Application of fluorescence spectra based on radial basis function neural network in identification of bacteria

Abstract

The results of research on fluorescence spectra of three kinds of probiotic bacteria (lactobacillus acidophilus, streptococcus mutans and lactobacillus bulgaricus) show that the bacteria emit fluorescence when irradiated by ultraviolet light. The spectra are in the range of 300—650　nm with the excitation wavelength of 290　nm. Wavelet transform is used to compress the 150 groups of spectra data, and the number of data points in each group is reduced from 1341 to 168, which not only keeps the character of original spectra unchanged, but also improves the processing speed of neural network. Radial basis function neural network is applied for processing the compressed data. 40 groups of data of each strain are used to be trained, and the other 30 groups of data of which the kinds of bacteria are not given are used for prediction. The result shows that radial basis function neural network can identify the unknown bacterial strain accurately.