Yilong Liu^1,2, Kun Zhou³, Dehe Weng³, Hua Guo⁴, and Ed X. Wu^1,2
1Laboratory of Biomedical Imaging and Signal Processing, The University of Hong Kong, Hong Kong, China, ²Department of Electrical and Electronic Engineering, The University of Hong Kong, Hong Kong, China, ³Siemens Shenzhen Magnetic Resonance Ltd, Shenzhen, China, ⁴Center for Biomedical Imaging Research, Department of Biomedical Engineering, Tsinghua University, Beijing, China

This study presents a calibrationless parallel imaging (CPI) reconstruction for simultaneous multi-slice (SMS) PROPELLER MRI of the upper abdomen. With simultaneously excited slices having different blipped-CAIPI shifts, the inherent incoherency of SMS PROPELLER data enables CPI reconstruction via low rank matrix approximation. The proposed method was evaluated with both simulated phantom and acquired abdominal MR data. Compared to conventional split slice-GRAPPA, the proposed method jointly reconstructs all blades, providing significantly improved SNR and reduced artifact level.

Zhilang Qiu^1,2, Sen Jia¹, Haifeng Wang¹, Lei Zhang¹, Xin Liu¹, Hairong Zheng¹, and Dong Liang^1
1Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen, China, ²University of Chinese Academy of Sciences, Beijing, China

Parallel imaging with wave encoded k-space trajectory can use most of the existing reconstruction methods such as SENSE, ESPIRiT and SPIRiT. However, these reconstruction methods are not suitable to reduced field-of-view (FOV) imaging scenario, where the imaging FOV is smaller than the object size, and severe aliasing artifacts remain in the reconstructed images. This work proposes to extend the existing reconstruction methods to treat the reduced FOV case, by accurately modeling and resolving the aliased components. Without increasing scan time, the artifact-free and full FOV images can be reconstructed by the proposed methods.

Gregory Lemberskiy¹, Jelle Veraart¹, Benjamin Ades-aron¹, Els Fieremans¹, and Dmitry S Novikov^1
1Radiology, NYU School of Medicine, New York, NY, United States

We propose a method of virtual coil compression using random matrix theory, MP-VCC, in which the Marchenko-Pastur distribution defines how many virtual coils may be discarded without loss beyond the PCA precision. MP-VCC is evaluated for partial Fourier, regular undersampling, and multiband acceleration. 

Fang Dong¹, Dehe Weng¹, and Nan Xiao^1
1Siemens Shenzhen Magnetic Resonance Ltd., Shen Zhen, China

PETRA sequence, with ultra-short TE and ultra-low acoustic noise, has wide applications such as dental imaging, lung imaging, bone imaging, pediatric imaging as well as the routine anatomical imaging. However, the long scanning time and the vulnerability to motion hampered its usage. In this study, the combining non-Cartesian parallel imaging and Compressed sensing technique is adapted for the hybrid trajectory PETRA sequence. The image quality is clinically adequate for 2-5 folds acceleration.

Charles Millard^1,2, Aaron T Hess², Jared Tanner¹, and Boris Mailhe^3
1Mathematical Institute, University of Oxford, Oxford, United Kingdom, ²Oxford Centre for Clinical Magnetic Resonance Research, University of Oxford, Oxford, United Kingdom, ³Digital Technology and Innovation, Siemens Healthineers, Princeton, NJ, United States

We present the Parallel Variable Density Approximate Message Passing (P-VDAMP) algorithm for compressed sensing MRI, which extends the recently proposed single-coil VDAMP algorithm to multiple coils. We evaluate the performance of P-VDAMP on eight datasets at a number of undersampling factors and find that it converges to a mean-squared error similar to the Fast Iterative Shrinkage Thresholding Algorithm (FISTA) with an optimally tuned sparse weighting, but in around 5x fewer iterations and without the need to tune model parameters.

Alexander Lin¹, Demba Ba¹, and Berkin Bilgic^2,3
1Harvard University, Cambridge, MA, United States, ²Department of Radiology, Massachusetts General Hospital, Martinos Center for Biomedical Imaging, Boston, MA, United States, ³Harvard Medical School, Boston, MA, United States

We propose Bayesian accelerated Compressed Sensing (BaCS) to improve the computational speed of Bayesian CS by two orders of magnitude. We achieve this by circumventing a costly matrix inversion problem using conjugate gradients and Monte Carlo sampling, which lend themselves well to parallel processing using GPUs. Exploiting parallelism renders BaCS even faster than sparseMRI, while having the ability to exploit similarities between multi-contrast images to improve reconstruction performance. Further, we extend BaCS to multi-channel reconstruction by synergistically combining it with SENSE to enable yet higher acceleration rates. 

