Photonic Data Science

Raman spectroscopy is an emerging tool for fast bacterial identification. However, Raman spectroscopy is depending on suitable preprocessing of the spectra, thereby background removal is a decisive step for conventional Raman spectroscopy. The background has to be estimated, which is challenging especially for high fluorescence backgrounds. Shifted excitation Raman difference spectroscopy (SERDS) eliminates the background through the experimental procedure and holds as promising approach for highly fluorescent samples. Bacterial Raman spectra might be especially complex because these spectra consist of a multitude of overlapping Raman bands from a large multiplicity of biomolecules and only subtitle differences between the species Raman spectra enable the bacterial identification. Here, we investigate the benefits of SERDS compared with conventional Raman spectroscopy specific for the study and identification of bacteria. The comparison utilizes spectra sets of four bacterial species measured with conventional Raman spectroscopy and SERDS and covers three processing approaches for SERDS spectra, for example, the reconstruction with a non-negative least square algorithm.