LAUSR

The safety of genetically modified (GM) food has attracted worldwide attention with a high-frequency appearance in people’s daily life. It is increasingly urgent to find a fast and efficient method to detect GM products to provide reference for safety evaluation. In the current study, we used laser-induced breakdown spectroscopy coupled with chemometrics methods to identify transgenic maize from their non-transgenic parent. One-hundred and twenty GM maize and 120 non-GM maize samples were firstly examined by laser-induced breakdown spectroscopy (LIBS) system. After obtaining the LIBS spectra, principal component analysis (PCA) was introduced to explore the separability of two kinds of samples, and 32 and 30 characteristic emission lines were selected using PCA loadings and weighted regression coefficients (BW), respectively. Classification models based on full spectra and characteristic emission lines were further built by applying partial least squares discrimination analysis (PLS-DA), soft independent modeling of class analogy (SIMCA), and extreme learning machine (ELM). The overall results demonstrated that all models achieved an excellent identification rate, especially the PCA loading-ELM model showed the best performance in the identification of GM maize, with 100% identification accuracy in both calibration and prediction sets. It can be concluded that LIBS combined with chemometrics methods provide a promising way for identification of transgenic maize.