Research On Bioluminescence Tomography Based On Multi-level Adaptive Finite Element Method

Bioluminescence tomography(BLT)is an important imaging modality for optical molecular imaging.It reconstructs the three-dimensional distribution of the bioluminescent source inside the birth object through the photon flux measured on the biological surface to reflect the internal pathology feature of the birth object.It is a highly ill-posed inverse problem to reconstruct the three-dimensional distribution of a light source based on only limited measurement data on the surface of the organism.After years of study,researchers have proposed a numerical solution based on Finite Element Method(FEM),but it is difficult to obtain high resolution and high accuracy reconstruction results.In recent years,researchers have successively proposed a series of multilevel adaptive finite element methods(MAFEM)based on FEM and improve the quality of reconstructed light sources significantly.However,in the reconstruction,the method of manually setting the energy threshold is often used to determine the permissible area and there is excessive manual intervention,leading to the position of the reconstructed light source deviating from the real light source,and the positioning error of the light source is excessive.In order to solve the problem that there are too many manual interventions in the MAFEM and lead to inaccurate positioning of the light source,a reconstruction framework of BLT based on multi-level adaptive finite element method and combined with permissible domain contraction(Permissible Region Shrinking-Multilevel Adaptive Finite Element Method,PRS-MAFEM)is proposed in the thesis.The framework introduces the permissible region shrinkage strategy into the reconstruction process of the MAFEM,and it could automatically determine the permissible region of the light source through iterative shrinkage,reduce manual intervention,and realize accurate and fast light source reconstruction by limited measurement data.To verify the effectiveness of the reconstruction framework,simulations of single-source and double-source based on digital mouse model are designed.The results of simulations show that the proposed reconstruction framework can significantly improve the positioning accuracy and energy density of the reconstructed light source.To explore the excellent properties of the PRS-MAFEM in BLT,the reconstruction algorithm parameter sensitivity and the applicability of different reconstruction algorithms are studied.Results show that the PRS-MAFEM is sensitive to the reconstruction algorithm parameters,but in the case of selecting suitable parameters,it could obtain high-quality reconstruction results for a variety of different reconstruction algorithms.The PRS-MAFEM have certain applicability in BLT.