Laser induced fluorescence (LIF) detection and discrimination of bacteria from oils, pollen, and chemicals: evaluation of medium sized sample sets and evaluation of classification robustness

Kurzfassung

Laser induced fluorescence (LIF) technology can be applied for instant detection and localization of bacterial aerosol clouds and bacteria on surfaces from distances exceeding 100 m. The technique enables the discrimination of bacteria from other substances in the environment like pollen or chemicals. Therefore, this sensitive technology is an excellent choice for the detection of anomalies. For identification of bacteria classical methods like polymerase chain reaction, DNA sequencing or immunological methods may be used in a second confirmation step. Different oils and bacterial species were excited with laser pulses of wavelengths of 266 nm and 355 nm. The fluorescence data have been analyzed by means of machine learning algorithms. Classification of test data of two classes oils and bacteria resulted in accuracies of 100 %. With more detailed classes (on the level of bacterial species) obtained accuracies were found higher to be than 90% within the set of samples. In addition from the large manifold of relevant samples, a set of 25 different chemicals and bio-agents has been examined under outdoor conditions with laser pulses of 280 nm and 355 nm wavelengths for excitation. The robustness of the LIF detection method has been evaluated: Three different bacterial species were freshly prepared in three different concentrations, repeated on three different days resulting in natural deviations of concentration and metabolic variations. LIF spectral data have been recorded and classification between bacteria and other substances resulted in 99.5% accuracy.