Research On Two New Methods Of MR Image Reconstruction

Magnetic Resonance Imaging(MRI)can provides a variety of information about the structure and function of human tissue through different imaging sequences and parameters.It has the advantages of no ionizing radiation,high sensitivity to soft tissue and multiple Imaging orientations,it is one of the most important imaging methods in clinical medicine.However,the main defect is the long acquisition time of signal,so the image undersampling technique based on compression perception has been paid widespread attention in recent years.In this method,an energy functional model is used to reconstruct the MRI accurately using the prior information of the image and part k-space data,but there are usually some problems such as low resolution and noise.For this purpose,a new reconstruction model is proposed in this dissertation,which can describe the sparse structure of the image using the total variation space and preserve the details of the image using shearlet.The main work is as follows:· Aiming at the problem that the spatial norm of L~2 for data fitting item can not keep the reconstructed image contrast effectively,a new total generalized variation(TGV)shearlet reconstruction model is proposed.· In view of the fact that the traditional total variation space can not effectively describe the local features of the image,the method of coupling the weighted matrix operator and the gradient operator is proposed to enhance the directivity of the gradient operator,and the anisotropic total variation shearlet reconstruction model is established.· Since the anisotropic total variation shearlet model is a convex nonsmooth problem coupled with multiple operators,by introducing auxiliary variables,the model is first transformed into a saddle point problem,which is then solved by split bregman method and verified the validity and rationality of the numerical method by numerical comparison.