@phdthesis{ramella04b, author = {J. C. Ramella-Roman}, title = {Imaging skin pathologies with polarized light: empirical and theoretical studies}, school = {Oregon Graduate Institute of Science and Technology}, year = {2004}, abstract = {The use of polarized light imaging can facilitate the determination of skin cancer borders before a Mohs surgery procedure. Linearly polarized light that illuminates the skin is backscattered by superficial layers where cancer often arises and is randomized by the collagen fibers. The superficially backscattered light can be distinguished from the diffused reflected light using a detector analyzer that is sequentially oriented parallel and perpendicular to the source polarization. A polarized image pol = parallel-perpendicular / parallel+perpendicular is generated. This image has a higher contrast to the superficial skin layer than simple total reflectance images. Pilot clinical trials were conducted with a small hand-held device for the accumulation of a library of lesions to establish the efficacy of polarized light imaging in vivo. It was found that melanoma exhibits a high contrast to polarized light imaging as well as basal and sclerosing cell carcinoma. Mechanisms of polarized light scattering from different tissues and tissue phantoms were studied in vitro. Parameters such as depth of depolarization (DOD), retardance, and birefringence were studied in theory and experimentally. Polarized light traveling through different tissues (skin, muscle, and liver) depolarized after a few hundred microns. Highly birefringent materials such as skin (DOD = 300\,$\mu$m at 696\,nm) and muscle (DOD = 370\,$\mu$m at 696\,nm) depolarized light faster than less birefringent materials such as liver (DOD = 700\,$\mu$ at 696\,nm). Light depolarization can also be attributed to scattering. Three Monte Carlo programs for modeling polarized light transfer into scattering media were implemented to evaluate these mechanisms. Simulations conducted with the Monte Carlo programs showed that small diameter spheres have different mechanisms of depolarization than larger ones. The models also showed that the anisotropy parameter $g$ strongly influences the depolarization mechanism. Large spheres will depolarize faster than smaller spheres if their anisotropy is smaller. A linearly polarized beam traveling through a solution of 1.07\,$\mu$m microsphere ($g = 0.9278$ at 543\,nm) depolarizes slower than in a solution of 2.03\,$\mu$m ($g= 0.8752$ at 543\,nm) microspheres . The Monte Carlo programs were also used to test two heuristic models of polarized light transport into scattering media. The models were based on a heuristic parameter $\chi$ [radians$^2$/Mfp], based on the apparent diffusivity of the angle of orientation for linearly polarized light. Large $\chi$ corresponded to highly depolarizing media, and small $\chi$ correspond to low depolarizing media. A high $\chi$ parameter was found for solutions of small spheres, and for highly birefringent biological media. Larger spheres showed lower values of $\chi$. A novel asymmetric illumination microscope was implemented to investigate precancerous nuclei enlargement. Preliminary experiments were conducted on latex microspheres, the diameter of the sphere could be obtained with only a priori knowledge of the sphere index of refraction.}, }