The Study On Magnetic Resonance Imaging Reconstruction By Using Shearlet Transform And High Degree Total Variation

With the increasing clinical application of magnetic resonance imaging technology,how to improve the quality and speed of magnetic resonance imaging reconstruction becomes a hot research topic.Due to it’s sensitive to the movement of the patient’s body and organs,motion artifacts can be generated by a slight movement in the process of magnetic resonance imaging.Therefore,how to shorten the reconstruction time and improve the reconstruction quality is a big problem to be solved.In this paper,a high degree total variation model(HDTV)is introduced to reconstruct magnetic resonance images.The experimental results show that,compared with the traditional total variation(TV)and the total generalized variation(TGV)methods,the HDTV based reconstruction method can improve the performances of the reconstructed magnetic resonance imaging with higher quality,less reconstruction error of MR images and better average structural similarity.Moreever,HDTV reconstruction method is more complexity than TV and TGV method,which takes the longest reconstruction time.Furthermore,a generalized high degree total variation(GHDTV)model was introduced to reconstruct the magnetic resonance images,and the fast alternating minimization method was used to solve the convex optimization problem of magnetic resonance image reconstruction.The experimental results show that under different acceleration factors,the GHDTV can reconstruct the MR imaging with higher signal to noise ratio,and smaller relative L2 norm error than HDTV method.In spite of the higher reconstruction complexity of GHDTV,the reconstruction time of the GHDTV method is less than the HDTV method.In addition,a new method combining shearlet transform with high degree total variation(ST-HDTV)model,was proposed to reconstruct the magnetic resonance images.The proposed model is solved by Fast Composite Splitting Algorithm and Fast Iterative Shrinkage Thresholding Algorithm.Two different sets of cardiac under-sampling magnetic resonance data were used to validate the reconstruction performance of ST-HDTV.The experimental results show that ST-HDTV method can improve the reconstruction quality of magnetic resonance image effectively,and can reduce the error of magnetic resonance imaging.Under different acceleration factors,compared with the HDTV and ST-TGV method,the ST-HDTV can reconstruct the MR imaging with higher signal to noise ratio,and smaller relative L2 norm error,which can eliminate the "staircase effect" and "texture loss effect" effectively.Moreover,the reconstruction time of the ST-HDTV method is less than the GHDTV and ST-TGV method.Therefore,the ST-HDTV algorithm has a significant reconstruction effect and application prospects.