Studies On Algorithms For Near-infrared Optical Tomographic Image Reconstruction

Near infrared optical imaging technique is a new imaging modality developed recently for medical purpose where the near infrared light sources are usually used to irradiate the tissues to be imaged. This imaging modality can be employed to obtain images corresponding to the optical parameters such as the scattering and the absorption coefficients. Since these coefficients are closely related to structural and functional changes in the tissues, such a technique can be used especially for early disease diagnosis. This imaging modality has some particular advantages due to its following characteristics: no radiation hazards, continuous monitoring of the patients, and the available of portable instruments. This technique can be potentially used for blood oxygenation measurements, gene therapy and small animal imaging and hence it is of great research value.The contributions of this thesis can be summarized as follows:Firstly, according to the fact that image reconstruction of the fluorescent optical tomography involves repeatedly solving of the large-dimension matrix equations corresponding to the forward and the inverse problems which are computationally expensive, especially for the case when there are large deviations in the optical properties between the target and the reference medium, a wavelet-based multiresolution reconstruction approach was proposed for the fluorescent optical tomographic image reconstruction in combination with a parallel forward computing strategy, in which both the forward and the inverse problem of fluorescent optical tomography was solved in the wavelet domain. Simulation results demonstrate that the proposed approach can significantly speed up the reconstruction process and improve the image quality of the fluorescent optical tomography.Secondly, the techniques of 3D near infrared optical image reconstruction were also studied. The forward problem described by the 3D diffusion equation was solved in the framework of the finite element method. For such a purpose, the volume of the tissue to be imaged was divided into voxels of the tetrahedrons without mutual overlapping. For the inverse problem of the 3D near infrared optical imaging, a conjugate gradient based image reconstruction algorithm was presented. To enhance the adaptability of the algorithm, an improved strategy for the construction of the searching directions was presented and the one dimensional searching process was introduced in the iterations to determine of the step size. By presenting an indirect method to compute the Jacobian matrix based on the Green's functions, the efficacy of the reconstruction was improved. Experimental results show that our algorithm presented for 3D optical image reconstruction in this thesis performs well in both the accuracy and the reconstruction speed.