dealing with non-cooperative subjects.

In this study, the GRE sequence was initially modified by adding three bipolar gradients along the read, phase, and slice-selection directions respectively. The bipolar gradients were used to trigger the position and orientation tracking sensor. A dynamic feedback loop mechanism was implemented into the sequence to receive the sensor position and orientation for real-time update of the imaging slice using an in-house developed application.  

We demonstrate that our PMC method in brain imaging using a passive magnetic sensor is capable of tracking the patient head motion with high accuracy.

Chandan Aladahalli¹, KS Shriram¹, Dattesh Shanbhag¹, Rakesh Mullick¹, Reem Bedair², Fiona Gilbert², Andrew Patterson², and Martin Graves^2
1GE Global Research, Bengaluru, India, ²University of Cambridge, Cambridge, United Kingdom

Multi-parametric longitudinal imaging is an important tool to monitor therapy response at tumor locations. The key challenge in longitudinal data is the synchronization of volumes across time points, as large variations are seen due to therapy response, weight loss/gain, patient positioning, and surgery. We employ a group-wise registration approach to synchronize the longitudinal volumes and compare/contrast it with a more typical pair-wise registration approach. Group-wise approach results in consistent registration across time-points as it does not require a reference image. The group-wise non-rigid registration approach is demonstrated on longitudinal MRI data used for assessing breast tumor therapy response.

Andrew J Coristine¹, Jerome Chaptinel¹, Giulia Ginami¹, Gabriele Bonanno¹, Simone Coppo¹, Ruud B van Heeswijk¹, Davide Piccini^1,2, and Matthias Stuber^1,3
1Department of Radiology, University Hospital (CHUV) / University of Lausanne (UNIL), Lausanne, Switzerland, ²Advanced Clinical Imaging Technology, Siemens Healthcare, Lausanne, Switzerland,³CardioVascular Magnetic Resonance (CVMR) research centre, Centre for Biomedical Imaging (CIBM), Lausanne, Switzerland

In respiratory self-navigation (SN), static structures, such as the arms or chest wall, may complicate motion detection due to the superposition of signal originating from different tissues. Even if motion detection is successful, the subsequent motion correction may introduce streaking artefacts when applied to static structures. Suppressing signal from those tissues may therefore improve image quality. In this study, we address the hypothesis that SN coronary MRA will benefit from the introduction of an outer volume suppressing "2D-T -Prep", and present results from a moving cardiac phantom and 10 healthy volunteers.

Hing-Chiu Chang^1,2, Mei-Lan Chu², and Nan-Kuei Chen^2
1Department of Diagnostic Radiology, The University of Hong Kong, Hong Kong, Hong Kong, ²Brain Imaging and Analysis Center, Duke University Medical Center, Durham, NC, United States

The Propeller technique is a useful acquisition scheme and reconstruction method to reduce motion artifact. However, the higher specific absorption rate (SAR) of RF pulse at high-field magnetic strength can limit the number of multiple slices for a given TR, which in turn reduces the efficiency of acquisition, especially for Propeller-FSE sequence. A possible solution is to reduce the echo-train-length with under-sampling of data of each blade. In this study, we propose to apply POCSMUSE instead of SENSE reconstruction and Propeller reconstruction, to reconstruct the image from all under-sampled blade data with reduced noise amplification. The data acquired from brain and liver using Propeller-GRE will be used to test purposed POCSMUSE algorithm.

Christopher M Rank¹, Thorsten Heußer¹, Andreas Wetscherek¹, Martin T Freitag², Heinz-Peter Schlemmer², and Marc Kachelrieß^1
1Medical Physics in Radiology, German Cancer Research Center (DKFZ), Heidelberg, Germany, ²Radiology, German Cancer Research Center (DKFZ), Heidelberg, Germany

To allow for MR acquisition times as short as 1 minute per bed, we propose a new method for PET/MR respiratory motion compensation (MoCo), which is based on strongly undersampled radial MR data. We acquired simultaneous PET/MR data of the thorax of three patients. Motion vector fields were estimated with a newly-developed algorithm, which alternates between MR image reconstruction and motion estimation. Subsequent 4D MoCo PET reconstructions employing the motion vector field derived from strongly undersampled MR data yielded a considerable visual and quantitative improvement compared to standard 3D PET and 4D gated PET reconstructions.

Thomas Koesters¹, Ryan Brown¹, Tiejun Zhao², Mattias Fenchel³, Peter Speier³, Li Feng¹, Yongxian Qian¹, and Fernando Emilio Boada^1
1Radiology, New York University, New York, NY, United States, ²Siemens Medical Systems, Malvern, PA, United States, ³Siemens Medical Systems, Erlangen, Germany

We demonstrate the use of an external RF signal as a mean to provide motion tracking information for motion-state sorting of k-space line during dynamic MRI scans.

Frédéric Gretsch¹, Tobias Kober^2,3,4, Maryna Waszak^2,3,4, José P. Marques⁵, and Daniel Gallichan^1
1CIBM, EPFL, Lausanne, Switzerland, ²Advanced Clinical Imaging Technology (HC CMEA SUI DI BM PI), Siemens Healthcare AG, Lausanne, Switzerland, ³Department of Radiology, University Hospital (CHUV), Lausanne, Switzerland, ⁴LTS5, EPFL, Lausanne, Switzerland, ⁵Donders Centre for Cognitive Neuroimaging, Radboud University, Netherlands

FID navigators (FIDNavs) and a few lines of highly accelerated double-echo 3D fat navigators (FatNavs) were measured each TR of a high-resolution 3D-GRE acquisition. High temporal resolution $$$B_0$$$ variations and motion parameters could be estimated by the FIDNavs using the lower temporal resolution FatNavs to derive calibration data. The cardiac cycle pattern emerged clearly in these estimates and retrospectively corrected images were of clearly improved quality, thereby demonstrating the potential for this hybrid approach.

Nathan White¹, Josh Kuperman¹, Neal Corson¹, Kazim Narisinh¹, Hauke Bartsch¹, David Karow¹, Ajit Shankaranarayanan ², and Anders Dale^1,3
1Department of Radiology, University of California, San Diego, San Diego, CA, United States, ²Global Applied Sciences Lab, GE Healthcare, Menlo Park, CA, United States, ³Department of Neuroscience, University of California, San Diego, San Diego, CA, United States

Free-breathing single-shot EPI data in the abdomen is confounded by respiratory motion artifact. Previous work has demonstrated the utility of using the Extended Kalman Filter (EKF) for real-time image-based tracking of spiral-navigated scans in brain. This this study we demonstrate the feasibility of using the EKF framework for real-time self-navigated motion tracking in the abdomen. 

Dominik Nickel¹, Xiao Chen², Boris Mailhe², Qiu Wang², Yohan Son³, Jeong Min Lee⁴, and Berthold Kiefer^1
1Siemens Healthcare GmbH, Erlangen, Germany, ²Medical Imaging Technologies, Siemens Healthcare, Princeton, NJ, United States, ³Siemens Healthcare Ltd., Seoul, Korea, Republic of, ⁴Department of Radiology, Seoul National University Hospital, Seoul, Korea, Republic of

Free-breathing dynamic contrast-enhanced liver MRI is explored using a Cartesian spoiled Volume-Interpolated Breath-hold Examination (VIBE) GRE sequence that also acquires a navigation signal. Images are iteratively reconstructed using a hard-gating approach as well as resolving the motion-states using an additional dimension. With both approaches yielding promising results, the latter appears to be more motion robust at the cost of computational effort.

Mads Andersen^1,2, Kristoffer H. Madsen¹, and Lars G. Hanson^1,2
1Danish Research Center for Magnetic Resonance, Copenhagen University Hospital Hvidovre, Hvidovre, Denmark, ²DTU Elektro, Technical University of Denmark, Lyngby, Denmark

A new prospective motion correction technique is presented that is based on signals from gradient switching, in an EEG-cap with interconnected electrodes the subject wears during scanning. The method has no line-of-sight limitations as optical methods, requires no interleaved navigator modules or additional hardware for sites already doing EEG-fMRI. Instead a training scan is performed were signals recorded with the EEG-system are correlated with motion parameters estimated by image realignment. Initial results from application of the method in a phantom are promising.

Niranchana Manivannan¹, Bradley D. Clymer¹, Anna Bratasz^2,3, and Kimerly A. Powell^2,4
1Department of Electrical and Computer Engineering, The Ohio State University, Columbus, OH, United States, ²Small Animal Imaging Shared Resources, The Ohio State University, Columbus, OH, United States, ³Davis Heart & Lung Research Institute, The Ohio State University, Columbus, OH, United States, ⁴Department of Biomedical Informatics, The Ohio State University, Columbus, OH, United States

In most small animal imaging studies both long axis (coronal or sagittal) and short axis (axial) images of the region of interest are obtained. The goal of this study is to explore whether combining two orthogonal views obtained with different slice-select directions could reduce the motion artifacts and improve image quality. The advantages of this method are that no a priori knowledge or external hardware is needed and it doesn’t increase the acquisition time. The results show that it is advantageous to use the available orthogonal image(s) to improve the image quality by reducing ghosting artifacts caused by motion.

Bhairav Bipin Mehta¹, Dan Ma¹, and Mark Alan Griswold^1
1Radiology, Case Western Reserve University, Cleveland, OH, United States

Motion is one of the biggest challenges in clinical MRI. The recently introduced Magnetic Resonance Fingerprinting (MRF) has been shown to be less sensitive to motion. However, it is still susceptible to patient motion primarily occurring in the early stages of the acquisition. In this study, we propose a novel reconstruction algorithm for MRF, which uses robust fitting and image registration algorithms to decrease the motion sensitivity of MRF. The evaluation was performed in numerical phantoms with simulated rigid motion.

