Detection of urinary tract infection (UTI)-causing bacteria uses conventional time-consuming microbiological techniques. The current need is to use a fast and reliable method of bacterial identification. In order to unambiguously distinguish the UTI-causing five bacterial species used in the current study, micro-Raman spectra were obtained from a home-assembled micro-Raman system and analyzed by multivariate statistical techniques such as principal component analysis (PCA), partial least square-discriminate analysis (PLS-DA), and support vector machine (SVM). Also, the micro-Raman spectra recorded from samples containing two and three bacterial species were tested and validated against the aforementioned calibration models using PLS-DA and SVM. The prediction accuracies of up to 73 and 89% were achieved with PLS-DA and SVM, respectively. Taken together, the present study depicts the capturing of unique micro-Raman spectral features manifesting from the biochemical content of each bacterium. Also, micro-Raman spectroscopy combined with multivariate data analysis can therefore be a reliable and faster technique for the diagnosis of UTI-causing bacteria. Graphical Abstract.
Keywords: Micro-Raman spectroscopy; Multivariate classification; PCA; PLS-DA; SVM; UTI-causing bacteria.