Two-Step Curve Fitting Combined with a Two-Layered Tissue Model to Quantify Intrinsic Fluorescence of Cervical Mucosal Tissue in Vivo

Fluorescence spectroscopy (FS) has been used to characterize tissue fluorophores in vivo for the diagnosis of precancers in the uterine cervix. In this study, a two-step curve fitting process is established to extract the intrinsic fluorescence intensity and spectrum of fluorophores including NADH and FAD in the epithelium and collagen crosslinks in the stroma. Forward Monte Carlo (MC) models of diffuse reflectance spectroscopy (DRS) and FS are replaced by artificial neuron networks to improve the computation speed. First, absorption and scattering coefficients of the two tissue layers and the epithelial thickness are estimated from DRS data. Second, the genetic algorithm is used to find the best set of intrinsic fluorescence parameters of the three fluorophores that best fit measured FS data. Results suggest that the two-layered tissue model outperforms conventional homogeneous tissue models in extracted tissue optical properties. Intrinsic fluorescence parameters are extracted from in-vivo spectra measured on 31 subjects.

Automated summary

A two-step curve fitting process is established to extract intrinsic fluorescence intensity and spectrum of tissue fluorophores, showing that the two-layered tissue model outperforms conventional homogeneous tissue models in extracting tissue optical properties. Intrinsic fluorescence parameters are extracted from in-vivo spectra measured on 31 subjects.