Daniel V Litwiller¹, Erik Tryggestad², Kiaran McGee³, Yuji Iwadate⁴, Lloyd Estkowski⁵, and Ersin Bayram^6
1Global MR Applications & Workflow, GE Healthcare, New York, NY, United States, ²Department of Radiation Oncology, Mayo Clinic, Rochester, MN, United States, ³Department of Radiology, Mayo Clinic, Rochester, MN, United States, ⁴Global MR Applications & Workflow, GE Healthcare, Hino, Tokyo, Japan, ⁵Global MR Applications & Workflow, GE Healthcare, Menlo Park, CA, United States, ⁶Global MR Applications & Workflow, GE Healthcare, Houston, TX, United States

Here we present a multi-slice, multi-phase, single-shot fast spin echo sequence with variable refocusing flip angle and interleaved cylindrical reference navigator for retrospective respiratory-guided image sorting for the purpose of managing motion in the context of MR-guided radiation therapy treatment planning.

Falk Lüsebrink¹, Alessandro Sciarra¹, Hendrik Mattern¹, Renat Yakupov¹, and Oliver Speck^1,2,3,4
1Biomedical Magnetic Resonance, Otto-von-Guericke University, Magdeburg, Germany, ²Leibniz Institute for Neurobiology, Magdeburg, Germany, ³Center for Behavioral Brain Sciences, Magdeburg, Germany, ⁴German Center for Neurodegenerative Disease, Magdeburg, Germany

Increasing the spatial resolution is of major importance to structural imaging as this may build bridges to optical microscopy and may lead to superior diagnostics and segmentations. Increasing the spatial resolution with sufficient SNR usually prolongs time of acquisition. This inevitably introduces more motion artifacts even with experienced subjects. However, this can be compensated by prospective motion correction. Increasing the resolution to a few hundred micrometers inherently reduces SNR such that reconstruction by sum of squares is not adequate anymore. Here we demonstrate our work on the acquisition and reconstruction of the currently highest resolution in vivo MPRAGE at 7T. 

Yixin Ma¹, Yoann Petibon², Joyita Dutta², Xucheng Zhu³, Rong Guo¹, Georges El Fakhri², Kui Ying¹, and Jinsong Ouyang^2
1Key laboratory of Particle and Radiation Imaging, Ministry of Education, Tsinghua University, Beijing, China, People's Republic of, ²Center for Advanced Medical Imaging Sciences, Division of Nuclear Medicine and Molecular Imaging, Department of Imaging, Massachusetts General Hospital, Boston, MA, United States, ³UCSF/UC Berkeley Bioengineering Graduate Group, University of California Berkeley, Berkeley, CA, United States

In this work, we exploited the low-rank and sparse properties of dynamic MRI data to accelerate the MR acquisition and assessed its impact on MR-based PET respiratory motion using a PET/MR oncologic patient study.  PET/MR motion correction workflow was accomplished, and good performance of partially sampled MR based motion correction attests the possibility of faster PET/MR data acquisition and better lesion estimation.

Jianing Pang¹, Yuhua Chen^1,2, and Debiao Li^1,3
1Biomedical Imaging Research Institute, Cedars-Sinai Medical Center, Los Angeles, CA, United States, ²Department of Computer and Information Science, University of Pennsylvania, Philadelphia, PA, United States, ³Bioengineering, University of California, Los Angeles, CA, United States

Current motion compensation strategies for whole-heart coronary MRA only accept data acquired during specific motion phases, thus significantly reduce the imaging efficiency. In this work, we developed a reconstruction framework based on 4D coronary MRA that allowed one to increase the cardiac gating efficiency by accepting cardiac phases beyond the quiescent period, while minimizing the associated cardiac motion artifacts through non-rigid motion correction. Preliminary results from healthy subjects showed that the proposed method significantly improved the imaging efficiency, aSNR, and coronary sharpness compared with images reconstructed from the quiescent period. 

Emre Kopanoglu¹, Haifeng Wang¹, Gigi Galiana¹, and Robert Todd Constable^1,2
1Diagnostic Radiology and Biomedical Imaging, Yale University, New Haven, CT, United States, ²Neurosurgery, Yale University, New Haven, CT, United States

Motion navigation using nonlinear gradient fields is demonstrated experimentally. The method makes use of the simultaneous multi-dimensional encoding capabilities of nonlinear gradient fields. A two-dimensional navigator image is obtained from a single-echo encoded using a nonlinear gradient field and multiple receiver coils. Without exceeding the maximum field generated by the linear gradient fields of a 3T scanner inside a 20cm isotropic field-of-view, the navigator can be acquired in under one millisecond, including its rewinder. The method can track both translational and rotational in-plane rigid body motion, as demonstrated in phantom experiments. Simulations show the method is applicable in oblique angles.

Thomas Martin^1,2, Andres Saucedo¹, Tess Armstrong¹, Holden Wu², Danny Wang³, and Kyunghyun Sung^2
1Biomedical Physics, UCLA, Los Angeles, CA, United States, ²Radiological Sciences, UCLA, Los Angeles, CA, United States, ³Neurology, UCLA, Los Angeles, CA, United States

Respiratory motion is one of the biggest confounders of liver DCE-MRI.  There are methods that use a 3D radial self-gated signal (SGS) to compensate for respiratory motion.  However, SGS includes both respiratory motion and contrast uptake in DCE-MRI, and it is not trivial to perfectly separate the two from SGS, leading to inaccuracies of respiratory motion. In this work, we propose a method to extract the respiratory motion only from SGS using golden angle radial acquisition with two-point Dixon separation. The proposed method utilizes the fact the fat-only SGS does not include contrast uptake while including the same respiratory motion.

Yilong Liu^1,2, Mengye Lyu^1,2, Yanqiu Feng^1,2, Victor B. Xie^1,2, and Ed X. Wu^1,2
1Laboratory of Biomedical Imaging and Signal Processing, The University of Hong Kong, Hong Kong, China, People's Republic of, ²Department of Electrical and Electronic Engineering, The University of Hong Kong, Hong Kong, China, People's Republic of

A navigator-free motion correction method for Cartesian FSE is proposed for intermittent motion compensation. It first divides all shots into several motion-free groups, then estimates and corrects inter-group motion. In vivo imaging results show that this approach has decent performance in dealing with intermittent motion problems. Though only demonstrated with Cartesian acquisition, it is also applicable for non-Cartesian acquisitions, such as spiral and radial acquisitions.

Denis Kokorin¹, Cris Lovell-Smith¹, Benjamin Knowles¹, Michael Herbst¹, Jürgen Hennig¹, and Maxim Zaitsev^1
1Medical Physics, University Medical Center Freiburg, Freiburg, Germany

In this work, we present a new k-space based approach for position and rotation detection to improve calibration of optical tracking systems used for prospective motion correction in MRI. 3D radial acquisition was employed to detect motion of a custom built phantom. The integrated method was tested for different motion types and showed a high precision along with a good robustness for rotation detection.

Dongxiao Li^1,2, Juerong Wu¹, Kofi M. Deh², Thanh D. Nguyen², Martin R. Prince², Yi Wang^2,3, and Pascal Spincemaille^2
1College of Information Science and Electronic Engineering, Zhejiang University, Hangzhou, China, People's Republic of, ²Department of Radiology, Weill Cornell Medical College, New York, NY, United States, ³Department of Biomedical Engineering, Cornell University, Ithaca, NY, United States

Liver dynamic contrast enhanced MRI (DCE-MRI) requires high spatial and temporal resolution such that all relevant enhancement phases are clearly visualized. Image quality is compromised when breathing occurs during the acquisition. This abstract presents a novel 4D respiratory Motion corrected Patch based Reconstruction of Under-sampled Data (M-PROUD) which uses 3D patch based local dictionaries for sparse coding and simultaneously estimates 3D nonrigid motion. Results on in vivo data demonstrated that the proposed method can significantly reduce motion blurring artifacts and preserve more details at a sub-second temporal frame rate in free breathing liver 4D DCE-MRI.

Robert Grimm¹, Dominik Nickel¹, Qiu Wang², Boris Mailhe², Berthold Kiefer¹, Kai Tobias Block³, and Tobias Heye^4
1Siemens Healthcare GmbH, Erlangen, Germany, ²Medical Imaging Technologies, Siemens Healthcare, Princeton, NJ, United States, ³Radiology, New York University School of Medicine, New York City, NY, United States, ⁴Klinik für Radiologie und Nuklearmedizin, Universitätsspital Basel, Basel, Switzerland

Although the radial sampling scheme of GRASP DCE-MRI leads to inherent motion robustness, strong respiration can still impair the image quality of free-breathing abdominal examinations. We propose to combine GRASP with respiratory gating to minimize motion artifacts within and between temporal frames, without prolonging the reconstruction time and maintaining a high spatio-temporal resolution. A validation in 10 clinical patients confirmed the improved sharpness of the gated reconstruction.

Barbara Dymerska¹, Siegfried Trattnig¹, and Simon Daniel Robinson^1
1High Field MR Centre, Department of Biomedical Imaging and Image-guided Therapy, Medical University of Vienna, Vienna, Austria

A common assumption for phase combination and dynamic distortion correction is that phase offsets ($$$?_0$$$), i.e. the phase of coil sensitivities, are temporally stable. We investigate the validity of this assumption at 7T for long measurements (40min) and large head rotations (up to 12°) made with multi-channel coils. We show that changes in separate-channel $$$?_{0,ch}$$$ have little effect on the combined phase images (unwarping errors of 0.2 voxel, reduction in phase matching quality by 2%). We thus conclude that the assumption of $$$?_0$$$ temporal stability holds and methods based on this assumption should work at 7T with substantial motion.

Ashley G Anderson III¹ and James G Pipe^1
1Imaging Research, Barrow Neurological Institute, Phoenix, AZ, United States

An inpainting technique using the fast marching method was used to implement a fast, robust algorithm for extrapolating ?$$$\Delta f_0$$$ map data into low signal areas.

