Rapid identification of peucedanum geographical growing areas through near infrared spectroscopy

Peucedanum geographical origin has significant relevance on its clinical efficacy. In this work, a rapid method of identificatiing peucedanum origin was established through near-infrared spectroscopy. 92 peucedanum samples grown from Anhui, Hubei and Henan province were collected. 61 samples were randomly selected as calibration set and the other 31 samples were as prediction set. Diffuse reflectance near-infrared spectroscopy of peucedanum was recorded, and was preprocessed by first-order differential and autoscale. Then, principal component analysis was applied to extract information; artificial neural network with principal component as input variables and partial least-squares discriminant analysis were used to build models. The results showed that the purpose of identifying the geographical origin of peucedanum was not achieved through the principal component analysis. Artificial neural network achieved 100% identification rate when 7 principal components were taken as input variables. PLSDA method also achieved 100% identification rate when 3 latent variables were taken in model. The VIP scores of the first 3 LVs on wavenumber were different, which suggested that the chemical ingredients in three region had significant difference. it was good way in rapid identifying peucedanum origin through near-infrared spectroscopy.