The need for novel techniques of rapid identification of pathogenic microorganisms arises from the massive spread of drug-resistant nosocomial strains and the emergence of centers for biohazard control. Fourier-transform infrared spectroscopy is a promising alternative to mass spectrometry as it is cost-effective, fast and suitable for field use. The aim of this work was to propose an algorithm for the identification of microorganisms in pure cultures based on the analysis of their Fourier transform infrared spectra. The algorithm is based on the automated principal component analysis of infrared spectra. Unlike its analogues described in the literature, the algorithm is capable of identifying bacteria regardless of the culture medium or growth phase. The training sample included the most prevalent causative agents of infections and sepsis in humans: Staphylococcus aureus (n = 67), Enterococcus faecalis (n = 10), Enterococcus faecium (n = 10), Klebsiella pneumoniae (n = 10), Escherichia coli (n = 10), Serratia marcescens (n = 10), Enterobacter cloacae (n = 10), Acinetobacter baumannii (n = 10), Pseudomonas aeruginosa (n = 10), and Candida albicans (n = 10). The model we built successfully passed a series of blind tests involving clinical isolates of 10 methicillin-resistant (MRSA) and 10 methicillin-sensitive (MSSA) Staphylococcus aureus strains as well as pair mixes of these cultures with clinical isolates of Pseudomonas aeruginosa, Escherichia coli, and Klebsiella pneumoniae.
Identification of Escherichia coli and Trueperella pyogenes isolated from the uterus of dairy cows using routine bacteriological testing and Fourier transform infrared spectroscopy
