Reconstruction Based On Sparse Representation For Fluorescence Molecular Tomography

Fluorescence Molecular Tomography(FMT)has been extensively developed as an important branch of molecular imaging technique due to its high sensitivity,high temporal resolution,no ionization,no radiation,low cost,handy imaging process and so on.In this imaging modality,a specific fluorescence probe was used as the contrast agent,which will emit fluorescent light under the illumination of an external excitation light.The three dimensional distribution of the fluorescent target within the organism was reconstructed through the measurements of light on the surface of the tissues in combination with a proper photon propagation model in the tissues,further revealing the information of the organism’s physiological and pathological characteristics.FMT has great application potential and essential research value in the early detection of cancer,drug distribution,and gene expression.The main work of this thesis focuses on the reconstruction of the fluorescence target in FMT with a continuous wave and is summarized as follows:1.A new model based on compressed sensing for fluorescence molecular tomography is constructed around the optimization of the measurement matrix in FMT.As some studies have shown that optimizing the measurement matrix based on compressed sensing theory can improve the performance of the reconstruction algorithm,the optimization of the measurement matrix(forward matrix)in FMT was studied.Firstly,based on the theories of optimizing the measurement matrix,several existing methods of the measurement matrix optimization are summarized,including the measurement matrix optimization based on the variant of the cross-correlation coefficient,the optimization of the measurement matrix based on the proximity to the unit matrix,and optimization of the measurement matrix based on the equiangular tight frame(ETF);it is verified by the experimental simulation.Secondly,Cyclic Algorithm(CA)and the thin QR decomposition were proposed to the optimization of the measurement matrix,the simulation experimental data,phantom experimental data and in vivo mouse experimental data to validate the analysis effect of the CA algorithm and thin QR decomposition on the optimization of the measurement matrix for FMT.2.Based on the above method of optimizing the measurement matrix,the application of the greedy reconstruction(GR)algorithm in the reconstruction of fluorescence molecular tomography is studied.We studied the reconstruction performance of the Regularized Orthogonal Least Squares(ROLS)algorithm and Stagewise Orthogonal Matching Pursuit(St OMP)algorithm in fluorescence molecular tomography.First,the greedy reconstruction algorithm is briefly summarized and analyzed.Then,based on the optimized measurement matrix,the application of the classical greedy reconstruction algorithm in FMT reconstruction is studied.Finally,the simulation experiments have been conducted to verify the reconstructed performance of the greedy reconstruction algorithm after using the CA algorithm and thin QR decomposition for the optimization of the measurement matrix,combined with the classical reconstruction algorithm(Simultaneous Algebraic Reconstruction Technique,SART).The GR-SART algorithm,the physical phantom experiments and the real mouse in vivo experiments have been conducted and the results have shown the potential of the proposed method for FMT.