Diffuse reflectance spectroscopy offers a noninvasive, fast, and low-cost alternative to visual screening and biopsy for
skin cancer diagnosis. We have previously acquired reflectance spectra from 137 lesions in 76 patients and determined
the capability of spectral diagnosis using principal component analysis (PCA). However, it is not well elucidated why
spectral analysis enables tissue classification. To provide the physiological basis, we used the Monte Carlo look-up table
(MCLUT) model to extract physiological parameters from those clinical data. The MCLUT model results in the
following physiological parameters: oxygen saturation, hemoglobin concentration, melanin concentration, vessel radius,
and scattering parameters. Physiological parameters show that cancerous skin tissue has lower scattering and larger
vessel radii, compared to normal tissue. These results demonstrate the potential of diffuse reflectance spectroscopy for
detection of early precancerous changes in tissue. In the future, a diagnostic algorithm that combines these physiological
parameters could be enable non-invasive diagnosis of skin cancer.
|