Development of diffuse optical tomography for imaging the brain

The transport of near-infrared light inside biological tissue can be well approximated as a diffusion process. The diffusing photons can be used to detect the global and local variations in the absorption and scattering properties of tissue and offer information such as spatial variation in the oxy-, and deoxy-hemoglobin concentrations of blood. This dissertation presents the author's work in developing method of diffuse optical tomography and clinical studies of detecting hematomas in piglet.; Through simulations, we first show the inaccuracy of near-infrared spectroscopy (NIRS) in quantifying local perturbations in the absorption coefficient of turbid medium and showed that DOT is required for quantitative study of such local changes.; We then studied the use of k-space tomography in reflectance mode in both simulations and experiment. We have developed a fast algorithm for a single source and 2D detector array placed on the tissue surface. The object function is de-convoluted from the background in spatial frequency space. We have shown that this algorithm has the capability of puesdo-3D reconstruction.; We have demonstrated the difficulty of quantitative image reconstruction using linear algorithms by investigating the systematic image errors resulting from background uncertainties and measurement noise. We showed that in determining the changes in concentrations of oxy- and deoxy-hemoglobin, ratio measurement is more accurate than an absolute measurement.; As diffuse optical measurement involves very weak signal changes, DOT systems need to be calibrated before or during the testing. We have developed a model-based calibration method, which enables calibration of the system without reference to a previous measurement. It is based on the assumption that the forward calculation that describes photon distribution inside tissue is accurate. Experiments and simulations on semi-infinite homogeneous turbid medium show that this approach can accurately measure tissue oxygen saturation and greatly enhances the quality of DOT images by reducing errors caused by the coupling problem, which in general is hard to control.; To solve the forward problem of photon distribution inside tissue with heterogeneous structures, we developed a finite difference approach for both the diffusion and transport equations. We studied the effect of a relatively clear material, like cerebral spinal fluid (CSF), together with the highly scattering tissue on the photon distribution inside the brain. We also calculated the 3D photon distribution in a real adult human head. We used the diffusion synthetic accelerated transport method borrowed from neutron scattering physics to speed up the calculation.; Finally, we present the results of our animal study in imaging intracranial hematomas in a piglet. We monitored the temporal development of intracranial hematomas using a CW system. The calibration method that we have developed was applied to process the experimental data and has resulted in greatly enhanced image quality. We also demonstrate a 2D constrained approach for 3D image reconstruction using a limited number of measurements.; Overall, we showed the challenge of quantitative DOT and demonstrated that ratio measurement is more accurate than the absolute measurement. Our calibration approach greatly enhances the accuracy of NIRS and DOT results. We showed through piglet experiments that the cerebral hematomas can be monitored through DOT using a simplified homogeneous semi-infinite model. The 2D constrained approach makes possible 3D reconstruction possible using a limited number of measurements. The finite difference approach we developed can be used to calculate 3D photon distribution inside heterogeneous tissue structures.