NIR spectroscopy for bacterial identification

Reliable bacterial identification is absolutely necessary procedure for disease diagnosis, food protection and environment screening. Most of current methods rely on phenotypic identification (Colony morphology, Gram stains, and Cells morphology), biochemical (Oxidase and Catalase test), and molecular tests (Polymerase chain reaction, Mass Spectrometry). Although mentioned methods are well-established, time consumption and need of trained personal are a main drawbacks. Since the time is crucial for clinical examinations, rapid technique for bacterial identification is desired. As a first step towards development of classification model based on NIR measurements, we explored the bacterial chemical composition and its impact to NIR spectrum. In contrast with middle-IR spectroscopy, NIR spectroscopy doesn’t require sample drying and complicated preparation. We found the dependencies between bacterial species and NIR peaks positions and amplitudes. Common food pathogens (Escherichia Coli, Listeria, and Salmonella) were analyzed by reflectance and transmittance techniques. As a supportive technique, we use middle-IR spectra of colonies and mass spectra. We observed interpretable spectra acquired from bacteria placed on glass fiber filter. Spectra of different bacterial strains are clearly distinguishable (classification error rate is 0 on limited sample set). Subsequently, we analyzed the liquid samples. Evaluated classification error rate was determined to 15%, but it strongly dependent on bacteria concentration. NIR spectroscopy seems to be possible tool for rapid bacterial identification. The classification power of evaluated model needs to be improved and validated by more samples and techniques. However, the results show clear correlation between bacterial chemical composition and NIR spectra.