Realization Of Image Reconstruction Algorithm On Parallel Mri Based On Frequency Domain Using VC

Magnetic resonance imaging is an important tool in the clinical diagnosis and research. The speed of MRI is limited by the traditional hardware-level, gradient switching speed and human tolerance limits, which limits its scope of application. For parallel magnetic resonance imaging, multiple coils are used for parallel scan. The data are under-sampled for reducing phase encoding steps. The speed is improved greatly by reducing acquisition time.The sub-sampled data make the FOV of reconstruction image reduced, resulting in a wrap-round artifact. Parallel MRI eliminates the artifact using the spatial information of phased array coils. SENSE and GRAPPA are commonly applied into commerce successfully. MRI inland falls behind abroad about ten years. Parallel MRI technology is not realized in the MR scanner of china.Firstly, the theory of parallel magnetic resonance imaging is introduced. SMASH and GRAPPA algorithm are introduced in detail. Two methods of coil sensitivity estimation are introduced.Then, the implementation method of pMRI using C++ is studied. Several key engineering issues are discussed, such as the sub-sampled data generation and storage, generalized FFT, the generalized matrix of the Moore-Penrose Pseudo-inverse solution, the storage and display of the reconstructed image. We developed SMASH class and GRAPPA class which package the key steps of SMASH and GRAPPA algorithm to facilitate the call and provide a reference for other pMRI algorithms.Finally, the SMASH class and GRAPPA class are tested. The reconstruction images are get successful. Different reconstruction parameters make different quality of the reconstruction image. Bigger acceleration factor and fewer ACS lines make the image quality more poor. The images reconstructed by GRAPPA are better than SMASH, but GRAPPA cost more time.We developed pMRI software based on MFC dialog using VS 2005. We can read the raw data, select pMRI algorithm, set reconstruction parameters conveniently. It is a basic platform for development of other pMRI algorithms.