Charachterization of human skin by combined photothermal radiometry and diffuse reflectance spectroscopy

In this thesis I combine two optical techniques to enable assessment of structure and composition of human skin in vivo: Pulsed photothermal radiometry (PPTR), which involves measurements of transient dynamics in mid-infrared emission from sample surface after exposure to a light pulse, and diffuse reflectance spectroscopy (DRS) in visible part of the spectrum. The analysis involves simultaneous fitting of measured PPTR signals and DRS spectra with corresponding predictions of a Monte Carlo model of light-tissue interaction. By using a four-layer optical model of skin (epidermis, papillary dermis, reticular dermis and subcutis) I obtain a good match between the experimental and model data when scattering properties of the epidermis and dermis are also optimized on an individual basis. The assessed parameter values correlate well with literature data and demonstrate the expected trends in controlled tests involving temporary obstruction of peripheral blood circulation using a pressure cuff, and acute as well as seasonal sun tanning. The obtained epidermal thickness values were tested by coregistration with a multiphoton microscope. Moreover, I evaluate the potential of this approach for quantitative evaluation of tattoos during laser removal treatment. For this purpose, I apply a three-layer optical model of skin consisting of epidermis, upper dermis, and bottom dermis which includes the tattoo ink. However, the involved iterative optimization of 14 skin model parameters using a numerical forward model (i.e., inverse Monte Carlo - IMC) is computationally very expensive. In order to overcome this drawback, we have constructed a very fast predictive model (PM) based on machine learning. The PM involves random forests, trained on $\sim$9,000 examples computed using my forward MC model.