Peter Andrew Hardy¹ and Erfan Akbari^2
1Radiology, University of Kentucky, Lexington, KY, United States, ²Radiation Medicine, University of Kentucky, Lexington, KY, United States

We developed an imaging and analysis method to use the spatial distortion in echo planar images to estimate the magnetic field inhomogeneity. The method requires the acquisition of two images from an unmodified echo planar sequence with the direction of the phase encode reversed between the two. After combining the images appropriately an undistorted and a displacement image are produced. The displacement image is a map of the local magnetic field inhomogeneity.

A Alhamud¹, Ernesta M. Meintjes¹, and André J.W. van der Kouwe^2
1Human Biology,MRC/UCT Medical Imaging Research Unit, University of Cape Town, Cape Town, South Africa, ²Massachusetts General Hospital, Charlestown, MA, United States

Most DTI acquisitions are based on 2D multislice EPI sequences. The spatial distribution of the field inhomogeneity may differ from region to region within the DTI volume. While the scanner and traditional prospective shimming methods shim over the whole volume, this may not be optimal for DTI. Further, changes in the shim in different regions across DTI volume in the presence of subject motion are yet to be explored. In this work, we introduce a technique to measure and correct the changes in the static field from region to region and for different sized regions across the DTI volume

Kay Nehrke¹, Arthur Felipe Nisti Grigoletto Borgonovi², and Peter Börnert^1,3
1Philips Research, Hamburg, Germany, ²Philips Healthcare, Best, Netherlands, ³Radiology, LUMC, Leiden, Netherlands

Simultaneous multi-slice imaging has been employed for the DREAM B₁⁺ mapping approach, allowing a multi-slice B₁⁺ map to be acquired in a fraction of a second. Basic feasibility has been shown in experiments on phantoms and in vivo using a clinical 3T MRI system. The presented approach potentially allows free-breathing multi-slice B₁⁺ mapping, freezing respiratory motion in a short single shot acquisition window. 

Vincent Gras¹, Alexandre Vignaud¹, Franck Mauconduit², Michel Luong³, Alexis Amadon¹, Denis Le Bihan¹, and Nicolas Boulant^1
1Neurospin, CEA/DSV/I2BM, Gif-sur-Yvette, France, ²Siemens Healthcare, Saint-Denis, France, ³IRFU, CEA/DSM, Gif-sur-Yvette, France

The standard approach to design radiofrequency pulses in MRI is to minimize the deviation of the flip angle from a target value. An alternative approach is proposed here for the MPRAGE sequence which uses the signal as a surrogate of the flip angle in the optimization of the excitation and inversion pulses. The results obtained in simulation and in the brain in vivo on a parallel transmission enabled 7T scanner show two possible applications of the method: an improvement in image quality or a significant reduction of the SAR at equivalent image quality.

Denis Kokorin¹, Jürgen Hennig¹, and Maxim Zaitsev^1
1Medical Physics, University Medical Center Freiburg, Freiburg, Germany

In this study, the feasibility of 2D spiral-encoded parallel excitation of limited slice profiles was investigated. The imaging experiments were performed in phantoms on a 3T MRI system with 8 RF channels for transmission and compared to 2D parallel excitation using EPI encoding. The resulting profiles revealed that 2D spiral-encoded parallel excitation is more robust against B1 deviations compared to EPI.

Xiaoping Wu¹, Gregor Adriany¹, Eddie J. Auerbach¹, Sebastian Schmitter¹, Kamil Ugurbil¹, and Pierre-Francois Van de Moortele ^1
1CMRR, Radiology, University of Minnesota, Minneapolis, MN, United States

Increased signal to noise ratio and tissue contrast are strong incentive for pushing toward higher magnetic fields. However, as the magnetic field increases, transmit B1 fields become more and more non-uniform, leading to spatially varying contrast and local signal dropouts. This problem can be addressed with parallel RF transmission (pTx). We have recently made operational the first 10.5 Tesla whole body MRI scanner which holds promise for a wide range of biomedical applications. In this study, we assessed the performance of the installed 16-channel pTx system by designing 2D Transmit SENSE pulses. Our results suggest that high fidelity excitation patterns can be attained after correction of system imperfections. 

Vincent Gras¹, Michel Luong², Alexis Amadon¹, and Nicolas Boulant^1
1Neurospin, CEA/DSV/I2BM, Gif-sur-Yvette, France, ²IRFU, CEA/DSM, Gif-sur-Yvette, France

The k_T-points parametrization is a powerful technique to mitigate the RF inhomogeneity at ultra-high field which can also be used to achieve homogeneous non-selective inversion profiles with less SAR than adiabatic RF pulses. The large flip angle regime thus is an interesting domain of application of the k_T-points technique. However, to fully exploit it, it is necessary to optimize the placement of the k_T-points in k-space. In this work, the simultaneous optimization of the RF and k-space trajectory coefficients is proposed and validated, whereby SAR and power constraints are handled explicitly.

Tijl van der Velden¹, Peter R Luijten¹, and Dennis W.J. Klomp^1
1Radiology, UMC Utrecht, Utrecht, Netherlands

Using multiple local excitation coils for different parts of the body, such as bilateral breast coils, opens up the possibility to perform frequency shimming: using multiple carrier frequencies to excite the different body parts without gradient based localization. In this work we demonstrate frequency shimming to improve fat suppressed breast MRI at 7 T.

Mustafa Cavusoglu¹, Klaas Paul Pruessmann¹, and Shaihan Malik^2
1Institute of Biomedical Engineering, ETH Zurich, Zurich, Switzerland, ²Kings Collage London, London, United Kingdom

Variable-rate selective excitation (VERSE) is a powerful method to control RF power and SAR that bounds to a key condition as retaining the RF-to-gradient amplitude ratio at each sample that preserves the rotational behavior of on-resonance spins (1). The maintenance of VERSE condition strictly depends on the fidelity of the local gradient fields implying that any deviation from the nominal VERSE’d gradients will modulate the spin rotations similar to off-resonances ultimately resulting excitation errors and the RF pulse to converge to a significantly different peak RF power.

Tingting Shao¹, Nikolai I. Avdievich¹, and Anke Henning^1,2
1Max Planck Institute for Biological Cybernetics, Tübingen, Germany, ²Institute of Biomedical Engineering, University and ETH, Zürich, Switzerland

This work presents experimental results of 3D volumetric parallel excitation at a 9.4T human whole-body MRI scanner. The approach and concept of a “trajectory container” is adopted to match practical considerations. The “trajectory container” is used to shape the k space trajectory and restrict it to a limited traversing area in the k space and therefore to constrain the pulse duration. A simplified and direct way of the definition of the “trajectory container” is proposed and verified with promising experimental results.

Denis Kokorin¹, Jürgen Hennig¹, and Maxim Zaitsev^1
1Medical Physics, University Medical Center Freiburg, Freiburg, Germany

In this work, we discuss off-resonance effects observed in zoomed DWI combined with 2D pulses based on EPI trajectories. Our main focus is placed on the use of parallel excitation (PEX) for shortening 2D pulses and further minimizing the distortion. Experimental data were obtained using a 3T MRI system with an 8-channel TxArray extension and analyzed in detail. We conclude that the use of PEX improves the matching between excitation and acquisition in zoomed EPI applications.

Martina F Callaghan¹, Frederic Dick², Patrick Grabher³, Tim Keller⁴, Patrick Freund^1,3,5, and Nikolaus Weiskopf^1,5
1Wellcome Trust Centre for Neuroimaging, UCL Institute of Neurology, London, United Kingdom, ²Birkbeck/UCL Centre for Neuroimaging, London, United Kingdom, ³Spinal Cord Injury Center Balgrist, University Hospital Zurich, Zurich, Switzerland, ⁴Department of Psychology, Carnegie Mellon University, Pittsburgh, PA, United States, ⁵Department of Neurophysics, Max Planck Institute for Human Cognitive and Brain Sciences, Leipzig, Germany

Unified segmentation based correction of R₁ brain maps for RF transmit field inhomogeneities (UNICORT) has previously been shown to reduce bias in R₁ maps caused by inhomogeneity in the RF transmit field (B₁⁺). This approach simultaneously estimates the B₁⁺ inhomogeneities and R₁ values from the uncorrected R₁ maps without the need for additional B₁⁺ calibration data. It employs a probabilistic framework that incorporates a physically informed generative model of smooth transmit field inhomogeneities and their multiplicative effect on R₁ estimates. However, different systems may require different priors, depending on the particular transmit coil used.  Here we show that these parameters can be estimated using a linear relaxometry model framework, without the need to acquire B₁⁺ mapping data.

Hyungseok Jang^1,2 and Alan B McMillan^1
1Department of Radiology, University of Wisconsin, Madison, WI, United States, ²Department of Electrical and Computer Engineering, University of Wisconsin, Madison, WI, United States

Accurate knowledge of the k-space trajectory is critical for artifact-free MR imaging, particularly in non-Cartesian imaging. In this study we propose a new gradient measurement technique based on single point imaging (SPI), which allows simple, rapid, and robust measurement of k-space trajectory. In the proposed technique, the zoom-in/out effect of SPI is used for k-space trajectory measurement. First, 1D SPI data are acquired from a targeted gradient in each axis, and then relative FOV scaling factors between encoding times are found, which represents relative k-space position. Improvements in image quality are demonstrated for UTE, spiral, and ramp-sampled IDEAL imaging.

Ying-Hua Chu¹, Yi-Cheng Hsu¹, and Fa-Hsuan Lin^1
1Institute of Biomedical Engineering, National Taiwan University, Taipei, Taiwan

We use a dense 2D field probe array with 24 probes moving over a 3D volume to estimate magnetic field distribution dynamically and demonstrate that trajectory calibration is slice-dependent. The image reconstructed using measured dynamic magnetic field information 7 cm away shows similar l₂ norm error as the image reconstructed without any dynamic magnetic field information.