Zhilang Qiu^1,2, Sen Jia¹, Shi Su¹, Yanjie Zhu¹, Xin Liu¹, Hairong Zheng¹, Haifeng Wang¹, and Dong Liang^1
1Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen, China, ²University of Chinese Academy of Sciences, Beijing, China

Wave encoding is less efficient in situations of high resolution and high bandwidth. In this work, a novel model (named as VCC-Wave) which combines virtual conjugate coil (VCC) and wave encoding (Wave) was proposed. It can not only combine both advantages of VCC and Wave, but also exploit more priors of Wave under the VCC framework. Further significant improvement is achieved, and the limitation of Wave in situations of high resolution and high bandwidth is alleviated.

Jae-Hun Lee¹, Jaeuk Yi¹, Kanghyun Ryu², Soozy Jung¹, and Dong-Hyun Kim^1
1Department of Electrical & Electronic Engineering, Yonsei Univ., Seoul, Korea, Republic of, ²Department of Radiology, Stanford Univ., Stanford, CA, United States

 Myelin water imaging using multi-cho GRE (mGRE) requires long acquisition times. In this study, we explored a multi-step reconstruction method using both advanced parallel imaging and deep learning, which can utilize joint information between the multi-echo images to further accelerate the acquisition. The proposed method shows acceptable image quality with improved quantitative values compared to conventional methods. The proposed method can achieve high acceleration factors for mGRE based 3D myelin water imaging.

Tzu-Cheng Chao¹ and James G. Pipe^1
1Department of Radiology, Mayo Clinic, Rochester, MN, United States

A temporal harmonic encoding scheme is proposed to improve T1W FLAIR imaging with TSE acquisition. The imaging method reduces blurring in T1W FLAIR and provides an extra T2W FLAIR contrast at the same time with no need of additional scan time. 

Michael Rawson¹, Xiaoke Wang², Ze Wang², Radu Balan^1,3, and Thomas Ernst^2
1Department of Mathematics, University of Maryland at College Park, College Park, MD, United States, ²Department of Diagnostic Radiology and Nuclear Medicine, University of Maryland School of Medicine, Baltimore, MD, United States, ³Center for Scientific Computation and Mathematical Modeling, University of Maryland at College Park, College Park, MD, United States

We introduce an approximation and resulting method called MGRAPPA to allow high speed MRI scans robust to subject motion using prospective motion correction and GRAPPA [1,2]. In experiments on both simulated data and in vivo data, we observe high accuracy and robustness to subject movement in L2 (Frobenius) norm error including a 41% improvement in the in vivo experiment. 

Michael Herbst^1
1Bruker BioSpin MRI GmbH, Ettlingen, Germany

Segmented EPI enables high resolution DWI. However, phase differences between segments can lead to severe artifacts. This work investigates an algorithm to enable reconstruction of interleaved segmented acquisitions without the need of additional calibration or navigator measurements. Given a limited number of interleaves, the initial phase estimates can be calculated by a traditional parallel imaging reconstruction, using the unweighted scan of the DWI measurement as a reference. The ASeDiWA jointly reconstructs all segments of one DWI frame maintaining their phase information. Therefore, the algorithm allows for an iterative improvement of the phase estimates included in the joint reconstruction.

Sen Jia¹, Zhilang Qiu^1,2, Lei Zhang¹, Haifeng Wang¹, Xin Liu¹, Hairong Zheng¹, and Dong Liang^1
1Paul C. Lauterbur Research Center for Biomedical Imaging, Shenzhen Institutes of Advanced Technology, Shenzhen, China, ²University of Chinese Academy of Sciences, Beijing, China

Equipping the existing parallel imaging methods such as ESPIRiT and SPIRiT with a full field-of-view (FOV) calibration could resolve the fold-over artefacts induced by reducing the imaging FOV to be smaller than the object size. Full-FOV images could be reconstructed by accurately resolving the aliased components in image space, or by reconstructing the kspace at a finer sampling interval corresponding to full-FOV. Both approaches requires a separate full-FOV calibration data which could be acquired efficiently. Reduced FOV Parallel imaging methods with full-FOV calibration may provide an alternative approach to treat the common FOV aliasing problem in practice.

Wei Liu¹, Simon Bauer², and Stephan Kannengiesser^2
1Siemens Shenzhen Magnetic Resonance Ltd, Shenzhen, China, ²Siemens Healthcare GmbH, Erlangen, Germany

We describe a more accurate analytical method to calculate the g-factor in dual kernel Slice-GRAPPA (SG-DK) reconstruction for blipped-CAIPI simultaneous multi-slice EPI data. To account for the effect of EPI phase correction, the Slice-GRAPPA kernels are phase corrected before the combination with the in-plane GRAPPA kernel. The experimental results based on a phantom study highlight that there is excellent agreement between SNR maps calculated with the standard pseudo multiple-replica method and the proposed method.

