A Comparative Study Of Bioluminescence Tomography Reconstruction Algorithms Based On Greedy Strategy

Molecular imaging can make vivo imaging of biological tissues in molecular,and provides a new method for the study of gene function,disease pathogenesis and therapeutic effect evaluation.Now it has been widely used in tumor detection,gene therapy and pathogenesis,drug development etc..As an important mode of optical molecular imaging,bioluminescence tomography(BLT)is based on the photon distribution measured by the body surface to invert the distribution of light source(target)in vivo.This is a typical inverse problem,especially because of the lack of measurement information,uncertainty exacerbate the ill-posedness of the reconstruction problem,making the accurate three-dimensional reconstruction of source becomes a challenging problem.In the existing bioluminescence tomography based on finite element method,because the distribution of the target is very sparse in biological tissue,the number of grid nodes in the light source area is far smaller than the number of nodes in the whole reconstruction domain.By referring to the theory and method of sparse signal recovery and compressed sensing reconstruction in signal processing,this paper makes a comparative study on the sparse reconstruction algorithm of high efficiency reconstruction of the bioluminescence signal.The existing compressed sensing reconstruction algorithm includes convex optimization algorithm,greedy algorithm and combination algorithm.The greedy algorithm is an iterative method,each time to construct and solve a local optimal solution to gradually approximate the global optimal solution,which has the advantages of low computational cost and high efficiency.In this paper,we focus on the comparison of several representative algorithms based on greed.including the orthogonal matching pursuit(OMP),Stagewise Orthogonal Matching Pursuit(StOMP),Regularized Orthogonal Matching Pursuit(ROMP),and Compressive Sampling Matching Pursuit(CoSaMP).Based on the finite element method,combined with the anatomical structure of the biological tissue,the representative algorithms are combined into the reconstruction of the sparse light source.In order to evaluate and compare the performance of each algorithm,a multi group simulation experiment was designed to test the performance of the single target and multiple targets.The experimental results show that they can accurately reconstruct the position of the light source in the case of noise.In particular,CoSaMP is used as a depth improved matching pursuit algorithm,which shows better robustness and stability against noise.This research can give guidance to the practical application of the bioluminescence tomography.