AUTHOR=Hu Yuxia , Zhu Yunhao , Hu Dun , Zhou Na , Xiu Lei , Li Weihua , Xie Jiaqi , Zhang Yiming , Yan Pu 

TITLE=Rapid identification of bacteria in water by multi-wavelength transmittance spectroscopy and the artificial neural network

JOURNAL=Frontiers in Environmental Science

VOLUME=Volume 11 - 2023

YEAR=2024

URL=https://www.frontiersin.org/journals/environmental-science/articles/10.3389/fenvs.2023.1274605

DOI=10.3389/fenvs.2023.1274605

ISSN=2296-665X

ABSTRACT=Multi-wavelength transmittance spectroscopy, in combination with artificial neur al network has been a novel tool to identify and classify microorganisms in re cent years. In our work, the transmittance spectra in the region from 200-900n m for four bacterial species of interest, Escherichia coli(E．Coli), Staphylococcu s aureus(S．Aureus), Klebsiella pneumoniae(K．Pneumoniae), and Salmonella typhimurium(S．Typhi), were recorded by using ultraviolet-visible spectrophotometer. Consider too much redundant data of full wave band spectra, the Characteristic w avelength variables were selected using the competitive adaptive reweighting al gorithm (CARS). Spectra of the initial training set of these targeted microorgan isms were used to create identification models representing the spectral variabili ty of each species using four kinds of neural networks (BP、RBF、GRNN and PNN). The blinded isolates spectra of targeted species were identified using the above four identification models. Compared to full-band modelling, after using CARS to screen the wavelength variables, four identification models are establ ished for the preferred 35 characteristic wavelengths, and the prediction perfor mance of four models is obviously improved. Among them, the CARS-PNN m odel is the best, and the identification rates of all targeted bacteria achieved wi th 100% accuracy, and the calculation time is just about 0.04s. The use of CA RS can effectively remove useless information from the spectra, reduce model complexity and enhance model prediction performance. Multi-wavelength transm ission spectroscopy, combined with CARS-PNN method, can provide a new me thod for rapid detection of bacteria in water, and could be readily extended for bacterial microbiological detection on blood and food.