Research On The Quantitative Techniques In Intravoxel Incoherent Motion Imaging And Chemical Shift Imaging

The introduction of magnetic resonance imaging(MRI)in the clinical and research community started a new era of medical imaging.Compared with X-ray,nuclear medicine,ultrasound-based imaging technologies,MRI technology,which is based on nuclear magnetic resonance(NMR),can provide images with a high soft-tissue contrast.Another technology based on NMR,which is called magnetic resonance spectroscopy(MRS),can be used to provide metabolic information.Both in MRI and MRS,there is an increasing interest in quantitative approaches.An important feature of MRI is that the image contrast is fundamentally multiparametric,pri-marily based on T1,T2 relaxation time and proton density.Furthermore,it was soon recognized that the diffusion and flow of water molecules in tissues can play an important role in the contrast of MRI images.One of the MRI methods exploiting the water diffusion in tissues is the intravoxel incoher-ent motion(IVIM)imaging,that can be used to assess simultaneously diffusion and perfusion.IVIM imaging has been proved of great value in diagnosis,staging and prognosis,and its clinical application in the human body such as head and neck,prostate,liver,kidney and other parts is increasing rapidly.However,the quantification of IVIM perfusion-related parameters still suffers from large variability and low repeatability.This is particularly true in organs such as liver,which contains a rich vascular network.The numerous blood vessels of different sizes in liver represent a confounding factor in the IVIM quantification of tissue perfusion.Between MRI and MRS,the MR spectroscopic imaging(MRSI)is an imaging technology that combines to some extent the metabolic information of MRS with the spatial coverage of MRI.The MRSI is also known as the chemical shift imaging(CSI).Compared with IVIM,which is used to quantify the parameters describing physical phenomena such as diffusion and perfusion in the tissue,CSI is mainly used to quantify the metabolite content in different organs such as brain,prostate,mus-cle,etc..In the CSI technology,there are two mainstream solutions for providing reference signals for quantitative quantification:in vitro physical phantom and in vivo reference signal.However,in vivo reference signals can be easily affected by pathology,and the use of physical phantom outside the body has a number of drawbacks,such as for instance the loss of CSI spatial resolution.This dissertation focused on these quantitative techniques in IVIM and CSI,specifically:1.Sparsity-based All-Voxel Tri-Exponential IVIM(SAVTE-IVIM)algorithmIn view of the current clinical interest in the quantification of the parameters of the IVIM diffusion-weighted images of liver,and the vascular structure with its confounding effect in the liver,we pro-posed a novel method,a sparsity-constrained all-voxel tri-exponential IVIM(SAVTE-IVIM)algo-rithm,that can automatically identify the existence of potential blood vessels in the target region-of-interest(ROI).In addition to the main goal of identifying potential blood vessels in a given ROI,the algorithm can simultaneously quantify the IVIM parameters of all voxels in the ROI to evaluate the diffusion,perfusion,and blood vessel confounding effect in each voxel.Specifically,we proposed a tri-exponential model based on sparse constraint to describe all voxels at the same time.Typically,the IVIM parameters are evaluated voxel-by-voxel(voxel-wise),while SAVTE-IVIM can quantify all the voxels in the ROI simultaneously.In addition,in order to solve the proposed new model,an optimization algorithm,based on the idea of Alternating Direction Multiplier(ADMM)together with the use of Levenberg Marquardt algorithm to deal with nonlinear problems,was proposed.Two strategies for the inherent non-negative constraints were also introduced.2.Implementation and Comparison of Five Fitting Algorithms for IVIM Quantification on Ver-tebral Bone MarrowSince Marchand et al.successfully applied the IVIM method to the quantification of bone mar-row in 2014,there has been an increasing interest in bone marrow IVIM in recent years.However,there are still issues with the current image quality of bone marrow IVIM;furthermore,little attention has been paid to the investigation of optimal algorithms for IVIM quantification of bone marrow.It is worth noting that an optimal algorithm can to some extent compensate for the low image quality.In view of the above considerations,i)we applied a recently proposed protocol that improves the image quality in bone marrow IVIM,using the RESOLVE(readout segmentation of long variable echo train)sequence and ii)we implemented five algorithms for the parameter quantification of vertebral bone marrow IVIM.Four algorithms,One-Step,Two-Step,Three-Step and Fixed-D~*algorithms are based on the idea of least squares(LSQ),and the fifth one is a Bayesian-based algorithm.A comparison among these algorithms was conducted.Furthermore,maps of the IVIM parameters were generated and compared.3.Virtual phantom chemical shift imaging(ViP CSI)Based on some shortcomings of the mainstream solutions for providing reference signals in CSI technology(including physical phantom in vitro and internal reference):the demand for an additional MR scan,decrease of the CSI resolution,etc.,we proposed to use the virtual phantom technique,which was extended from ERETIC(Electronic REference To access In vivo Concentrations)technology to design a virtual phantom,to provide in CSI the reference signal for quantification.The amplitude and frequency can be custom designed,to simulate the FID signal which is going to be acquired with CSI to provide a reference signal.In the classic CSI technology,with different phase encoding gradients,a repeated NMR signal acquisition process is required.For example,to generate an 8×8 matrix,data acquisition needs to be repeated 64 times.However,the proposed scheme can generate the same reference signals in all 64 voxels by transmitting the ViP signal only once in one of the 64 data acquisitions.In this way,an additional MR scan is no more needed,simplifying the CSI imaging protocol.It can also be further applied to different data acquisition schemes,such as elliptical CSI or weighted CSI.In addition,the ViP reference signal can be customized in its amplitude and frequency and displays an excellent spatial uniformity and time stability for all voxels in the sample.