Artificial neural network-assisted Fourier transform infrared spectroscopy for differentiation of Salmonella serogroups and its application on epidemiological tracing of Salmonella Bovismorbificans outbreak isolates from fresh sprouts

ABSTRACT

Salmonellae represent one of the most common bacterial infection reagents in both humans and animals. For detection and epidemiological elucidation of Salmonella infections, determination of Salmonella serotypes and differentiation between different Salmonella isolates is crucial. In the first part of this study, Artificial Neural Network (ANN)-assisted Fourier transform infrared (FTIR) spectroscopy was used to establish a method for subtyping Salmonella isolates according to their serogroups. For this, 290 Salmonella strains from 35 different serogroups were used to establish an ANN for differentiation between infrared spectra of 10 different Salmonella serogroups (B, C1, C2-C3, D1/D2, E1, E4, F, G, H, O:55) vs. the remaining serogroups. In the final ANN, sensitivity values ranged between 90 and 100% for most of the 10 serogroups under investigation. In the second part of this study, ANN-assisted FTIR spectroscopy was applied for epidemiological distinction of Salmonella Bovismorbificans outbreak isolates from fresh sprouts vs. isolates from other sources. Four Salmonella Bovismorbificans isolates from human and food origin in the context of a Southern German outbreak were successfully discriminated from other S. Bovismorbificans isolates from various sources. ANN-assisted FTIR spectroscopy is thus an effective tool for discrimination of Salmonella isolates at or even below serogroup level.