Reconstruction Of Fluorescence Molecular Tomography Based On Sparse Regularization

With the development of medical imaging technology,the molecular imaging,emerging as a novel imaging mode,could visualize the biological and chemical processes in molecular level of intact living subjects.Fluorescence Molecular Tomography(FMT)is one of the molecular imaging.It can use the specificity of fluorescent probes to perform three-dimensional imaging or tracing of specific targets at the cellular molecular level,and quantify the fluorescent data.This technology has been widely applied to the study of biochemistry,already has a good application in tumor resection,has potential clinical value in the field of metabolism,visual cells and tissue.The research of this paper focuses on the major topic to improve reconstruction time,accuracy and robustness of FMT.There are three methods to solve the FMT.The main contributions are listed as follows:1.A novel trace norm regularization method for FMT was proposed.This method is based on the structured sparsity of the fluorescent regions for FMT.And the trace norm is defined as the sum of the singular values of the matrix,it can approximate instead of matrix rank constraints to make full use of effective information.As this characteristic,it could efficiently reconstruct the sparse distribution of fluorescence.In the process,the accelerated proximal gradient algorithm was used to accelerate the computation.The simulation results show that the method could efficiently get accurate results.2.A novel elastic net regularization method for FMT was proposed.The elastic net(Enet)regularization combines the advantages of L1-norm and L2-norm.It could achieve the balance between the sparsity and smooth by simultaneously employing the L1-norm and L2-norm.To solve the problem effectively,the proximal gradient algorithms was used to accelerate the computation.To evaluate the performance of the proposed method,numerical phantom experiments are conducted.The simulation study shows that the proposed method achieves accurate and is able to reconstruct image effectively.3.A novel reconstruction method for FMT based on L1-norm primal accelerated proximal gradient was proposed.This method was based on L1 norm regularization.At each iteration,it utilizes the last two iterations to obtain a search point.In order to get fast convergence,we adopted primal accelerated proximal gradient(PAPG)methods to efficiently solve the search point.And then the L1-norm regularized projection(L1RP)was performed to obtain the robust and accurate results.Several simulation phantom experiments were designed to verify the performance.The comparative result shows that the proposed method has advantages for robustness,accuracy and efficiency.The in vivo experiment was also performed,and the result was able to indicate the potential for the proposed method of FMT.