Parallel Imaging: Coil Sensitivities, Gradients, & Applications
Thursday 23 April 2009
Room 314 16:00-18:00

Moderators:

 Anja C. Brau and Fa-Hsuan Lin
 
16:00 759. Virtual Body Coil Calibration for Phased-Array Imaging
    Martin Buehrer1, Peter Boesiger1, Sebastian Kozerke1
1Institute for Biomedical Engineering, University and ETH Zurich, Zurich, Switzerland
    Image-based parallel imaging methods such as sensitivity encoding require accurate coil sensitivity information. Obtaining coil sensitivity maps requires a homogeneous reference image to remove anatomy related structures in the single coil images. Commonly, such a reference image is provided by the body coil. This procedure, however, makes a separate reference scan necessary. In this work we present a method which extracts the complex sensitivity information from fully sampled single coil images without the need for acquiring additional body coil data, allowing for phase preserving self calibrated SENSE reconstruction.
     
16:12 760. Automated Coil Subset Selection for Improved GRAPPA Reconstruction
    Keith Heberlein1
1Siemens AG, Healthcare Sector, Erlangen, Germany
    Conventional wisdom in parallel MR holds that increasing the number of coil elements always improves the image reconstruction. While that likely still holds when looking at the coil array as a whole after parallel reconstruction, the results presented here show that GRAPPA is improved when optimizing the reconstruction parameters on a coil by coil basis. Previous GRAPPA optimization methods have looked predominantly at determining the kernel support in k-space dimensions. The new findings presented here suggest the same procedure can be extended to a synthetic coil direction.
     
16:24 761. Efficient “O-Space” Parallel Imaging with Higher-Order Encoding Gradients and No Phase Encoding
    Jason Peter Stockmann1, Pelin Aksit Ciris1, Robert Todd Constable2
1Biomedical Engineering, Yale University, New Haven, CT, USA; 2Diagnostic Radiology & Neurosurgery, Yale School of Medicine, and Biomedical Engineering, Yale University, New Haven, CT, USA
    "O-space" projection imaging using a combination of linear gradients and the Z2 spherical harmonic is described. Gradient encoding functions are chosen so as to complement the information provided by a circumferential array of surface coils. Using the resulting hybrid encoding function, images are reconstructed from simulated data using the Kaczmarz iterative algorithm. The O-space results show better resolution and lower artifact levels than undersampled radial k-space data, demonstrating that valuable spatial information is provided by the addition of the Z2 field.
     
16:36 762. Noise Behavior of Cartesian PatLoc Reconstruction
    Gerrit Schultz1, Maxim Zaitsev1, Peter Ullmann2, Heinrich Lehr2, Jürgen Hennig1
1Dept. of Diagnostic Radiology, Medical Physics, University Hospital Freiburg, Freiburg, Germany; 2Bruker Biospin MRI GmbH, Ettlingen, Germany
    This contribution deals with an important aspect of imaging: Noise. In particular noise in PatLoc imaging. This imaging modality is a new concept, where the gradients are replaced by coils, which generate nonlinear, non-bijective encoding fields. Non-bijective encoding leads to ambiguities in the resulting images, which can be resolved by means of parallel imaging reconstruction methods. A common feature of such methods is that the SNR varies from point to point. An explicit mathematical expression of SNR is derived for a reconstruction algorithm, which is applicable to Cartesian sampling patterns. The theoretical predictions are then validated by comparing them to simulation results.
     
16:48 763. 3D Undersampled Golden-Radial Phase Encoding Using Iterative Reconstructions and Inherent Regularization
    Claudia Prieto1, Sergio Uribe1, Philip Batchelor1, Pablo Irarrazaval2, Reza Razavi1, David Atkinson3, Tobias Schaeffter1
1Division of Imaging Sciences, King's College London, London, UK; 2Departamento de Ingenieria Electrica, Pontificia Universidad Catolica de Chile, Santiago, Chile; 3Centre for Medical Imaging Computing, University College of London, London, UK
    Iterative SENSE reconstruction for non-Cartesian undersampling leaves residual aliasing and produces noise amplification because the ill-conditioned nature of the inverse problem. We proposed a new undersampled dynamic acquisition and reconstruction method for 3D DCE-MRI. It combines a 3D radial phase-encoding trajectory with the golden angle profile order, providing explicit regularization for iterative reconstructions. This approach takes advantage of the trajectory's properties to generate a regularization image. Moreover, since the acquisition is based on the golden angle, the regularization can be changed adaptively for each frame. The method was tested with synthetic phantoms and in-vivo showing excellent results with undersampling factors up to 49.
     
