Technical Note: Sequential combination of parallel imaging and dynamic artificial sparsity framework for rapid free-breathing golden-angle radial dynamic MRI: K-T ARTS-GROWL

Purpose: To develop and validate a fast dynamic MR imaging scheme. A novel approach termed K-T ARTificial Sparsity enhanced GROWL (K-T ARTS-GROWL) is proposed that integrates dynamic artificial sparsity and GROWL-based parallel imaging (PI). Methods: Golden-angle radial k-space data are acquired with the free-breathing sampling scheme and then sorted into a time series by grouping consecutive spokes into temporal frames. The reconstruction framework sequentially applies PI and dynamic artificial sparsity. In the implementation, GROWL is taken as a special PI instance for its high computational efficiency and the K-T sparse is exploited to improve the PI reconstruction performance, because the dynamic MR images are often sparse in the x-f domain. In the final reconstruction procedure, artificial sparsity is constructed and fed back to the previous reconstruction. Results: The K-T ARTS-GROWL results in high spatial and temporal resolution reconstructions. By exploiting dynamic artificial sparsity, the acceleration capability is further improved compared to the PI alone. The experimental results demonstrate that K-T ARTS-GROWL leads to significantly better image quality (P < 0.05) than the frame-by-frame GROWL and frame-by-frame ARTS-GROWL for invivo liver imaging. Compared with the tested K-T reconstruction algorithms, the K-T ARTS-GROWL results in a better or comparable image quality and temporal resolution with greatly decreased computational costs. Conclusion: The proposed technique enables sparse, fast imaging of high spatial, high temporal resolutions for dynamic MRI. (C) 2017 American Association of Physicists in Medicine