Infrared spectrum difference analysis and rapid identification of Paris polyphylla var. yunnanensis from different geographical origin

Abstract

Paris polyphylla var. yunnanensis (PPY) is an important medicinal plant resource, but its quality is greatly affected by the growing environment. There are many mixed uses in the market, but it's difficult to distinguish the good from the bad. Therefore, rapid geographical origin traceability of PPY is of great significance for the safety and efficacy of medication. In this study, through the analysis of conventional Fourier transform infrared spectroscopy (FI-IRS), second-derivative infrared spectroscopy (SD-IRS) and two-dimensional correlation spectroscopy (2DCOS) images, the spectral differences of PPY from different origin were investigated. Hierarchical (HCA) and principal component analysis (PCA) were used to conduct a preliminary exploration of the clustering formation of PPY samples from different places, and then, partial least squares discriminant analysis (PLS-DA), support vector machine (SVM), extreme learning machine (ELM), decision tree (DT), back propagation neural network (BPNN), residual convolutional neural network (ResNet) 6 machine learning algorithms were used to trace the origin of PPY, aiming to provide a rich method reference for the research of PPY from different places. The results showed that FT-IRS could characterize the difference of PPY in different places, PLS-DA, SVM and ResNet all obtained good results, and ResNet model could reach 100 % accuracy. The performance of other models may be related to the size of the data set. The results of this study can promote the rapid quality detection of PPY and provide guarantee for the drug safety of PPY.