Research On MRI Super-resolution Reconstruction Based On Feedback Mechanis

Magnetic Resonance Imaging(MRI)technology,as an important auxiliary means in clinical medicine,is a commonly used medical imaging method.It can quantitatively measure human tissue structure and function information;it has the advantages of high contrast and no radiation damage to the human body.Therefore,high-definition magnetic resonance images are of great significance for disease diagnosis.During the acquisition of magnetic resonance images,in order to obtain high-definition images,it is often necessary to consume a long scanning time,which will increase the pain of patients.If the acquisition process is shortened,the acquired data will be reduced,the resolution of the reconstructed image will be reduced,and the detailed information in the magnetic resonance image will be blurred and difficult to distinguish,thus affecting the doctor’s diagnosis and treatment of the disease.Based on the above problems,this paper proposes two methods for superresolution reconstruction of MRI,so that MRI has the function of short-time scanning and high-precision imaging,and reconstructs high-resolution images.The main research work of this paper is as follows:1)MRI super-resolution reconstruction based on a feedback mechanism.In order to improve the resolution of reconstructed MR images,based on the U-Net network,a feedback mechanism and an attention mechanism are innovatively introduced.In this algorithm,the feedback reconstructed image features are fused with the input image features,and the U-net network with residual channel attention is added to perform feature extraction to realize super-resolution reconstruction of MR images.The experimental results show that,compared with other networks,the proposed network can not only obtain sharper super-resolution MR images,but also significantly improve the peak signal-tonoise ratio and structural similarity.2)MRI super-resolution reconstruction based on dual regression network.This chapter introduces the loop network based on the third chapter,so that the reconstructed highdefinition image can be enlarged as a whole on the basis of complete details,which is convenient for doctors to better observe the details of the diseased tissue.Reconstruct the edge information of the image.Comparing the reconstructed image obtained by the dual regression network algorithm with the existing mainstream algorithms,it is proved that the proposed method can make the texture of the image clearer and the details of the reconstruction complete,which is feasible.This paper evaluates the experimental results through subjective evaluation and objective evaluation,and intuitively shows the performance of the proposed network.The experimental results show that the MRI super-resolution reconstruction algorithm based on feedback mechanism can obtain reconstructed images with high resolution;the MRI superresolution reconstruction based on double regression network can obtain reconstructed images with large size and high resolution.