Ryan K Robison¹ and James G Pipe^1
1Imaging Research, Barrow Neurological Institute, Phoenix, AZ, United States

Gradient impulse response functions can predict gradient behavior in the presence of eddy currents and other sources of error. This work presents a rapid method for generating the 0th and 1st order gradient impulse response functions using existing software measurement techniques. The proposed method is applied to spiral imaging but could be similarly applied to EPI, projection reconstruction, or any other imaging sequence.

Andreas Lesch¹, Kristian Bredies², Clemens Diwoky³, and Rudolf Stollberger^1,4
1Institute of Medical Engineering, Graz University of Technology, Graz, Austria, ²Institute of Mathematics and Scientific Computing, University of Graz, Graz, Austria, ³Institute of Molecular Biosciences, University of Graz, Graz, Austria, ⁴BioTechMed Graz, Graz, Austria

In this work we present an approach for chemical shift based fat/water-separation using variational methods for the estimation of the underlying B₀-field inhomogeneity. We show that the fat/water-problem can be solved by the application of total-generalized-variation (TGV) regularization on the underlying B₀-field to enforce piecewise smoothness. With this approach we are able to model huge B₀-deviations as well as discontinuities in the B₀-field at tissue boundaries with different susceptibility parallel to the main-field. This is shown on different datasets, including two of the ISMRM fat/water-challenge of 2012.

Catherine J Moran¹, Bragi Sveinsson^1,2, Brady Quist², Marcus T Alley¹, Bruce Daniel¹, and Brian A Hargreaves^1
1Radiology, Stanford University, Stanford, CA, United States, ²Electrical Engineering, Stanford University, Stanford, CA, United States

The double echo steady state (DESS) acquisition has the potential to provide high-resolution and distortion-free T2 and diffusion-weighted images in the breast.  Initial investigations of DESS in the breast have been limited by the presence of ghosting artifacts. Respiratory-induced B0 variation is one source of these artifacts. A method utilizing an in-vivo time-varying off-resonance estimate along with an image entropy metric to assess the level of artifact is described and investigated for correction of ghosting due to respiration-induced B0 variation in DESS in the breast.

Shengzhen Tao¹, Paul T Weavers¹, Joshua D Trzasko¹, Yunhong Shu¹, Seung-Kyun Lee², Lou Frigo³, Scott Hinks³, and Matt A Bernstein^1
1Radiology, Mayo Clinic, Rochester, MN, United States, ²GE Global Research, Niskayuna, NY, United States, ³GE Healthcare, Milwaukee, WI, United States

Based on Maxwell’s equation, the spatial encoding gradient fields must be accompanied by spatially-varying higher-order concomitant fields. Different from conventional gradient systems whose concomitant fields contain second-order and higher spatial dependence, some MRI platforms employ asymmetric gradient systems, and their concomitant fields also include zero- and first-order terms. The first-order terms cause artifacts including image blurring and ghosting in spiral acquisition and echo shifting in EPI. Here, an efficient gradient pre-emphasis scheme suitable for real-time implementation is demonstrated to simultaneously correct all the first-order concomitant fields with a real-time implementation on gradient firmware typically used for eddy current compensation. 

Paul Polak^1,2, David Lindner¹, Jannis Hanspach¹, Michael Dwyer¹, Niels Bergsland¹, Nicola Bertolino^1,2, Robert Zivadinov^1,2, and Ferdinand Schweser^1,2
1Neurology, Buffalo Neuroimaging Analysis Center, State University of New York at Buffalo, Buffalo, NY, United States, ²Molecular and Translational Imaging, Clinical Translational Research Center, Buffalo, NY, United States

Gradient “unwarping” is the removal of image distortions caused by non-linearity of the imaging gradient fields. Exasperated by increasing distance from isocenter, the unwarping process is applied to every image by the manufacturer and generally is a “black-box” process occurring near the end of the image processing pipeline. This is problematic for researchers who source their data from a more primitive step, e.g. from raw k-space, since this data is generally “warped”. Presented here is a method to reverse engineer the spherical harmonic coefficients that describe the warp field and are used by the vendor’s black-box process, allowing the researcher to perform the gradient unwarping off-line as a post-processing step.

Saikat Sengupta¹, David S Smith¹, and Edward Brian Welch^1
1Department of Radiology, Vanderbilt University, Nashville, TN, United States

In this abstract, we present an approach for rapid mapping of the whole body B₁ field at 3 Tesla. We use a combination of dual interleaved TR B₁ mapping with continuously moving table imaging to measure the B₁ field in the whole body in 90 seconds. Considerable variation is observed in the B₁ distribution in different sections of the body. This measurement will serve as the basis for dynamic B₁field shimming with the moving table for whole body imaging in future work.  

Kevin D Harkins¹, Mary Katherine Manhard^1,2, William A Grissom^1,2,3, and Mark D Does^1,2,3,4
1Institute of Image Science, Vanderbilt University, Nashville, TN, United States, ²Biomedical Engineering, Vanderbilt University, Nashville, TN, United States, ³Electrical Engineering, Vanderbilt University, Nashville, TN, United States, ⁴Radiology and Radiological Sciences, Vanderbilt University, Nashville, TN, United States

Transient gradient waveform errors can degrade MRI quality. This work presents an automatic implementation of gradient waveform predistortion, which has been applied to half-pulse excited 2D-ultrashort echo time imaging. The predistortion method reliably improves image quality and quantification of ultrashort T₂ species by reducing out-of-slice signal that contaminates half-pulse excited images. 

Maysam M Jafar¹, Christopher E Dean¹, Malcolm J Birch¹, and Marc E Miquel^1
1Clinical Physics, Barts Health NHS Trust, City of London, United Kingdom

Major hardware-related geometric distortions in MRI arise from gradient field non-linearity and static field inhomogeneity.  For an accurate mapping of geometrical distortion in 3D, the number of control points must be sufficiently large to provide a comprehensive mapping of the spatial variation of distortion and the positional accuracy of these control points must be ensured. In this study, the spatial accuracy of coordinates in 3D space is assessed across six clinical MRI scanners at both 1.5T and 3T field strengths using a previously reported 3D-printed grid phantom. This is a cost-effective approach to determine the spatial accuracy of control points.

Rosie Goodburn¹, Nicholas Powell^2,3, James O’Callaghan², Simon Walker-Samuel², and Karin Shmueli^1
1Department of Medical Physics and Biomedical Engineering, University College London, London, United Kingdom, ²UCL Centre for Advanced Biomedical Imaging, Division of Medicine, London, United Kingdom, ³UCL Centre for Medical Imaging Computing, London, United Kingdom

Susceptibility artifacts hamper gradient-echo MRI of the mouse brain at 9.4 Tesla. Here, we characterised and modelled magnetic field maps aiming to improve preclinical mouse MRI. To characterise field perturbations, we measured and coregistered field maps in 7 mice to create a group-average field map. Next, a Fourier forward model was used to simulate a field map based on a susceptibility distribution constructed from the group-average magnitude image. The measured and modelled field maps showed similar qualitative field patterns, especially around the aural air cavities, but had quantitative differences. Work is ongoing to improve the accuracy of the model.

Jan-Willem Beenakker¹, Joep Wezel¹, Gregorius Luyten¹, Andrew Webb¹, and Peter Boernert^1,2
1Leiden University Medical Centre, Leiden, Netherlands, ²Philips Research Laboratories, Hamburg, Germany

MRI is becoming an increasingly valuable non-invasive tool in ocular tumour assessment and treatment planning. ophthalmology. High resolution images acquired at high field provide multi-dimensional information on tumour size. However, image quality is often compromised by eye motion which is often triggered by gradient noise In the present work the use of magnetization prepared 3D Zero Echo Time imaging (ZTE) is proposed, enabling for almost silent volumetric scanning at isotropic resolution. An initial validation showing the potential of the ZTE approach at 7T for is shown in volunteers and tumour patients.

Yuhui Chai¹, Jingwei Sheng¹, Bing Wu², Yang Fan², and Jia-Hong Gao^1
1Center for MRI Research, Peking University, Beijing, China, People's Republic of, ²GE Healthcare, MR Research China, Beijing, China, People's Republic of

In-vivo detection of oscillatory magnetic field may lead to fundamental development of functional MR imaging. The recently developed spin-lock oscillatory excitation (SLOE) method exhibited sub-nanotesla level of sensitivity. However its application in vivo is troubled by main field inhomogeneity. In this work, an oscillatory-selective detection (OSD) is proposed to overcome this limitation and hence to improve sensitivity of in-vivo detection of oscillatory magnetic field. OSD has also been verified as a viable tool in mapping transcranial alternating current.

Mengye Lyu^1,2, Victor B. Xie^1,2, Patrick G. Peng^1,2, Edward Hui³, and Ed X. Wu^1
1Laboratory of Biomedical Imaging and Signal Processing, The University of Hong Kong, Hong Kong SAR, China, People's Republic of, ²Department of Electrical and Electronic Engineering, The University of Hong Kong, Hong Kong SAR, China, People's Republic of, ³Department of Diagnostic Radiology, The University of Hong Kong, Hong Kong SAR, China, People's Republic of

We propose a method to simultaneously acquire spin echo and gradient echo and form two images with distinct contrasts. The spin echo and gradient echo are created by phase-cycled RF pulses, so that they are shifted from each other in image space. In reconstruction, coil sensitivity information can be used to separate them. This study demonstrates the feasibility of extracting multiple echo components using controlled aliasing and parallel imaging reconstruction.

