Research On Bioluminescence Fault Imaging Reconstruction Algorithm Based On L1 Norm Function

Bioluminescence tomography is a new kind of optical molecular imaging technique. It could offer the possibilities to monitor physiological and pathological processes at molecular level in biological tissue by locating and reconstructing fluorescent light source of the body to provide an important reference for disease diagnosis, the cell lesion detection and drug development.Bioluminescence tomography includes the forward process and the reconstruction process. The forward process is an important basis of bioluminescence tomography reconstruction process. It uses the known light sources distribution and optical parameters of biological simulation organizations in small animals or imitation of biological tissue to obtain the small animals or imitation of body surface transmission of biological tissue optical signal distribution by solving the mathematical model of the photon transmission. The reconstruction process exploits the biological tissue optical signal distribution of the small animals or its body imitation to obtain the light source position and light source density distribution of the biological tissues. Bioluminescence tomography light reconstruction is an ill-posed inverse problem because of the complexity of the light transmission in the biological tissues and normal distribution limitation of measured data. In this paper, several different reconstruction algorithms were introduced to research the advantages and disadvantages of all bioluminescence tomography reconstruction process algorithms.In this dissertation, the main content mainly focuses on the following aspects:(1) Based on the forward process, this paper introduced the mathematical model of the photon transmission——diffusion model. This model was the first-order spheric harmonic approximation model of the radiative transfer equation, which had high computing efficiency.(2) The accuracy of the reconstruction results of the reconstruction algorithm based on L2norm objective function were discussed,. Based on the compression perception reconstruction algorithm, we directly put forward the L1norm objective function considering the sparse characteristic of the light source distribution and the serious shortage of surface measurement data.(3) Based on L1norm objective function, we obtained forward the Split Bregman iterative algorithm, the Gradient Projection Sparse Reconstruction algorithm and the L1Regularized Least Squares algorithm. We discussed the feasibility of those algorithms applied to bioluminescence tomography.(4) Using the bioluminescent tomography equipment and the numerical simulation platform, we obtained evaluate parameter of the reconstructed location of absolute error, relative error, reconstructed time, reconstructed density, reconstructed energy etc to the digital mouse model experiment, Those evaluate parameter could be used to the Split Bregman iterative algorithm, the Gradient Projection Sparse Reconstruction algorithm and the L1Regularized Least Squares algorithm. At the same time, we discussed the influence of noise robustness in Split Bregman iterative algorithm and the influence of regularization parameter in Split Bregman iterative algorithm detailed.This article research conclusion includes:(1) The Split Bregman iterative algorithm has a better result in the influence of noise robustness. The influence of regularization parameter of the Split Bregman iterative algorithm have great influence to the reconstruction results.(2) The consumed time of the Split Bregman iterative algorithm is short and the reconstruction result of the light source is accurate.