The Denoising Algorithm And Reconstruction Of Mr Imaging By Using Sparse Constraints

Compressed Sensing Magnetic Resonance Imaging(CS-MRI) is a promising technique to accelerate dynamic cardiac MR imaging(dCMRI). For dCMRI, the CS-MRI usually exploits image signal sparsity and low-rank property to reconstruct dynamic images using the under-sampled K-space data. For reconstructing MR image by using the sparse constraints, the sparser of the MR image, the higher the accuracy of the reconstructed image. So how to improve magnetic resonance imaging sparseness has important significance in improving the quality of the reconstructed image. In addition, with the demand increasing of magnetic resonance imaging in the clinical application, how to reduce the effects of noise in magnetic resonance imaging has far-reaching significances. Therefore, the magnetic resonance image denoising algorithm and reconstruction algorithm have been investigated by using the sparse constraints.Total generalized variation(TGV) regularization model is one of the most effective method of MR image denoising. However, for 3D dynamic MR image, the TGV regularization model cannot use the correlated information of each slice. In this paper, a three-dimensional TGV(3D-TGV) de-noising method was proposed to denoise different kinds of noise from dynamic MR images. Experimental results show that, compared to the total variation(TV) and 3D-TGV has better denoising effect, with a higher signal to noise ratio(SNR) and smaller artifacts.In order to improve the sparsity of the dynamic magnetic resonance image, a sparse constraint method, combining 3D TGV method with high order singular value decomposition(HOSVD) based tensor decomposition method:k-t- TGV-TD(Total Generalized Variation and Tensor Decomposition), was proposed in this paper. Meanwhile, in order to improve the reconstructed image reconstruction speed, the fast complex splitting algorithm(FCSA) was used to solve complex sparse convex optimization problem by decomposing the upcoming complex convex optimization problem into several simple sub-problems. And then get the solutions of the convex optimization problem by a linear combination of the sub-problem solutions. Finally, thein-vivo cardiac cine and cardiac perfusion MR datasets were used to validate the performance of the proposed method. Experimental results show that, compared to k-t SLR(Sparisty and Low Rank)method and HOSVD method, the proposed method can significantly improve the quality of reconstruction, not only with a higher signal-to-noise ratio but also a relatively lower relative l2 norm error(RLNE).