The Research Of The Magnetic Of Resonance Imaging Reconstruction Based On Compressed Sensing

MRI is one of the important clinical imaging modalities at this stage,which is widely used because of its advantages in diagnosis accuracy and harmlessness to human body.However,the long data acquisition time is the key to the further development of MRI.Therefore,finding a new fast imaging method is an urgent problem in research and development under the premise of ensuring the quality of the original signal image reconstruction.Compressed sensing theory is a brand new theory of signal sampling which is produced under the background of the development of this era.This theory proposes the epoch-making significance of the signal acquisition and processing.Compressive sensing is aimed at the sparse signal,which can realize the accurate reconstruction of the original signal with a small amount of sampled data.It provides a new idea for the rapid imaging of magnetic resonance images.Therefore,this thesis applies compressive sensing theory to MRI and improves its compression reconstruction algorithm to reduce the complexity of the algorithm and to improve the visual effect of reconstructed images.Finally,this thesis realizes that the original image signal has been reconstructed accurately using a few samples of the original data.The sparse representation of signals,the design of measurement matrix and the choice of reconstruction algorithms are the three core technologies in the process of compressive sensing image reconstruction.Among them,the reconstruction algorithm has a direct impact on the reconstruction of the visual effects of the image.In this paper,the success rate of signal reconstruction and relative error are analyzed in detail.Four types of greedy matching pursuit algorithms,MP algorithm,OMP algorithm,St OMP algorithm and GOMP algorithm are analyzed emphatically.Experimental results show that the algorithm has the best reconstruction effect.Due to the traditional inner product matching criterion,the important data of the signal can not be amplified well when the similarity of the signals is measured.Therefore,the traditional inner product matching method is improved and an improved generalized orthogonal matching pursuit algorithm is proposed,which can effectively accelerate the convergence of the algorithm and improve the robustness of the reconstruction algorithm.In addition,if we reconstruct the whole image directly,the computation is large and the greedy algorithm only deals with the correlation between columns and does not consider the relationship between columns.Therefore,in this paper,the image is divided into block processing to further improve the image reconstruction accuracy,reduce the complexity of the algorithm.Finally,the simulation experiment selects three different MRI images,and reconstructs the original image using the improved generalized orthogonal matching pursuit algorithm.The simulation results show that the improved compressed sensing reconstruction algorithm proposed in this paper can be used to reconstruct the nuclear magnetic resonance image very well and the accurate reconstruction of the original image can be achieved with a lower sampling rate.