Research And Application Of Multimodal Magnetic Resonance Imaging Simulation Algorithms

Multimodal magnetic resonance imaging(MRI)has become the main nondestructive method for detecting the structure,function and micro-material content of human organs.However,in the absence of ground-truth of the structure and function of human tissues,it is difficult to evaluate the accuracy of the subsequent processing algorithms of multimodal MRI.To deal with this issue,this work intends to investigate the simulation algorithm of multimodal MRI and design the corresponding simulation system,which will provide evaluation criteria for the post-processing algorithm of multimodal MRI.The main research works of the paper are as follows:(1)The digital model of tissues is firstly established.Then,based on the basic principle of MRI,the MRI simulation modules with different imaging sequences are designed,which can finally obtain and display the T1-weighted,T2-weighted and proton density-weighted images.(2)The geometrical model of fibrous tissues is firstly constructed,and then based on the basic principle of diffusion magnetic resonance imaging,the diffusion-weighted images of such virtual fiber model are simulated with Monte Carlo method.Subsequently,the corresponding diffusion tensor images and high angle resolution diffusion magnetic resonance images are simulated.Finally,the fiber tracking was implemented and visualized.(3)According to the basic principle of quantitative susceptibility imaging(QSM),the amplitude and phase images of the tissues with susceptibility are firstly simulated.Then,the magnetic susceptibility maps are calculated by phase unwrapping algorithm.Based on which,the simulation module of QSM is designed.(4)The simulation algorithms are validated firstly,and then using the simulation data to train the deep learning model in order to realize the application of our proposed simulation system in the field of denoising for magnetic resonance images.