Three-dimensional Bioluminescence Tomography Of In Vivo Animal

Bioluminescence tomography (BLT), a newly, rapidly developing and promising technology, isan imporant branch of molecular imaging technology. It can monitor the growth and metastasis oftumors in organism and detect the expression of specific genes and many other biological processes atthe molecular and genetic level. With the advantages of non-invasive, high sensitivity, simpleoperation and low cost, BLT can be widely used in the fields of tumor detection, gene therapy anddrug development.Bioluminescence tomography consists of two major problems, forward problem and inverseproblem, of which the forward problem is the basis of the inverse problem. The forward problem is tofind out the energy and the distribution of surface lights with given source in the organism,physiological anatomy and tissues’ optical parameters, such as the absorption coefficient and thescattering coefficient. The aim of the inverse problem is to reconstruct the source inside organismaccording to the given physiological anatomy, tissues’ optical parameters and the photons’ distributionon the organism’s surface. The main methods to solve the forward problem are Monte Carlo method,finite difference method and Green function method. Due to the complication of photon’stransportation in tissues and the illness of the inverse problem, there are many problems with thereconstruction algorithm. Mouse model is widely used in the field of bioluminescence tomography.The physiological anatomy of mouse which can be obtained from segmentations of CT (computedtomography) or MRI (Magnetic Resonance Imaging) can help to reduce the illness of the inverseproblem. However, since it is hard to segment each tissue and organ accurately, modeling becomesvery difficult. According to the problems mentioned above, the main points of this paper aresummarized as follows:(1) This paper transforms the radiation transfer equation into the diffusion equation forsimplification, and uses finite element method (FEM) to solve the forward problem. A point sourcephantom and a ball source phantom are applied to testify the feasibility of FEM with MMCM(Mesh-based Monte Carlo Method) and COMSOL Multiphysics software.(2) This paper solves the inverse problem using the finite element method and applies thepermissible source region method and regularization method to reduce the illness of the inverseproblem. This paper selects the L curve method to choose the regularization parameter and reconstructthe source. The reconstruction results show the more accurate information of the source inside the phantom.(3) This thesis classifies the tissues of digital mouse into many groups to built different opticalmodels and compares the difference in simulation results between models by setting evaluationcriteria. The simulation results find out the optimal model and also provide references for modelingbased on physiological anatomy.