MR Image Denoising Method Based On Low-rank Characteristic

Magnetic resonance imaging(MRI)is an imaging technology which is capable of displaying the internal structural information of a human body,and it has been widely used in many fields such as clinical diagnosis and scientific research.However,due to the inevitable influence of noise in the MR imaging process.Therefore,how to remove noise and improve image SNR is an important research direction in MR image processing.Considering about the high redundancy and self-similarity of MR images,the denoising method based on image structure information has attracted lots of attention.In this paper,the MR imaging denoising method based on the low-rank characteristics of the MR image is studied in depth,and the main research work is reflected by the following two aspects:A new static MR image denoising method based on low-rank decomposition based Hankel matrix and image TV is proposed.The proposed method utilizes the low-rank characteristics of the MR image matrix and the sparsity of the MR image in the difference domain.By constraining the kernel norm of the Hankel matrix and as well as the TV L₁ norm and minimizing them at the same time,an objective function is established.Furthermore,by introducing the Alternating Direction Method of Multipliers(ADMM),the objective function is solved,and the process efficiency is greatly improved.In addition,the computer simulation results show that compared with the denoising method only using lowrank characteristics,the method has good denoising effect for different noise types and has a wide range of application.A new dynamic MR image denoising method based on tensor singular value decomposition is proposed.The proposed method considers the dynamic MR image as thirdorder tensors,and it utilizes the high redundancy of the MR image as well as the sparsity of the MR image difference results in the 2-D space domain and one dimensional(1-D)time domain,by minimizing the tensor-nuclear-norm as well as the TV L₁ norm,an image denoising model is established.Furthermore,by introducing the Alternating Direction Method of Multipliers(ADMM),the objective function is solved.Computer simulation results show that the proposed method has good performance in dynamic MR image denoising.