Research On Magnetic Resonance Imaging Data Acquisition And Reconstruction Algorithm

Magnetic resonance imaging (MRI), which is based on the principles of nuclear magnetic resonance (NMR), is an imaging technique used primarily in medical settings to produce high quality images of the inside of the human body. It is one of the non-invasion diagnostic methods and a powerful tool for medical and molecular biology research. It is often possible to diagnose disease at very early stage without any harmful radiation, and before it is visible by other means. It is important to develop new k-space trajectories and corresponding image reconstruction algorithms for higher imaging speed. In this thesis we compared three kinds of gridding reconstruction algorithms and developed a MRI reconstruction software. We discussed the convolution gridding algorithm, matrix-inversion gridding algorithm and next neighbor gridding algorithm. Next neighbor algorithm is modified to speed up imaging. Above algorithms are applied to numerical models, phantoms, and in vivo experiments on human bodies with the conclusion: (1) Next neighbor algorithm is time efficient with acceptable image quality. It does not require special gradient system and can be easily applied in current systems. (2) Modified next neighbor algorithm runs 3 times faster than the original one with parallel structure, which can be implemented on a parallel system for real-time imaging. A MRI reconstrucition software is developed which can read in raw data, display the trajectory, compensate non-uniform sampling density, reconstruct image and display the result. It is useful to new trajectory design and new image reconstruction algorithm development.