Bioluminescence Tomography Sparse Reconstruction Method Combined With Iterative Strategy

Bioluminescence tomography is to reconstruct the distribution of the fluorescent light source in organ with known optical parameters of biological tissue and biological surface light rate distribution.This technology is the combination of medical imaging technology and the modern molecular biology,and in the analysis of protein and oncology research has a certain application in clinical application.As a new molecular imaging mode,bioluminescence tomography has forward and inverse two solving process.Aiming at the photons transmission characteristics of high scattering and low absorption,based on the first-order spheric harmonic expansion of Radiative Transfer Equation,which is the Diffusion Approximation,we use mathematical method to establish and solve the mathematical model of linear relationship--about biological surface light rate distribution and fluorescent light source distribution in the body under the constraint of the corresponding boundary conditions.Because transfer process of the near-infrared light in biological tissue is complex,and the measured light rate on the surface is limited,inverse process of bioluminescence tomography is an ill-posed problem.So bioluminescence tomography is full of challenges.It is necessary to integrate as much as possible priori information and take strategies to reduce the size of the solution and to improve the accuracy of the reconstruction.This paper revolves around overcoming the seriously ill-posed problem of reconstruction of bioluminescence tomography in order to improve the quality of reconstruction.This paper mainly take a variety of strategies based on some certain algorithm,in order to improve the locating precision,stability,light source recognition etc of reconstruction to improve the reconstruction results.The main work is as follows:(1)About transmission of photons in biological tissue,this paper introduces the first order approximation model of spheric harmonic expansion of the radiative transfer equation namely the diffusion approximation model and simulates the simulation experiment of forward process.(2)We probes into the method based on two-step iterative shrinkage thresholding algorithm to reconstruct the target of BLT.This algorithm included Iterative Shrinking/Thresholding algorithm and Iterative Re-Weighted Shrinkage,updates the current value by using the first two iterative.We study the method combined with L1 norm and TV norm respectively,and investigate the simulation experiment to verify the effectiveness.In the simulation experiment,localization accuracy is about 1.1 mm?1.2 mm.(3)Probes into the localization performance and dependency of regularization parameter based on Incomplete Variables Truncated Conjugate Gradient method.And on the basis of the algorithm,we present a light source iteratively shrinking permissible region method combined with multispectral strategy,which could reduce the ill-posedness with the more multispectral date.By a lot of simulation experiments,we investigate localization performance with single light source and ability to resolve two neighboring targets with double light sources.Simulation results show that only using single view measurement the proposed reconstruction method can yield more stable reconstruction no matter in single source cases or in double sources cases.At the same time,in the single light source simulation experiments the spherical spatial distribution is keeped well in 3d reconstruction,and in the double light sources simulation experiments double light sources can be better discriminated.