Chemical Exchange Saturation Transfer MRI Based On PROPELLER Acquisition And Neural Network Reconstruction

Magnetic resonance imaging(MRI)is a medical imaging technology with high spatial resolution and non-invasiveness,which has been widely used in clinical practice.Chemical exchange saturation transfer(CEST)provides a new contrast mechanism for MRI to effectively detect exchangeable protons with low concentrations,which is of great significance for the detection of many kinds of diseases.However,CEST MRI requires long time for data acquisition and image reconstruction,which severely limits its clinical application.Echo planar imaging(EPI)is the most widely used ultrafast MRI technique.It greatly reduces the data acquisition time,and has high temporal and spatial resolution.However,EPI images are susceptible to magnetic field inhomogeneity and chemical shift effect,which cause image distortion and artifact.Periodically rotated overlapping parallel lines acquisition and enhanced reconstruction(PROPELLER)technique proposed by Pipe in 1999 has high resistance to magnetic field inhomogeneity and can dramatically reduce or eliminate motion artifact and magnetic susceptibility artifact.Its unique sampling pattern also has great potential in accelerating acquisition.In addition,most of the existing algorithms for reconstructing CEST contrast images require long calculation time and are vulnerable to multiple undesirable factors.In recent years,with the rapid development of deep learning,the application of neural network in medical image processing has become a hot research topic.Deep neural network can learn complex nonlinear mapping relationship from a large number of medical image data and extract specific features,so it is widely used in medical image reconstruction,medical image segmentation,etc.In this thesis,we investigate the application of PROPELLER and deep neural network on CEST MRI.The main contents are as follows:1.The mechanism and characteristics of CEST MRI are described in detail.The cause of CEST effect,the mathematical model of proton chemical exchange process,the pulse sequences for CEST MRI,and the commonly used quantification methods for CEST effect are introduced.2.A fast CEST MRI method based on PROPELLER acquisition(called CESTPROPELLER)is proposed.The CEST contrast image is reconstructed from the undersampled CEST images by using deep neural network.Experimental results show that the data acquisition time can be greatly reduced and high-quality CEST contrast image can be obtained by the new method.3.The new method is applied to the nuclear Overhauser effect(NOE)MRI.NOE usually occurs together with CEST effect.Although NOE MRI is not as widely used as CEST MRI at present,it can also provide a special contrast mechanism for MRI and may find potential application.Here the feasibility of the new method in accelerating NOE imaging and NOE quantification is preliminarily tested.