Cost-efficient wide-bandwidth fluorescence spectroscopy for precise identification of biogenic hazardous substances

Hazardous substances originating from biological sources such as bacteria, coronaviruses, and biotoxins are known to cause environmental contamination and exert deleterious effects on human health. Detection techniques based on fluorescence spectroscopy have manifested substantial potential for the rapid and non-destructive identification of biogenic hazardous substances such as bacteria, coronaviruses, and biotoxins. Notwithstanding, the fluorescence signatures generated by different excitation modes exhibit distinct discriminative ability and measurement cost differences. This study was designed to conduct a comprehensive and systematic analysis of the identification capabilities of 15 biogenic hazardous substances within the framework of the wide-bandwidth fluorescence spectral mode. Three-dimensional fluorescence spectroscopy and synchronous fluorescence spectroscopy were introduced as comparative references, and scrupulous consideration was given to the potential background noise inherent in actual measurement signals. The background noise contamination present in the signals was simulated by artificially adding interference data. Fluorescent data feature processing was implemented via the Principal Component Analysis method. In combination with the Multi-Layer Perceptron model, it enabled spectral identification, while also constructing filters and database enhancement technologies to counteract the impact of noise at varying intensities. For the identification of species and categories under database enhancement, the wide-bandwidth fluorescence spectroscopy mode is comparable to the synchronous fluorescence spectroscopy mode (ranging from 50%–55% and 80%–85% respectively), but both are lower than the three-dimensional fluorescence spectroscopy mode (close to 100%). However, the wide-bandwidth fluorescence spectral mode exhibits a lower degree of excitation-wavelength dependence and a reduced measurement time cost when compared with the synchronous and three-dimensional fluorescence spectral modes. As supercontinuum laser light sources continue to undergo progressive evolution, the wide-bandwidth fluorescence spectral mode is anticipated to promote the identification of biogenic hazardous substances. • Multi-mode fluorescence spectroscopy reveals biogenic hazardous substances. • Spectral identification is implemented via machine learning algorithms. • Identification accuracy demonstrates substantial dependence on the database. • Cost-efficient wide-bandwidth fluorescence excitation exhibits application potential.