| Near-infrared (NIR) light provides potential for a new approach to non-invasive detection, diagnosis and clinical management of breast cancer. Using NIR spectroscopic imaging techniques, the physiological information about breast tissue composition can be quantitatively estimated, including hemoglobin, water and lipid concentrations, together with scattering properties. In this thesis work, strategies to improve the accuracy of NIR imaging have been explored experimentally and numerically. A novel Ti:Sapphire laser-based frequency domain tomography system was developed to achieve maximum spectral information, using intrinsic phase-locked detection of the signal propagation. The improvement in quantification through addition of more wavelengths was demonstrated in simulations and in tissue-phantom experiments. A hybrid NIR tomography system combining frequency domain and continuous wave spectroscopy approaches was implemented for imaging healthy subjects and women with malignant breast tumors. Adding measurements at wavelengths above 850nm with the continuous wave method significantly improved the accuracy of water and lipid estimation. When used in cancer imaging in vivo, the NIR contrast information is consistent with physiological and pathological values expected in tumor as seen by investigational studies with Magnetic Resonance Imaging (MRI) and pathology analysis. |
