Research On Shape Diffuse Optical Tomography Based On B-spline Curve Description

Diffuse optical tomography (DOT) has received considerable attention which canprovide vast quantities of the functional information of tissues. However, thereconstruction problem of the traditional voxel-based DOT is highly ill-posed, whichseriously affects the algorithm’s accuracy and sensitivity. In this paper, theshape-based image reconstruction algorithm of DOT is proposed based on B-splinecurve description, and the boundary element method is proposed for solving theforward problem.Researches carried out in this thesis are as follows:Firstly, the boundary element method is proposed for numerically computing theforward problem, which derives the boundary integral equation. In addition, theellipsoid parametric description and B-spline curve description are developed todescribe the shape of the organization. Compared with the numerical resultscalculated by the finite element method and Monte Carlo simulation, the boundaryelement method is proved to implement the calculation of light flow effectively.Secondly, the framework of the shape-based DOT image reconstruction algorithmis constructed. Levenberg-Marquardt is employed for solving the least squaresmethod. And optical parameters and shape parameters are reconstructedsynchronously. Compared with the results calculated by the voxel-based imagereconstruction algorithm, the shape-based image reconstruction algorithm is provedthe reliability and superiority in tissue optical imaging.Thirdly, the three groups of the numerical simulation with different shapedescription methods are designed. By the use of ellipsoid parametric description,B-spline curve description and the traditional Fourier series expansion description, thecontrast results validate the convergence ability and the robustness to the noise.What’smore, the influence of various parameters on the reconstruction results are discussed.Finally, the phantom experiments are implemented by the steady-statemulti-channel single photon counting system. The experimental data are processedusing the difference method. By the use of the ellipsoid parametric description andB-spline curve description, image reconstructions are conducted. From thecomparison of the results, both descriptions can be implemented effectively to reconstruct the shape parameters of the inhomogeneity and the optical parameters ofeach region. But the B-spline curve is relatively more flexible in describing theorganizational shape.The shape DOT reconstruction algorithm based on the B-spline curve descriptioncan reconstruct any flexible organization shape. With the advantages of local support,B-spline curve description will promote the combination of the tissue optical imagingand modern medicine detection methods, which will improve the sensitivity and theresolution in clinical diagnosis. In addition, a more reliable quantification image ofthe tumor oxygen characteristic will be received. So this method has a more broadapplication prospects.