Zhifeng Chen1, Ling Xia1,2, Feng Liu3, Qiuliang Wang4, Yi Li4, Xuchen Zhu4, Feng Huang5. 1. Department of Biomedical Engineering, Zhejiang University, Hangzhou, Zhejiang, People's Republic of China. 2. State Key Lab of CAD & CG, Zhejiang University, Hangzhou, Zhejiang, People's Republic of China. 3. School of Information Technology and Electrical Engineering, The University of Queensland, Brisbane, QLD, Australia. 4. Division of Superconducting Magnet Science and Technology, Institute of Electrical Engineering, Chinese Academy of Sciences, Beijing, People's Republic of China. 5. Philips Healthcare, Suzhou, Jiangsu, People's Republic of China.
Abstract
PURPOSE: To improve the performance of non-Cartesian partially parallel imaging (PPI) by exploiting artificial sparsity, the generalized autocalibrating partially parallel acquisitions (GRAPPA) operator for wider band lines (GROWL) is taken as a specific example for explanation. THEORY: This work is based on the GRAPPA-like PPI having an improved performance when the to-be-reconstructed image is sparse in the image domain. METHODS: A systematic scheme is proposed to artificially generate the sparse image for non-Cartesian trajectory. Using GROWL as a specific non-Cartesian PPI method, artificial sparsity-enhanced GROWL (ARTS-GROWL) is used to demonstrate the efficiency of the proposed scheme. The ARTS-GROWL consists of three steps: 1) generating synthetic k-space data corresponding to an image with smaller support, that is, artificial sparsity; 2) applying GROWL to the synthetic k-space data from previous step; and 3) recovering the final image from the reconstruction with the processed data. RESULTS: For simulation and in vivo data, the experiments demonstrate that the proposed ARTS-GROWL significantly reduces the reconstruction errors compared with the conventional GROWL technique for the tested acceleration factors. CONCLUSION: Taking ARTS-GROWL, for instance, experimental results indicate that artificial sparsity improved the signal-to-noise ratio and normalized root-mean-square error of non-Cartesian PPI. Magn Reson Med 78:271-279, 2017.
PURPOSE: To improve the performance of non-Cartesian partially parallel imaging (PPI) by exploiting artificial sparsity, the generalized autocalibrating partially parallel acquisitions (GRAPPA) operator for wider band lines (GROWL) is taken as a specific example for explanation. THEORY: This work is based on the GRAPPA-like PPI having an improved performance when the to-be-reconstructed image is sparse in the image domain. METHODS: A systematic scheme is proposed to artificially generate the sparse image for non-Cartesian trajectory. Using GROWL as a specific non-Cartesian PPI method, artificial sparsity-enhanced GROWL (ARTS-GROWL) is used to demonstrate the efficiency of the proposed scheme. The ARTS-GROWL consists of three steps: 1) generating synthetic k-space data corresponding to an image with smaller support, that is, artificial sparsity; 2) applying GROWL to the synthetic k-space data from previous step; and 3) recovering the final image from the reconstruction with the processed data. RESULTS: For simulation and in vivo data, the experiments demonstrate that the proposed ARTS-GROWL significantly reduces the reconstruction errors compared with the conventional GROWL technique for the tested acceleration factors. CONCLUSION: Taking ARTS-GROWL, for instance, experimental results indicate that artificial sparsity improved the signal-to-noise ratio and normalized root-mean-square error of non-Cartesian PPI. Magn Reson Med 78:271-279, 2017.