Congbo Cai¹, Yuchuan Zhuang², Shuhui Cai³, Jianhui Zhong², and Zhong Chen^3
1Department of Communication Engineering, Xiamen University, Xiamen, China, People's Republic of, ²Department of Imaging Sciences, University of Rochester, ROCHESTER, NY, United States,³Department of Electronics Science, Xiamen University, Xiamen, China, People's Republic of

Magnetic Resonance parameters mapping can provide useful quantitative information for characterization of tissue properties. However, the long acquisition time usually hinder the real-time MR parameter mapping. In this abstract, a novel single-shot T2 mapping method was proposed based on spin-echo EPI method. Two overlapping echo signals with the different T2 weighting were obtained simultaneously. A detachment algorithm based on joint sparsity constraint was proposed to separate the two echo signals. The robustness and efficiency of the sequence were demonstrated through phantom experiments. The reliable T2 mapping can be obtained in the order of milliseconds.

Yun Jiang¹, Jesse I. Hamilton¹, Katherine L. Wright², Dan Ma², Nicole Seiberlich¹, Vikas Gulani^1,2, and Mark A. Griswold^1,2
1Department of Biomedical Engineering, Case Western Reserve University, Clevleand, OH, United States, ²Department of Radiology, Case Western Reserve University, Clevleand, OH, United States

The purpose of this study is to develop a method for simultaneous quantification of T₁, T₂, and diffusion within the MR Fingerprinting (MRF) framework. Multiple diffusion-weighted driven-equilibrium modules are inserted into the MRF-FISP acquisition in this study. The results from the prostate show the promising ability of this combination of MRF-FISP and diffusion preparation modules to quantify relaxation parameters along with diffusion in one scan.

Xiao Chen¹, Christopher C. Cline^1,2, Boris Mailhe¹, Qiu Wang¹, and Mariappan S. Nadar^1
1Medical Imaging Technologies, Siemens Healthcare, Princeton, NJ, United States, ²Biomedical Engineering, University of Minnesota, Minneapolis, MN, United States

In MRF, a single image is reconstructed from data collected from each short TR due to the non-repeatable magnetization history. The highly-undersampled single-shot imaging leads to high levels of noise and artifacts. In this study, an SR preparation module was introduced to MRF, enabling multi-shot MRF without a waiting time for magnetization recovery. The SR prepared multi-shot MRF can achieve similar or even better accuracy than the original single-shot IR prepared MRF with the same amount of data collected.

Eric Van Reeth¹, Hélène Ratiney¹, Michael Tesch², Steffen Glaser², and Dominique Sugny^3
1CREATIS, Université de Lyon ; CNRS UMR5220 ; Inserm U1044 ; INSA-Lyon ; Université Claude Bernard Lyon 1, Lyon, France, ²Department of Chemistry, Technische Universität München, Munich, Germany, ³Laboratoire Interdisciplinaire Carnot de Bourgogne (ICB), UMR 5209 CNRS-Université de Bourgogne, Dijon, France

Magnetic Resonance Imaging (MRI) uses the difference in tissue relaxation times to create contrast. Various image weightings can be obtained by tuning acquisition parameters which are usually empirically defined. In this article, optimal control theory is used to design excitation pulses that produce the optimal contrast between given tissues. The designed pulses are tested on numerical phantoms with and without magnetic field inhomogeneities and for the first time in vitro on a small-animal MRI. The reasonable match between simulation and real experiments is promising for the development of such pulses in further in vivo applications.

Kwan-Jin Jung¹, Andrea Willhite², and Susan Harkema^2
1Radiology, University of Louisville, Louisville, KY, United States, ²Neurosurgery, University of Louisville, Louisville, KY, United States

The phase offset in the phase contrast MR flow imaging was corrected using an image-based method in order to account for the spatially inhomogeneous and subject-dependent phase offset. The phase shift on the flow region was estimated iteratively from the phase shift of the stationary tissue using the low spatial distribution of the phase offset. This phase offset correction method with an automated segmentation and iterative estimation of the phase offset allowed us to study the cerebrospinal fluid flow in the spinal subarachnoid space of ten healthy and ten spinal cord injury participants reliably without elaborate manual effort.

Kyong Hwan Jin¹, Dongwook Lee¹, Paul Kyu Han¹, Juyoung Lee¹, Sung-Hong Park¹, and Jong Chul Ye^1
1Dept. of Bio and Brain Engineering, KAIST, Daejeon, Korea, Republic of

In this paper, we propose a sparse and low-rank decomposition of annihilating filter-based Hankel matrix for removing MR artifacts such as motion, RF noises, or herringbone artifacts. Based on the observation that some MR artifacts are originated from k-space outliers, we employ a recently proposed image modeling method using annihilating filter-based low-rank Hankel matrix approach (ALOHA) to decompose the sparse outliers from the low-rank component. The proposed approach can be applied even for static images, because the k-space low rank component comes from the intrinsic image properties. We demonstrate that the proposed algorithm clearly removes several types of artifacts such as impulse noises, motion artifacts, and herringbone artifacts.

Uten Yarach¹, Daniel Stucht¹, Hendrik Mattern¹, Frank Godenschweger¹, and Oliver Speck^1
1Department of Biomedical Magnetic Resonance, Otto-von-Guericke University Magdeburg, Magderburg, Germany

Patient motion during an MRI of the brain can result in non-diagnostic image quality. Even with perfect prospective motion (PMC) tracking and correction, the varying coil sensitivity, gradient non-linearity, and B0 field shift can cause significant artifacts that cannot be corrected prospectively. Recently, a model-based MR image reconstruction via iterative solver was employed to minimize the sensitivity misalignment due to physiological movement. In this study, we extended the mentioned model by gradient-warped and B0-induced distortion corrections to reconstruct the PMC MR data. The result shows the improvement that is remarkably reduced artifacts after a few iterations of the proposed technique.

Jiazheng Wang¹, Bing Wu², and Yongchuan Lai^3
1Radiology Department, University of Cambridge, Cambridge, United Kingdom, ²GE healthcare MR Research China, Beijing, China, People's Republic of, ³GE heathcare China, Beijing, China, People's Republic of

Usually aper-volume or even per-slice basis reference scan is usually needed for correcting the phase inconsistency between odd and even shots attributed to eddy current in EPI. The phase correction is usually first order. In this work, the concept of multiplxed SENSE (MUSE) is extended for EPI N/2 ghost calibration. 

Andreia S Gaspar^1,2, Giulio Ferrazzi¹, Rita G Nunes^1,2, Emer J Hughes^3,4, Shaihan J Malik¹, Laura McCabe^3,4, Kelly Pegoretti^3,4, Mary A Rutherford^3,4, Joseph V Hajnal^1,4, and Anthony N Price^1,4
1Biomedical Engineering, King's College London, London, United Kingdom, ²Instituto de Biofísica e Engenharia Biomédica, Faculdade de Ciências, Universidade de Lisboa, Lisbon, Portugal, ³Perinatal Imaging and Health, King's College London, London, United Kingdom, ⁴Centre for the Developing Brain, King's College London, London, United Kingdom

Effective suppression of maternal fat is critical for functional imaging of the fetal brain with echo planar imaging (EPI). Localized image-based shimming (IBS) for the fetal brain is required but can provoke high field variation in maternal adipose regions causing fat suppression to fail.  We have addressed this issue by using IBS of the fetal brain with linear constraints across maternal fat regions and optimization of saturation pulse frequency offset. The results showed that is possible to obtain more complete fat supression when combining an optimized pulse offset with a constrained shimming approach without compromising fetal brain shim.

Xueqing Liu¹, Zhihao Li², Shiyang Chen², and Xiaoping Hu^2
1Department of Biomedical Engineering, Columbia University, New York, NY, United States, ²Coulter Department of Biomedical Engineering, Georgia Tech and Emory University, Atlanta, Georgia

We describe a novel global signal removal method, sparse parameter global signal regression (SP-GSR), for fMRI data preprocessing. We assume the global signal to be low-rank and the remaining signal can be decomposed into orthogonal regressors with spatially sparse parameters. We demonstrated by simulation that SP-GSR can remove global signal and recovery true correlations without introducing anti-correlations. Application of this method to experimental data led to a more prominent and focused default mode network with isolated negative correlations.

Hyun-Soo Lee¹, Seung Hong Choi², and Sung-Hong Park^1
1Department of Bio and Brain Engineering, Korea Advanced Institute of Science and Technology, Daejeon, Korea, Republic of, ²Department of Radiology, Seoul National University College of Medicine, Seoul, Korea, Republic of

The quality of balanced steady-state free precession is vulnerable to eddy-currents and transient oscillations. However, the conventional centric phase-encoding (PE) scheme makes these artifacts severe, thus needs additional compensation strategies. In this study, we propose an improved PE scheme where k-space is encoded from center to periphery in a group-wise manner (PE-grouping). This reduces related artifacts by preventing big jumps in k-space along PE direction. Also proposed were various averaging strategies that could further eliminate the residual artifacts by averaging two full images acquired not only with the PE-grouping, but also with the conventional centric and pairing schemes.

JaeKyun Ryu^1,2, Joonsung Lee^1,2, Seong-gi Kim^1,2, and Jang-Yeon Park^1,2
1Center for Neuroscience Imaging Research, Institute for Basic Science, Suwon, Korea, Republic of, ²Department of Biomedical Engineering, Sungkyunkwan University, Suwon, Korea, Republic of

RASER(Rapid Acquisition by Sequential Excitation and Refocusing) sequence acquires all the echoes with same TE by using two refocusing pulses [1], whereas other spatiotemporal-encoding(SPEN) techniques using a single-refocusing pulse offers a shorter effective TE than RASER but with varying TE, thereby causing signal variation in the SPEN dimension [2]. Here, we propose an effective way of overcoming this problem of TE variation in single-refocusing SPEN imaging.