17:00 764. Automatic Design of Radial Trajectories for Parallel MRI and Anisotropic Fields-Of-View
    Alexey A. Samsonov1
1Radiology, University of Wisconsin-Madison, Madison, WI, USA
    In recent years, non-Cartesian k-space trajectories have gained increased attention owing to a number of unique properties. Challenges of non-Cartesian imaging include the lack of methods to optimize their performance for anisotropic Fields-of-View (FOV) arising from slab-selective excitation and anatomy, and to maximize parallel imaging performance for a given coil array. We developed a new fast approach to the design of non-Cartesian trajectories. As an example, we applied the developed theory to optimize radial trajectories. The new method has demonstrated ability to reduce level of noise and undersampling artifacts in simulated and real data studies.
     
17:12 765. Feasibility of Five-Minute Comprehensive Cardiac MR Examination Using Highly Accelerated Parallel Imaging with a 32-Element Coil Array
    Jian Xu1,2, Ricardo Otazo3, Sven Zuehlsdorff4, Daniel Kim3, Xiaoming Bi4, Qi Duan3, Sonia Nielles-Vallespin5, Monvadi Barbara Srichai3, Thoralf Niendorf6, Renate Jerecic4, Bernd Stoeckel1, Yao Wang2, Li Feng2, Kellyanne Mcgorty3, Daniel K. Sodickson3
1Siemens Medical Solutions USA Inc., New York, NY, USA; 2Polytechnic Institute of NYU, Brooklyn, NY, USA; 3Center for Biomedical Imaging, Dept. of Radiology,New York University School of Medicine, New York, NY, USA; 4Siemens Medical Solutions USA Inc., Chicago, IL, USA; 5Siemens Medical Solutions, Erlangen, Germany; 6Division of Experimental Magnetic Resonance Imaging, RWTH Aachen, Aachen, Germany
    Highly accelerated 3D and 4D imaging techniques for comprehensive cardiac imaging in a total time of 5 minutes were developed and assembled into a single streamlined protocol, which was evaluated for feasibility in healthy adult subjects.
     
17:24 766. Highly-Accelerated Cardiac Cine MR Imaging Using Kats ARC (Autocalibrating Reconstruction for Cartesian Sampling with K- & Adaptive-T-Space Data Synthesis)
    Peng Lai1, Anja C. Brau1, Philip J. Beatty1, Ajit Shankaranarayanan1
1Applied Science Laboratory, GE Healthcare, Menlo Park, CA, USA
    In this study, a new fast imaging approach (kats ARC: Autocalibrating Reconstruction for Cartesian Sampling with k- & adaptive-t-space data synthesis) was developed for fast cardiac MRI. On 5 healthy volunteers, kats ARC was simulated by decimating full cine k-space datasets offline and compared with conventional ARC, sliding window and k-t ARC. Based on our quantitative comparison and visually assessment, kats ARC generated images with the lowest reconstruction error and the highest temporal fidelity, especially with high acceleration factors. This work demonstrates that kats ARC is a promising technique for highly-accelerated cardiac MRI.
     
17:36 767. Real-Time Non-Gated Cardiac MRI Using PARADISE: Doubly Adaptive Accelerated Imaging
    Behzad Sharif1, John Andrew Derbyshire2, Anthony Z. Faranesh2, Robert J. Lederman2, Yoram Bresler1
1Coordinated Science Lab, Department of Electrical & Computer Engineering, University of Illinois at Urbana-Champaign, Urbana, IL, USA; 2Cardiovascular Branch, NHLBI, National Institutes of Health, DHHS, Bethesda, MD, USA
    Patient-Adaptive Reconstruction and Acquisition Dynamic Imaging with Sensitivity Encoding (PARADISE), is a highly accelerated non-gated dynamic imaging scheme that enables artifact-free imaging while providing performance guarantees on achievable SNR and spatio-temporal resolution. In addition to parallel imaging, the method gains acceleration from a sparse spectral support model (in x-y-f space) that is adapted to the imaged subject; hence it is doubly accelerated. It uses the support information and coil sensitivities to adapt both the acquisition and reconstruction; hence it is also doubly adaptive. Results demonstrate the superior spatio-temporal resolution and SNR of PARADISE compared to non-gated TSENSE.
     
17:48 768. Auto-Calibrated Parallel Imaging Using the Unused Echo in Alternating TR SSFP
    Hsu-Lei Lee1, Yoon-Chul Kim1, Ajit Shankaranarayanan2, Krishna S. Nayak1,3
1Ming Hsieh Department of Electrical Engineering, University of Southern California, Los Angeles, CA, USA; 2Global Applied Science Lab, GE Healthcare, Menlo Park, CA, USA; 3Keck School of Medicine, University of Southern California, Los Angeles, CA, USA
    Low-resolution high-SNR k-space data can be acquired during the often-unused short TR echo of alternating TR SSFP. Although exhibiting a slightly different tissue contrast from the final image, this data can still be used to perform auto-calibration of the coil sensitivity map and GRAPPA kernel. We demonstrated R=2 GRAPPA reconstructed cardiac cine images using this method, where this ACS signal is acquired at the same time as the reduced-FOV image and eliminates the need for additional scan time.