Indexed on: 10 Jan '21Published on: 27 Nov '20Published in: Journal of near infrared spectroscopy
Journal of Near Infrared Spectroscopy, Ahead of Print. In recent years, near infrared (NIR) spectroscopy has gained interest as a tool for bacteria strain identification. Although some promising results suggest good applicability of the technique, a better interpretation of the NIR bacterial spectra is still needed. In order to analyze the NIR spectrum of biological samples, a correlation analysis between the NIR and the mid-infrared (mid-IR) spectra was performed. In total, 28 spectra of 8 bacterial strains were acquired and correlated in the NIR and the mid-IR spectral ranges. Some molecular bands (Amide I, P = O stretching, C-H stretching/deformation of polysaccharides) were well correlated, and the effect of concentration changes in these molecules were investigated. Moreover, a model for the NIR spectra classification was created with an overall 85% correct classification rate. Subsequently, only NIR wavelengths with high correlation to important mid-IR peaks were selected. This led to an increase in the correct classification rate to 94%. By correlation between well-established mid-IR peaks and NIR spectra, some relationships in the NIR spectra of biological samples were revealed, which was a step towards better understanding and interpretation of the NIR spectra of biological samples.