Fast bacterial strain identification by laser induced breakdown spectroscopy and neural networks

摘要

A method for rapid bacterial strain identification based on Laser Induced Breakdown Spectroscopy (LIBS) and Neural Networks (NN) is reported. The study includes bacterial strains of the most relevant bacteria causing Hospital Acquired Infections (HAI), i.e. Pseudomonas aeruginosa, Escherichia coli, Klebsiella pneumoniae, Salmonella typhimurium and Staphylococcus aureus. LIBS/NN methodology was evaluated for its capacity to discriminate different bacterial strains using their characteristic LIBS spectra from changes in their elemental composition as a result of genetic variations. The samples were measured for two different days to evaluate the time-dependent classification capacity of the methodology. A successful classification of bacterial strains by the proposed LIBS/NN method, with accuracy above 95%, shows its potential to address the safety and social-cost HAI-related issue.