Paul T. Weavers¹, Shengzhen Tao¹, Kiaran McGee¹, Joshua Trzasko¹, Yunhong Shu¹, Erik Tryggestad², Ken-Pin Hwang³, Seung-Kyun Lee⁴, Thomas KF Foo⁴, and Matt Bernstein^1
1Mayo Clinic, Rochester, MN, United States, ²Radiation Oncology, Mayo Clinic, Rochester, MN, United States, ³MD Anderson, Houston, TX, United States, ⁴GE Global Research, Niskayuna, NY, United States

Radiation therapy, especially proton beam therapy requires exacting spatial accuracy to deliver a sterilizing dose of ionizing radiation to the target volume with confidence.  The superior soft tissue contrast afforded by MRI vs. CT has increased interest in using MRI for treatment planning.  However, gradient non-linearities reduce the spatial accuracy of MRI.  We have developed a fiducial phantom based calibration procedure to map these gradient nonlinearities on a system-specific basis and generate up to 9th order spherical harmonic coefficients for correction.  These coefficients show improved spatial accuracy vs. standard 5th order, especially at distances >400mm from magnet and gradient isocenter.

Julianna D. Ianni^1,2 and William A. Grissom^1,2,3,4
1Department of Biomedical Engineering, Vanderbilt University, Nashville, TN, United States, ²Vanderbilt University Institute of Imaging Science, Vanderbilt University, Nashville, TN, United States,³Department of Radiology and Radiological Sciences, Vanderbilt University, Nashville, TN, United States, ⁴Department of Electrical Engineering, Vanderbilt University, Nashville, TN, United States

A method for automatic correction of EPI trajectory errors is presented.  The method is an iterative parallel imaging reconstruction which uses k-space data consistency to correct image artifacts. An advantage of the method is that it requires no calibration data and allows correction of non-static errors such as those due to gradient coil heating.

Matthew Robert Smith¹, Jin Jung Kweon², Eun Sang Choi², Curtis Wiens¹, Nathan Artz³, and Scott B Reeder^1,4
1Radiology, University of Wisconsin, Madison, WI, United States, ²Florida State University, Tallahassee, FL, United States, ³St. Jude's Children's Hospital, Memphis, TN, United States, ⁴Medical Physics, University of Wisconsin, Madison, WI, United States

Despite important recent development, further progress of MR imaging around metallic prostheses is dependent on the ability to model the field perturbations surrounding the prostheses. These calculations require knowledge of the magnetic susceptibility of the metal which is not reported by the manufacturers. The purpose of this work was to estimate the magnetic susceptibility of commonly implanted metal alloys by measuring the magnetic moment across a range of clinical field strengths (0-5 Tesla) using a SQUID (Superconducting QUantum Interference Device) magnetometer. Linearity of the susceptibility across field strengths was also assessed.

Daniel Daniel Hernandez¹, Eric Michel¹, Ki Soo Kim¹, and Soo Yeol Lee^1
1Bio-medical engineering, Kyung Hee University, Yogin, Korea, Republic of

We propose a method to correct phase drift artifact from MR thermometry measurements with the use of Magnetic field monitoring (MFM). Field variations are measured with an array of MFM probes and correction maps are computed from the frequency shifts of FID signals. This method allows to have better resolution and can be used with any MR thermometry pulse sequence.

Ying Chen¹, Song Chen¹, Hui Liu², Zhong Chen³, and Jianhui Zhong^1,4
1Center for Brain Imaging Science and Technology, Zhejiang University, Hangzhou, China, People's Republic of, ²MR Collaboration Northeast Asia, Siemens Healthcare, Shanghai, China, People's Republic of, ³Department of Electronic Science, Xiamen University, Xiamen, China, People's Republic of, ⁴Collaborative Innovation Center for Diagnosis and Treatment of Infectious Diseases, Zhejiang University, Hangzhou, China, People's Republic of

Single-shot SPEN MRI is a technique capable of retaining the time efficiency of single-shot EPI but with significantly reduced geometric distortions. Akin to EPI, the phase inconsistency between even and odd echoes also result in Nyquist ghosts in SPEN images. This work is to present a scheme with more reliable performance than the previously reported Nyquist ghost correction method. Experimental results of human brains and in vivo rats show that the proposed method can remove ghosts without introducing blurring, and unwarping procedures can be conducted on the ghost-corrected data for further distortion correction.

Mark BYDDER¹, Wafaa Zaaraoui¹, and Jean-Philippe Ranjeva^1
1Aix Marseille Université, MARSEILLE, France

Several algorithms for choosing the radial spoke directions were evaluated for use in 3D UTE imaging. It was observed that methods that produce a highly regular set of directions result in higher aliasing than those that are more irregular.

Tilman Schubert¹, Peter Bannas², Samir Sharma³, Sonja Kinner¹, Mahdi Rahimi³, Frank Korosec³, and Scott Reeder^1
1Radiology, University of Wisconsin Madison, Madison, WI, United States, ²Radiology, University Medical Center Hamburg-Eppendorf, Hamburg, Germany, ³Medical Physics, University of Wisconsin Madison, Madison, WI, United States

Chemical shift based two-point “Dixon” MRI with bipolar readout gradients may produce flow induced fat-water misallocation artifacts. These artifacts have the potential to mimic intravascular thrombus. We reviewed 100 cases of two-point body MRI exams to characterize the incidence, location and severity of these artifacts. Artifacts appeared in 46% of the cases, with  severe artifacts in 20% and mild artifacts in 26% of the patients. Given this high number of potentially thrombus-mimicking, flow-induced artifacts, radiologists should be aware of this potential pitfall when using two-point fat-water separation methods.

Francesco Santini^1,2, Mathieu D Santin³, Paulo Loureiro de Sousa⁴, and Oliver Bieri^1,2
1Radiological Physics, University of Basel Hospital, Basel, Switzerland, ²Department of Biomedical Engineering, University of Basel, Basel, Switzerland, ³Institut du Cerveau et de la Moelle épinière, Hôpital Pitié-Salpêtrière, Paris, France, ⁴CNRS, ICube Laboratory, FMTS, Université de Strasbourg, Strasbourg, France

This work presents a method to combine the signal from multiple coils in order to obtain a coherent phase image. The methods is agnostic to the acquisition protocol and the coil geometry, and does not require operator interaction.

Hiroshi Toyoda¹, Sosuke Yoshinaga², Naoya Yuzuriha², and Hiroaki Terasawa^2
1CiNet, NICT, Suita, Japan, ²Department of Structural BioImaging, Kumamoto University, Kumamoto, Japan

We proposed an Image-based phase correction for dual-band EPI with slice-GRAPPA using point-by-point procedures in k-space. The results showed the usefulness and robustness of the proposed method compared with the conventional approach.

Xiaowei Zhuang¹, Virendra Mishra¹, Karthik Sreenivasan¹, Charles Bernick¹, Sarah Banks¹, and Dietmar Cordes^1,2
1Cleveland Clinic Lou Ruvo Center for Brain Health, Las Vegas, NV, United States, ²Department of Psychology and Neuroscience, University of Colorado Boulder, Boulder, CO, United States

Clinical applications of brain abnormality detection with supervised machine learning techniques are limited due to less and unbalanced sample sizes as compared to rich feature sets in patient population. We proposed a new combinatorial model approach, fs-RBFN, involving sampling from multivariate joint distribution, LASSO feature selection, RBFN cross validation, and inverse probability weighting to solve this problem. The proposed approach was validated against a ground truth phantom and further tested on a multimodal MRI dataset for cognitively impaired and non-impaired professional fighters. Our results suggest superior performance of this technique over several other out-of-the-bag feature selection algorithms.

Kevin K. Zhang^1,2, Shivani Kumar^2,3, Robba Rai³, Armia George³, Bin Dong^1,4, and Gary P. Liney^1,2,3,4
1Ingham Institute for Applied Medical Research, Sydney, Australia, ²South Western Sydney Clinical School, University of New South Wales, Sydney, Australia, ³Department of Medical Physics, Cancer Therapy Centre, Liverpool Hospital, Sydney, Australia, ⁴Centre for Medical Radiation Physics (CMPR), University of Wollongong, Sydney, Australia

Real-time lung tumour tracking and motion analysis is important in MRI-based radiotherapy planning to inform treatment margins and to permit accurate delivery for developing MR-Linac technology. This work describes a template matching approach to provide 3D motion assessment of lung tumours from real-time 2D images. Compared to previous work the TRAC technique utilises a multi-angled correlation analysis of the target region to correctly identify the tumour position. Results in both a moving phantom and in lung cancer patients show that the technique is feasible, accurate and can be easily adopted in widely used single plane cine imaging.

Nicole Wake¹, Jeremy C Lim², Artem Mikheev¹, Jas-mine Seah³, Elissa Botterill², Shawna Farquharson⁴, Henry Rusinek¹, and Ruth P Lim^2,5
1Bernard and Irene Schwartz Center for Biomedical Imaging, Center for Advanced Imaging Innovation and Research, Department of Radiology, New York University School of Medicine, New York, NY, United States, ²Department of Radiology, Austin Health, Melbourne, Australia, ³Department of Endocrinology, Austin Health, Melbourne, Australia, ⁴Florey Neuroscience Institute, Melbourne, Australia,⁵The University of Melbourne, Melbourne, Australia

A semi-automatic renal segmentation technique for non-contrast T1-weighted MR images was developed.  Renal segmentation and volumetric analysis was tested in ten healthy volunteers and ten Type I diabetic patients. We found that this segmentation tool is fast, reliable, and requires minimal user interaction. Upon further validation, this method has clinical potential for monitoring renal status in appropriate patient populations.

Jan L Bruse¹, Abbas Khushnood¹, Tain-Yen Hsia¹, Andrew M Taylor¹, Vivek Muthurangu¹, and Silvia Schievano^1
1Centre for Cardiovascular Imaging, UCL Institute of Cardiovascular Science & Great Ormond Street Hospital for Children, London, United Kingdom

We present a novel method for hierarchical 3D shape clustering of aortic arch shape models segmented and reconstructed from CMR imaging data. We apply the method to a cohort of 45 patients post aortic coarctation repair in order to explore previously unknown arch shape patterns that may relate to clinical outcome. Exploring a pathologic shape population using data mining and statistical shape modeling techniques can provide novel insight for improved diagnosis and treatment strategies and can thereby assisst in clinical decision making when analysing complex cases.

