3D Parallel Imaging with Transition Auto-Calibration (3D-PITA)

Magnetic Resonance Imaging (MRI) has drawn increasing attention from investigators in the past several years because of the fast development of modern biomedical engineering. A well designed MRI scanner can provide more information with respect to soft issues in human body compared with Ultrasound, Computed Tomography (CT) or X-ray. However, current MRI algorithms are still facing some challenges, like computation time and image quality issue. In this thesis, the author proposes a novel 3D MRI images reconstruction method based existing 2D MRI algorithms to shorten the acquisition time and improve the quality of 3D reconstructed images as well. This method, named 3D-PITA, implements a new sampling pattern which is specifically designed. To further reduce noise and suppress artifact, nonlinear GRAPPA is implemented with new sampling pattern. In addition, random Projection also plays a significant role in reducing computation time. The specific procedures for these two algorithms and discussions have been included within this thesis. Then, the core ideal of 3D-PITA and detailed procedure are well illustrated. Finally, in vivo experimental results with discussion have been provided as a final result for the thesis. The deliverables for the thesis would be considered valuable for the further research of volumetric MRI reconstruction because 3D-PITA can significantly improve the reconstruction quality compared to the conventional 3D GRAPPA at high reduction factors.