Characterization of the cortisol biomarker with Raman spectroscopy

Abstract

In this research, the biomarker cortisol was characterized using normal Raman spectroscopy at concentrations of 100%, 50%, 10%, 5%, and 1%. Cortisol solutions were prepared in ethyl alcohol (ethanol). Normal Raman spectra were obtained by varying integration time, photobleaching, signal preprocessing with Whitaker-Henderson smoothing, and fluorescence extraction. Computational tools for spectral analysis included the recovery of the Raman spectrum of cortisol in the presence of ethanol using spectral deconvolution. Additionally, unsupervised machine learning techniques, specifically Principal Component Analysis (PCA), were employed to identify patterns and group spectra of different concentrations based on their characteristics (intensity, spectral shape, and Raman bands). Key results showed the recovery of cortisol spectra with correlation coefficients above 95% for concentrations of 100%, 50%, 10%, and 5%. PCA effectively grouped and differentiated each concentration of cortisol spectra without overlap in the spectral ranges of 200 cm^-1 to 1800 cm^-1, as well as in the region of 1550 cm^-1 to 1650 cm^-1.