Soumya Ghose¹, Jason Dowling¹, Robba Rai², Benjamin Schmitt³, and Gary Liney^2,4,5,6
1eHealth, CSIRO, Brisbane, Australia, ²Liverpool Cancer Therapy Centre, Liverpool, Australia, ³Siemens Healthcare Pty Ltd, Macquarie Park, Australia, ⁴Medical Physics, Ingham Institute, Liverpool, Australia, ⁵UNSW Australia, Liverpool, Australia, ⁶University of Wollongong, Wollongong, Australia

Accurate conversion of MRI into attenuation correction maps is of current interest in PET-MR and MR-only radiotherapy planning in particular, where electron density calculation is particularly demanding and usually derived from CT. MRI methods to date have usually involved building a patient atlas and/or use of multiple imaging sequences and are time intensive. We propose a new single sequence approach based on ultra-short echo time to identify tissue classes of air, bone and soft-tissue in combination with a dynamic clustering regression based model that provides a direct CT conversion which is both efficient and accurate. 

Mohammad Mehdi Khalighi¹, Audrey Peiwen Fan², Gaspar Delso³, Praveen K. Gulaka², Bin Shen⁴, Aileen Hoehne⁴, Prachi Singh², Jun-Hyung Park⁴, Dawn Holley², Frederick T. Chin^2,4, and Greg Zaharchuk^2,4
1Applied Science Lab, GE Healthcare, Menlo Park, CA, United States, ²Radiology Department, Stanford University, Stanford, CA, United States, ³Applied Science Lab, GE Healthcare, Zurich, Switzerland,⁴Molecular Imaging Program, Stanford University, Stanford, CA, United States

Accurate measurement of Arterial Input Function (AIF) is essential in quantitative analysis of cerebral blood flow (CBF) using ¹⁵O-H₂O PET imaging. The time-of-flight enabled Signa PET/MR scanner (GE Healthcare, Waukesha, WI, USA) provides quality PET images during the arrival of ¹⁵O-H₂O tracer to the brain arteries, which can be used for carotid artery segmentation. The optimal time frame to segment these brain arteries for image-based AIF, is found by binning the PET list file every second and plotting the total number of true and scatter coincident events over time.

Phillip G. D. Ward^1,2, Audrey P. Fan³, Parnesh Raniga¹, David G. Barnes^2,4, David L. Dowe², and Gary F. Egan^1,5
1Monash Biomedical Imaging, Monash University, Clayton, Australia, ²Faculty of Information Technology, Monash University, Clayton, Australia, ³Lucas Center for Imaging, Department of Radiology, Stanford University, Stanford, CA, United States, ⁴Monash eResearch Centre, Monash University, Clayton, Australia, ⁵ARC Centre of Excellence for Integrative Brain Function, Melbourne, Australia

Partial volume effects impede the use of quantitative susceptibility maps for assessing small veins. Oxygen extraction fraction measures are particularly sensitive to these effects. We propose a geometric technique for calculating partial volume from binary venograms. The technique is able to calculate accurate partial volume maps, and vessel geometry, on simulated veins of sub-voxel radius. These partial volume maps are used to adjust for partial volume effects in estimating venous magnetic susceptibility.

Jean-Christophe Brisset¹, Louise E Pape¹, Ricardo Otazo¹, and Yulin Ge^1
1Radiology, New York University School of Medicine, New York, NY, United States

Since human gray matter cortex is a relatively thin structure and has a complex folding pattern blended with white matter and cerebrospinal fluid (CSF), partial volume effect is always considered a challenging issue for precise tissue segmentation. Super-resolution (SR) is a common method that is often used in the picture world to recover a high-resolution image from low-resolution images. This study was performed to test whether a newly developed sparsity-guided SR algorithm can be adapted on standard clinical MRI images to improve brain tissue segmentation by decreasing partial volume effect.

Allison B Randolph V¹, Valentina Pedoia¹, Lorenzo Nardo¹, and Sharmila Majumdar^1
1Radiology & Biomedical Imaging, UCSF, San Francisco, CA, United States

Voxel-based relaxometry (VBR) allows for MR relaxtion time analysis without the sometimes deletorious assumtions of traditional ROIs. However, VBR introduces potentially new analysis issues, such as noise and map heterogeneity. In this study we propose to use VBR significance thresholding in conjunction with cluster-extent based thresholding to define data-driven  regions of interest (ROIs) that include the most critical information in Statistical Parametric Maps (SPM), controlling the aforesaid issues. The results suggests that the data driven voxel cluster ROIs and predefined traditional ROIs have unique, separate anatomical locations, and that the data-driven clusters perform better when correlated to osteoarthritis (OA) disease markers. 

Jun Shen¹, Thomas Baum², Christian Cordes², Beate Ott³, Claudia Eichhorn³, Thomas Skurk^3,4, Hendrik Kooijman⁵, Ernst J Rummeny², Hans Hauner^3,4, Bjoern H Menze¹, and Dimitrios C Karampinos^2
1Department of Computer Science, TU Munich, Munich, Germany, ²Department of Radiology, TU Munich, Munich, Germany, ³Else Kröner Fresenius Center for Nutritional Medicine, TU Munich, Munich, Germany, ⁴ZIEL Research Center for Nutrition and Food Sciences, TU Munich, Munich, Germany, ⁵Philips Healthcare, Hamburg, Germany

The accumulation and regional distribution of abdominal adipose tissue and organ fat plays an important role in several diseases including obesity, metabolic syndrome and diabetes. The present work proposes a fully automatic method for abdominal organ segmentation and adipose tissue classification and measurement based on chemical shift encoding-based water-fat MR images. The results from the automatic method showed very good agreement with the manually created references. The developed automatic algorithm allowed the detection of regional differences in changes of adipose tissue depots in a study of 20 obese women undergoing a calorie restriction intervention.

Leili Riazy¹, Simone Fritzschi^2,3, Arthur Stötzner^2,3, Fabian Mühlberg^2,3, Luisa Schmacht^2,3, Matthias Dieringer^1,2,4, Florian von Knobelsdorff-Brenkenhoff^2,3, Thoralf Niendorf^1,2, and Jeanette Schulz-Menger^2,3
1Berlin Ultrahigh Field Facility (B.U.F.F.), Max-Delbrueck Center for Molecular Medicine, Berlin, Germany, ²Working Group on Cardiovascular Magnetic Resonance, Experimental and Clinical Research Center (ECRC), Berlin, Germany, ³Department of Cardiology and Nephrology, HELIOS Klinikum Berlin Buch, Berlin, Germany, ⁴Siemens Healthcare GmbH, Erlangen, Germany

Late Gadolinium Enhancement (LGE) is the noninvasive gold standard for focal fibrosis, parametric mapping with and without contrast-media enable detection of diffuse fibrosis. We developed a non-rigid registration method to superimpose LGE images and T1-Maps allowing for pixel-wise comparison of LGE extent and abnormal T1 times. We observed significantly larger regions of ECV, T1 native and post-contrast abnormalities than LGE positive areas. However, LGE was not always completely covered by abnormalities of any of the mentioned parameters.

Marie Anne Richard¹, Réjean Lebel¹, Jérémie P. Fouquet¹, and Martin Lepage^1
1Centre d'imagerie moléculaire de Sherbrooke, Université de Shebrooke, Sherbrooke, QC, Canada

Positron emission tomography (PET) images suffer from statistical noise, especially in short time frames. For applications requiring high temporal resolution, such as dynamic studies, efficient edge-preserving denoising algorithms such as the non-local means filter (NLMF) are needed. Because this filter relies on structural data, coregistered MR images were used to guide the NLMF. This novel method was compared to conventional PET-guided NLMF and proved superior in terms of increased contrast-to-background ratio and signal-to-noise ratio in a phantom model. It also increased small structure resolution in a rat model.

Jason Su¹, Thomas Tourdias², Manojkumar Saranathan³, Pejman Ghanouni⁴, and Brian Rutt^4
1Electrical Engineering, Stanford University, Stanford, CA, United States, ²Neuroradiology, Bordeaux University Hospital, Bordeaux, France, ³Radiology, University of Arizona, Tucson, AZ, United States,⁴Radiology, Stanford University, Stanford, CA, United States

The efficacy of the Thalamus Optimized Multi-Atlas Segmentation (THOMAS) algorithm for segmentation of thalamic nuclei with white-matter-nulled MP-RAGE images is studied in 3T and 7T variants of the image contrast. 5 subjects are evaluated at both field strengths and ground truth manual delineations of nuclei are performed on the 7T images. We demonstrate that the algorithm performs as well on 3T images as on 7T within a dice coefficient of ±0.1 as evaluated against the ground truth. This indicates that THOMAS can now reach a much wider audience of interested groups.

Adam K Featherstone^1,2, James P B O'Connor^2,3, Ross A Little¹, Yvonne Watson¹, Sue Cheung¹, Kaye J Williams^2,4, Julian C Matthews^1,2, and Geoff J M Parker^1,2,5
1Centre for Imaging Sciences, The University of Manchester, Manchester, United Kingdom, ²CRUK & EPSRC Cancer Imaging Centre in Cambridge and Manchester, Cambridge and Manchester, United Kingdom, ³Institute of Cancer Sciences, The University of Manchester, Manchester, United Kingdom, ⁴School of Pharmacy, The University of Manchester, Manchester, United Kingdom, ⁵Bioxydyn Ltd., Manchester, United Kingdom

DCE-MRI and OE-MRI scans were performed on 8 preclinical U87 tumour xenografts. Heuristic features (area-under-curve and rate-of-enhancement) were calculated from tumour voxel enhancement curves for each imaging modality. Clustering algorithms (k-means clustering and Gaussian mixture modelling) were applied to these features and native tissue T₁ to investigate their utility in characterising physiological heterogeneity in tumours. Efficacy in identifying large regions where there is agreement between features is shown. Further optimisation is needed to optimise the approach to characterise smaller, and potentially important, regions where there is a lack of concordance between features.

