Design And Implementation Of MRI Medical Image Reconstruction System Based On Deep Learning

In recent years,deep learning technologies have developed rapidly,and various deep learning algorithms have been applied to all many fields in society.The application of deep learning technology in the medical industry has become the trend of modern medicine.As an important basis for disease diagnosis,MRI has been widely used in various medical institutions.However,motion artifacts will be generated during MRI,which will seriously affect the accuracy of disease diagnosis.This paper makes full use of the characteristics of deep learning technology and MRI technology to design a visualization system based on web technology for removing MRI image artifacts.This paper mainly involves the following aspects:1.This article first studied some basic knowledge of deep learning technology and MRI technology,laid the foundation for the design of removing artifact models,and then processed and arranged the MRI images collected in the hospital to form a dataset which can be used for training the model.2.This paper uses two different deep learning algorithms to design models to remove artifacts.The first is to design a network model based on a generative adversarial network.In the selection of the backbone network,this article uses resnet and U-net for comparison experiments.In the selection of loss functions,this article performs ablation experiments on four loss functions.This paper designs a new loss function,EdgeLoss.3.The second method is to design a model based on a multi-scale neural network.The final network structure is selected by comparing multiple different multi-scale neural networks.The ability of the model to extract image features is improved by comparing different residual module structures.A deeper and wider residual module is designed and applied to the image removal artifact model.4.In the design of the visual system based on the removal of artifacts based on web technology,this paper uses the SpringMVC network architecture,the front end uses html+css+javascript technology,and the server uses the Java language to call Python code for design.The system implements the functions of online model call,output of model results and webpage display pictures.The purpose of this study is to learn the features of MRI images through deep learning models,and to learn a mapping from artifact images to artifact-free images based on the training mechanism.This article uses two different deep learning ideas to compare and innovate the components used in the model at the same time to improve the model’s effect of removing artifacts.Based on the above,a web-based visualization system for removing artifacts from MRI images is designed to simplify the use steps to a certain extent and helps doctors diagnose the disease better and more accurately.