Studies On Algorithms For Fluorescent Molecular Tomographic Image Reconstruction

Fluorescent molecular tomography (FMT) is a very important imaging modality for molecular imaging. In this imaging modality, a fluorescent probe used is usually used as the contrast agent, which will emit the fluorescent light under the illumination of the excitation sources. The spatial distribution images of the fluorescent optical properties and the optical parameters inside the tissues are reconstructed through the measurements of light on the surface of the tissues in combination with a proper photon propagation model in the tissues. The advantages of this imaging modality lie in its no radiative injurer, simple equipment and low cost. It can be potentially used in many fields such as tumor detection, gene expression analysis, protein molecular detection as well as the detection of the functional changes in bodies and therefore it is of great value for research.In this thesis, a parallel algorithm valid for any Stokes shift is firstly proposed for the two coupled differential equations describing the forward problem of the FMT, which are usually solved in the sequential manner. In this algorithm, the coupled equations are decoupled by introducing a multiplier matrix. Experimental results show that the proposed algorithm can not only improve the speed and the precision of the forward computation, but it can also improve the whole reconstruction efficiency of the FMT. Then, based on the above parallel algorithm, an extension of the Green's function method for the Jacobian matrix computation suitable for two coupled equations is proposed, which can reduce the computational requirements significantly. In addition, a closed form for the Jacobian matrix is derived to further reduce the computational complexity. Finally, a wavelet-based multiresolution reconstruction algorithm is proposed, in which the matrix equation is decomposed using the wavelet transform. Both the speed and quality of reconstruction can be improved through the iterative computation between high and low resolutions.