Mahmoud Mostapha¹, Boris Mailhe¹, Simon Arberet¹, Dominik Nickel², and Mariappan S. Nadar ^1
1Digital Technology and Innovation, Siemens Healthineers, Princeton, NJ, United States, ²Magnetic Resonance, Siemens Healthineers, Erlangen, Germany

Parallel Imaging (PI) is a crucial technique for accelerating data acquisition in Magnetic Resonance Imaging (MRI), which is exceedingly time-consuming. With current SENSE-based MRI reconstruction formulated as a trainable unrolled optimization framework with several cascades of regularization networks and varying data consistency layers, coils sensitivity maps (CSMs) are needed at each cascade. Therefore, we propose a deep sets CSM estimation network (DS-CSME in short), enabling an end-to-end deep learning solution that allows for further MRI acceleration while preserving the overall reconstructed image quality.

Silu Han¹, Chidi Patrick Ugonna¹, Mahesh Bharath Keerthivasan^2,3, and Nan-kuei Chen^1,3
1Biomedical Engineering Department, The University of Arizona, Tucson, AZ, United States, ²Siemens Medical Solutions USA, New York, NY, United States, ³Medical Imaging Department, The University of Arizona, Tucson, AZ, United States

A novel two-dimensional (2D) coil-signature-based phase cycled correction method has been developed for Nyquist artifact removal in echo planar imaging (EPI).  Our method uses already available coil sensitivity information, without requiring extra reference scans, to correct 2D phase errors and can be applied equally well to single-shot and multi-shot EPI. Our results show that the developed method can effectively reduce Nyquist artifacts in EPI data acquired using a variety of acceleration schemes, such as through-plane Multi-band Imaging (MB) and in-plane parallel SENSitivity Encoding Imaging (SENSE).

Zaccharie Ramzi^1,2,3, Philippe Ciuciu^1,2, Jean-Luc Starck³, and Alexandre Vignaud^1
1Neurospin, Gif-Sur-Yvette, France, ²Parietal team, Inria Saclay, Gif-Sur-Yvette, France, ³Cosmostat team, CEA, Gif-Sur-Yvette, France

We perform a qualitative analysis of performance of XPDNet, a state-of-the-art deep learning approach for MRI reconstruction, compared to GRAPPA, a classical approach. We do this in multiple settings, in particular testing the robustness of the XPDNet to unseen settings, and show that the XPDNet can to some degree generalize well.

Tobias C Wood¹, Emil Ljungberg¹, and Mark Chiew^2
1Neuroimaging, King's College London, London, United Kingdom, ²Wellcome Centre for Integrative Neuroimaging, FMRIB, Nuffield Department of Clinical Neurosciences, University of Oxford, Oxford, United Kingdom

We present a method for filling the dead-time gap in ZTE imaging using a parallel imaging technique. We present results in phantoms and a human brain demonstrating high fidelity reconstructions comparable to existing methods that require collection of additional data.

Tyler E Cork^1,2, Matthew J Middione¹, Michael Loecher¹, Kévin J Moulin¹, John M Pauly³, and Daniel B Ennis^1,4
1Radiology, Stanford University, Stanford, CA, United States, ²Bioengineering, Stanford University, Stanford, CA, United States, ³Electrical Engineering, Stanford University, Stanford, CA, United States, ⁴Radiology, Veterans Affairs Health Care System, Palo Alto, CA, United States

Echo planar imaging (EPI) is amongst the fastest ways to reduce scan times for long imaging techniques, but it is also susceptible to geometric distortion caused by phase accrual while traversing k-space over long echo times (TE). The objective was to determine the feasibility of a new EPI data acquisition technique that is able to: (1) maintain the SNR efficiency compared to traditional EPI methods, and (2) create a break in the trajectory linearity to allow for geometric distortion correction from one acquisition. Herein, we demonstrate a feasible reconstruction pipeline.

Yumna Bilal^1,2, Ibtisam Aslam^1,3, Muhammad Faisal Siddiqui¹, and Hammad Omer^1
1Medical Image Processing Research Group (MIPRG), Department of Electrical & Computer Engineering, COMSATS University Islamabad, Islamabad, Pakistan, ²Department of Electrical Engineering, University of Gujrat, Gujrat, Pakistan, ³Service of Radiology, Geneva University Hospitals and Faculty of Medicine, University of Geneva, Geneva, Switzerland

This work aims at reconstructing the undersampled non-Cartesian k-space signals by generating a cloud of randomly located additional points through GRAPPA Operator Gridding (GROG) facilitated Bunch Phase Encoding (BPE) collectively termed as GROG-BPE scheme. Gridding of this data is performed onto a Cartesian grid. Inherent randomness in the gridded BPE data is exploited using Compressed Sensing (CS) to obtain the solution image at higher acceleration factors and the results are compared with conventional CG-INNG method. Every step in the proposed method (right from BPE generation to reconstruction) is self-calibrating and does not require additional calibration signals.