Data fusion of Laser-induced breakdown spectroscopy and Near-infrared spectroscopy to quantitatively detect heavy metals in lily

Abstract

Heavy metal detection is important for food safety assurance. Due to the complexity of the matrix, plant foods often affect the accuracy of heavy metal detection by Laser-induced breakdown spectroscopy (LIBS). By using Near-infrared (NIR) and LIBS data, prediction models were established between spectra and corresponding heavy metal content in this study. The data fusion models provided more accurate results compared to the models based on a single method, with prediction accuracy improved to over 0.90. The mid-level data fusion models achieved the best prediction of heavy metals. The optimal prediction models for Zn, Cu and Pb achieved coefficient determination values of 0.9858, 0.9811 and 0.9460, respectively, and root mean square error of prediction (RMSEP) values of 4.3047 mg/kg, 4.9592 mg/kg and 8.3881 mg/kg, respectively. The results show that LIBS combined with NIR data fusion analysis is highly feasible for the rapid detection of heavy metals in lily.