Jun Ma1,2, Bernhard Gruber3,4, Xinqiang Yan1,5, William A Grissom1,2,5. 1. Vanderbilt University Institute of Imaging Science, Nashville, TN, USA. 2. Department of Biomedical Engineering, Vanderbilt University, Nashville, TN, USA. 3. A. A. Martinos Center for Biomedical Imaging, Massachusetts General Hospital, Harvard Medical School, Charlestown, MA, USA. 4. Division MR Physics, Center for Medical Physics and Biomedical Engineering, Medical University Vienna, Vienna, Austria. 5. Department of Radiology and Radiological Sciences, Vanderbilt University, Nashville, TN, USA.
Abstract
PURPOSE: To accelerate the design of (under- or oversampled) multidimensional parallel transmission pulses. METHODS: A k-space domain parallel transmission pulse design algorithm was proposed that produces a sparse matrix relating a complex-valued target excitation pattern to the pulses that produce it, and can be finely parallelized. The algorithm was applied in simulations to the design of 3D SPINS pulses for inner volume excitation in the brain at 7 Tesla. It was characterized in terms of the dependence of computation time, excitation error, and required memory on algorithm parameters, and it was compared to an iterative spatial domain pulse design method in terms of computation time, excitation error, Gibbs ringing, and ability to compensate off-resonance. RESULTS: The proposed algorithm achieved approximately 80% faster pulse design compared to the spatial domain method with the same number of parallel threads, with the tradeoff of increased excitation error and RMS RF amplitude. It reduced the memory required to store the design matrix by 99% compared to a full matrix solution. Even with a coarse design grid, the algorithm produced patterns that were free of Gibbs ringing. It was similarly sensitive to k-space undersampling as the spatial domain method, and was similarly capable of compensating for off-resonance. CONCLUSIONS: The proposed k-space domain algorithm accelerates and finely parallelizes parallel transmission pulse design, with a modest tradeoff of excitation error and RMS RF amplitude.
PURPOSE: To accelerate the design of (under- or oversampled) multidimensional parallel transmission pulses. METHODS: A k-space domain parallel transmission pulse design algorithm was proposed that produces a sparse matrix relating a complex-valued target excitation pattern to the pulses that produce it, and can be finely parallelized. The algorithm was applied in simulations to the design of 3D SPINS pulses for inner volume excitation in the brain at 7 Tesla. It was characterized in terms of the dependence of computation time, excitation error, and required memory on algorithm parameters, and it was compared to an iterative spatial domain pulse design method in terms of computation time, excitation error, Gibbs ringing, and ability to compensate off-resonance. RESULTS: The proposed algorithm achieved approximately 80% faster pulse design compared to the spatial domain method with the same number of parallel threads, with the tradeoff of increased excitation error and RMS RF amplitude. It reduced the memory required to store the design matrix by 99% compared to a full matrix solution. Even with a coarse design grid, the algorithm produced patterns that were free of Gibbs ringing. It was similarly sensitive to k-space undersampling as the spatial domain method, and was similarly capable of compensating for off-resonance. CONCLUSIONS: The proposed k-space domain algorithm accelerates and finely parallelizes parallel transmission pulse design, with a modest tradeoff of excitation error and RMS RF amplitude.
Authors: William Grissom; Chun-yu Yip; Zhenghui Zhang; V Andrew Stenger; Jeffrey A Fessler; Douglas C Noll Journal: Magn Reson Med Date: 2006-09 Impact factor: 4.668
Authors: Stephan Orzada; Klaus Solbach; Marcel Gratz; Sascha Brunheim; Thomas M Fiedler; Sören Johst; Andreas K Bitz; Samaneh Shooshtary; Ashraf Abuelhaija; Maximilian N Voelker; Stefan H G Rietsch; Oliver Kraff; Stefan Maderwald; Martina Flöser; Mark Oehmigen; Harald H Quick; Mark E Ladd Journal: PLoS One Date: 2019-09-12 Impact factor: 3.240