Research Of Magnetic Resonance Electrical Properties Tomography For Nonuniform Human Tissues

The dielectric properties(EPs)of biological tissues,which consist of electrical conductivity and permittivity,describe how electromagnetic waves behave in the tissues.Clinical research shows that the EPs of abnormal human tissue are different from normal tissue.EPs can potentially be used as biomarkers indicating the health condition of the tissue,which is helpful for the early diagnosis of diseases.In addition,EPs are critical to estimate electric current and electromagnetic distribution inside the body in applications utilizing electromagnetic stimulation for treatment.Thus,tissue EPs provide an important significance in clinical research.The quantitative calculation of specific absorption rate(SAR)requires knowledge of the EPs distribution.Improvements in the reliability of SAR estimation would in return unleash the full ability of ultra-high-field MRI that is otherwise limited by conservative safety constraints.The reported literature value of EPs predominantly pertains to ex-vivo measurements performed with dielectric probes.There are no mature non-invasive EPs measurement methods for the in vivo tissue.Since 2009,the magnetic resonance electrical properties tomography(MREPT)method gained great interest owning to reconstruct EPs' distribution non-invasively.However,there are challenges for existing MREPT methods.For example,the common methods were based on the homogeneous assumption: the distribution of EPs is uniform.However,this assumption is invalid for most cases,especially for the diseased tissue with varied EPs;besides,the MERPT needs to calculate second-order derivative of measured radiofrequency(RF)field,making it sensitive to the measurement with a degraded quality of reconstruction.Moreover,the magnitude and phase of the RF field were obtained through the MR scanning for the MREPT.Meanwhile,the magnitude of MRI images was also gained simultaneously.Hence,the MR structure information was not used in the EPs reconstruction.To address the above issues,this dissertation focuses on the research of MREPT.Especially,we investigates the conductivity imaging method for non-uniform human tissue.The main research contents include the following three parts:(1)This dissertation investigated the conductivity reconstruction method considering the non-uniform distribution of EPs of human tissues.The traditional MREPT method was based on the homogeneous Helmholtz equation,which depends on the assumption that EPs distribution is locally constant.However,it results in artifacts in the boundary of different tissues.This dissertation considers that the distribution of EPs of human tissue is non-uniform,leading to an increase in the reconstruction equation's complexity.Thus,a novel conductivity reconstruction method based on double constraints was proposed.Firstly,the reconstruction equation was discretized via a finite-difference method and then constructed an optimization problem solved by the iterative method;secondly,the wavelet transform and total variation regularization terms are introduced to suppress the spurious oscillations.Experiments show that the proposed method can reduce the reconstruction error and improve the quality of conductivity reconstruction.(2)A novel approach was introduced to improve the noise robustness of the conductivity imaging method.The conductivity reconstruction equation contains the first and second-order derivatives of transceive phase,and the second-order derivative is sensitive to noise.This dissertation modifies the equation via the rule of divergence;the modified equation calculates the first-order derivative of transceive phase twice.To solve the modified equation,the total generalization variation based conductivity reconstruction method was proposed.Firstly,the equation was discretized through the finite difference method,and then the problem was transformed into a linear inverse problem.The second-order total generalization variation was introduced as a regularization term,and the optimization problem was solved by the alternating direction method of the multipliers method.The proposed method was verified by simulations and clinical experiments.The simulation results showed that the proposed could suppress the noise,and the experiment results illustrated that the proposed method shows advantages over other existing methods.(3)The MRI structure information was applied to the MREPT reconstruction to improve the EPs image detail.The MR structure images were obtained simultaneously when acquiring the RF Field via MRI scanning.A prior information based MREPT method was proposed.Firstly,the dissertation defined a new prior to extracting the location and direction of edge information from the MR structure image using the windowed variation calculation;a weighted total generalized variation regularization term was constructed by using the similarity between EPs and MR structure images,then the reconstruction model containing prior information was obtained.The experimental results show that the proposed method preserves more edge details of the conductivity images.