Acta Scientiarum Polonorum Technologia Alimentaria

Background. FTIR spectroscopy is becoming an important tool in the differentiation and identification of bacteria. In the present study, lactic acid bacteria of the genus Leuconostoc were differentiated and identified with the use of Fourier transform infrared spectroscopy (FTIR) and artificial neural networks (ANNs). The aim of the study was to expand the existing library of FTIR spectra of lactic acid and propionic acid bacteria, and to develop multilayer artificial neural networks as part of the same structure.

Material and methods. The material for this study were 10 reference strains of the genus Leuconostoc, and 24 strains isolated from food products. The isolated pure cultures were identified with species specific pairs of primers by PCR technique, as a reference method. Bacterial strain samples were subjected to spectroscopic measurements by the transmission method at a wavelength of 4000 cm^-1 to 500 cm^-1 using a FTIR spectrophotometer. Digitized spectral data were submitted to neural networks training, until an error of less than 0.05 was obtained and than used for identification of isolates.

Results. The utility of neural networks has been determined based on the identification of 10 reference strains and 24 bacterial strains of the genus Leuconostoc isolated from food products. The isolated strains have been identified by PCR-based method using species-specific primers. The use of artificial neural networks in FTIR spectral analyses as the most advanced chemometric method supported the correct identification of 83-92% bacteria of the genus Leuconostoc at the species level.