Diffuse optical tomography with multiple priors

In human tissue, the propagation of near-infrared light beyond a depth of 0.5-cm can be modeled as diffusive photons migrating through randomly distributed particles. The exiting photon measurements are highly sensitive to the variation of hemoglobin content and sensitive to water and adipose content. These enable dynamic, noninvasive quantification of the distribution of tissue physiology. In the past two decades, evolving from the development of pulse oximetry, laser Doppler blood-flowmetry, and near infrared spectroscopy, Diffuse Optical Tomography (DOT) has emerged as a newly established functional imaging modality.; Like most of the other functional imaging modalities, DOT has a limited spatial resolution and is sensitive to measurement noise. In mathematical terms, the inverse problem of DOT is an ill-posed problem. This thesis explores the improvement of DOT image quality, within the framework of regularization for a linear inverse problem, by incorporating spectral and spatial priors in the reconstruction. The author attempted to exploit the flexibility of DOT's instrument design and experimental set-up to validate and use priors provided by other structural image modalities, such as X-ray and MRI. Simulation, phantom and clinical studies were conducted to test the utility of the methods in the context of breast imaging.