Variable-FOV Reconstruction for 3D Non-Cartesian Parallel Imaging

摘要

Conventional parallel MRI reconstruction methods can be used in the case of variable-density non-Cartesian acquisitions, but they often require iterations with high complexity, which limits their applicability, especially for 3D reconstructions. A technique based on PILS that utilizes variable FOVs to obtain faster reconstructions for 2D non-Cartesian data has recently been proposed. Here we apply this new method to 3D variable density non-Cartesian data, and further process it using wavelet regularization to get rid of remaining artifacts. We compare the performance of using different parameters for regularization, and then, we compare the performance of the overall reconstruction technique to SOS conventional gridding, PILS, and ESPIRiT reconstructions. Results indicate that its performance is superior to the compared reconstruction methods for different acceleration values.