 

Ferran Prados^1,2, Bhavana S Solanky², Patricia Alves Da Mota², Manuel Jorge Cardoso¹, Wallace J Brownlee², Niamh Cawley², David H Miller², Xavier Golay³, Sebastien Ourselin¹, and Claudia Angela Michela Gandini Wheeler-Kingshott^2,4
1Translational Imaging Group, Medical Physics and Biomedical Engineering, University College London, London, United Kingdom, ²NMR Research Unit, Queen Square MS Centre, Department of Neuroinflammation, UCL Institute of Neurology, University College London, London, United Kingdom, ³Brain Repair & Rehabilitation, UCL Institute of Neurology, University College London, London, United Kingdom, ⁴Brain Connectivity Center, C. Mondino National Neurological Institute, Pavia, Italy

Quantitative sodium magnetic resonance imaging (²³Na-MRI) enables the non-invasive measurement of in vivo total ²³Na concentration (TSC) in the human brain. This involves a complex process of reconstructing datasets acquired to calculate a TSC map. Quantitative TSC map calibration relies on external reference phantoms with known concentration for linear calibration. This commonly involves manually segmenting the phantoms by trained raters, hindering automatic image analysis, and presenting a bottleneck in the TSC computation. We propose to substitute the manual segmentation by OPAL, a novel, fast, robust and reliable technique for segmenting sodium phantoms that allows fully-automatic reconstruction of TSC maps.

Mingze Xu^1,2, Shiyang Chen², Bing Ji^2,3, Jiuquan Zhang⁴, Huaiqiu Zhu¹, Yi Zhang⁵, Yonggui Yuan⁶, Jiahong Gao¹, Yijun Liu¹, and Xiaoping Hu^2
1Biomedical Engineering, Peking University, Beijing, China, People's Republic of, ²Biomedical Engineering, Emory University & Georgia Institute of Technology, Atlanta, GA, United States, ³University of Shanghai for Science & Technology, Shanghai, China, People's Republic of, ⁴Department of Radiology, Southwest Hospital, Third Military Medical University, Chongqing, China, People's Republic of,⁵School of Life Science and Technology, Xidian University, Shaanxi, China, People's Republic of, ⁶Department of Psychosomatics and Psychiatry, ZhongDa Hospital, School of Medicine, Southeast University, Nanjing, China, People's Republic of

We conducted dynamic whole-brain connectivity analysis in Late-onset depression (LOD) to investigate alterations in brain networks. All subjects’ ROI-to-ROI dynamic FC were explored using a data-driven method to obtain the most explanatory states. Each state indicate a particular ROI-to-ROI FC pattern. The property of each state were determined based on its scores across time. Besides decreased FC in normal state, we found LOD patients switch between brain states more frequently and tend to enter LOD-risk states, due to and its high states variance and dominating increased FC in LOD-risk states. These results suggest neural mechanisms of disorder from dynamic perspective.

Shivaprasad Ashok Chikop¹, Vimal Chandran², Imam Shaik¹, Rashmi Rao¹, Mauricio Antonio Reyes Aguirre², and Sairam Geethanath^1
1Medical Imaging Research Center, Dayananda Sagar Institutions, Bangalore, India, ²Institute of Surgical Technology and Biomechanics, University of Bern, Bern, Switzerland

The step size of the parameters used for simulation of dictionary determines the parameters being determined. Partial Least squares (PLS) can be used as a general frame work for fast and robust dictionary matching. Regression co-efficient matrix obtained from PLS can be used for localizing the different brain tissue types thus avoiding iterative searching.  The increase in contrast between the grey matter and white matter can be attributed to the intermediate values generated by PLS based matching. PLS matches comparatively better at low SNR images compared to the straight forward dot product method.

Saba N Elias¹, Guang Jia², Firas G Petros³, Huyen Nguyen¹, Debra L Zynger⁴, Zarine K Shah⁵, Ronney Abaza⁶, and Michael V Knopp^1
1Radiology/Wright Center of Innovation, The Ohio State University, Columbus, OH, United States, ²Department of Physics & Astronomy, Louisiana State University, Baton Rouge, LA, United States,³Urology, The Ohio State University, Columbus, OH, United States, ⁴Pathology, The Ohio State University, Columbus, OH, United States, ⁵Radiology, The Ohio State University, Columbus, OH, United States, ⁶Robotic Urologic Surgery, OhioHealth Dublin Methodist Hospital, Dublin, OH, United States

Feasibility of classifying PCa into clusters based on microcirculatory features has the potential to predict outcome and assist in therapeutic treatment of PCa.

Dante PI Capaldi¹, Anthony Lausch², Khadija Sheikh¹, Fumin Guo¹, David G McCormack³, and Grace Parraga^1
1Robarts Research Institute, The University of Western Ontario, London, ON, Canada, ²Credit Valley Hospital, Mississauga, ON, Canada, ³Department of Medicine, The University of Western Ontario, London, ON, Canada

In this proof-of-concept demonstration, we developed and generated multimodal-parametric-response-mapping (mPRM) from CT and MRI pulmonary measurements to phenotype chronic obstructive pulmonary disease (COPD).  We performed principal component analysis of the voxel distribution generated from co-registered inspiration or expiratory CT with  ³He MRI SV cluster maps and ³He MRI ADC maps for ex-smokers with and without COPD.  Further work is necessary to determine the appropriate combination of imaging biomarkers generated from MRI and CT to provide useful information in deeply phenotyping COPD.

Tiago Constantino^1,2,3, André Santos Ribeiro⁴, Ricardo Maximiano³, John Mcgonigle⁴, David Nutt⁴, and Hugo Alexandre Ferreira^3
1Lisbon School of Health Technology-ESTeSL, Lisbon, Portugal, ²Spitalzentrum Biel, Biel, Switzerland, ³Institute of Biophysics and Biomedical Engineering, Faculty of Sciences of the University of Lisbon, Lisbon, Portugal, ⁴Centre for Neuropsychopharmacology, Imperial College London, London, United Kingdom

In this work we propose a comparison study between “Scans Without Evidence for Dopaminergic Deficit" (SWEDD) and Parkinson’s Disease (PD) patients against healthy subjects using the MIBCA toolbox. Here, we studied the difference in imaging and connectivity metrics obtained from anatomical (T1-weighted) and structural (Diffusion Tensor Imaging) data between the three groups. Results showed increased mean diffusivity in the frontal pole, rostral middle frontal gyrus and superior frontal gyrus between SWEDD and PD patients,  which can be related with the dopaminergic mesocortical pathway degeneration in PD. These preliminary results help clarify the differences between SWEDD and PD patients.

Kees M. van Hespen¹, Dirk H.J. Poot^1,2, Harm A. Nieuwstadt¹, and Stefan Klein^1
1Departments of Medical Informatics and Radiology, Erasmus MC, Rotterdam, Netherlands, ²Imaging Science and Technology, Delft University of Technology, Delft, Netherlands

We have recently developed an optimized T₁ mapping protocol for carotid atherosclerotic plaque imaging, using a combination of inversion and recovery prepared acquisitions. This protocol requires less images to be taken (and thus shorter acquisition time) for precise T₁ estimation than conventional inversion-prepared or saturation-prepared acquisition schemes. However, estimating T₁ from magnitude data, acquired with the optimized settings, causes bimodality of T₁ estimates, due to the ambiguity in sign of the inversion prepared magnitude images. Simulations and experiments on a hardware phantom and a volunteer show that the ambiguity resolves when we fit a complex-valued model to the complex data.

Tim Luo¹, Shrushrita Sharma², Mark Polivchuk³, Peng Zhai⁴, and Yunyan Zhang^4
1Bachelor of Health Sciences, University of Calgary, Calgary, AB, Canada, ²Biomedical Engineering Program, University of Calgary, Calgary, AB, Canada, ³Computer Science, University of Calgary, Calgary, AB, Canada, ⁴Radiology and Clinical Neurosciences, University of Calgary, Calgary, AB, Canada

Changes in myelin integrity are associated with many neurological diseases. We acquired 9.4T MRI from healthy mouse brain to evaluate the utility of texture heterogeneity in T2-weighted MRI for assessing myelin integrity, in comparing with proposed measures including magnetic transfer ratio and myelin water fraction. Measurements were focused on the corpus callosum, with both anatomical (genu, body, splenium) and hemispheric (left, center, right) locations evaluated. All 3 methods showed the uniformity of myelin in corpus callosum between hemispheres, and no significant differences between anatomical locations were detected. Texture heterogeneity showed the best consistency between animals and deserves further verification.

Jan-Gerd Tenberge^1
1University of Münster, Münster, Germany

Some imaging software packages do not accurately display datasets due to difficulties in color mapping. We  show some of the shortcomings an three of the most widely used tools (FSL, SPM, FreeSurfer) and provide an easy fix that can be applied to correct the images output by these tools.

Mustapha Bouhrara¹ and Richard G. Spencer^1
1NIA, NIH, Baltimore, MD, United States

We introduce two Bayesian-based analyses that use spatial information as a prior to improve the quality of voxel-by-voxel T₁-mapping from spoiled gradient recalled echo (SPGR) imaging data. These approaches, called BASIC, combine voxel-by-voxel fitting with region-of-interest (ROI) parameter estimation. ROI parameters act as a constraint, while voxel fitting mitigates blurring and detail loss. The results were compared with those derived using a conventional nonlinear least-squares-based algorithm. Estimation of T₁ from SPGR imaging data was markedly improved through use of the BASIC methods.