Unveiling the origin and quality traits of Angelica sinensis: Hyperspectral imaging combined with chemometrics and information fusion strategies

Abstract

The quality of Angelica sinensis (Oliv.) Diels (AS) is strongly influenced by its geographical origin and the levels of key functional compounds, such as ferulic acid (FeA), levistilide A (LEA), and chlorogenic acid (CCA). This study aimed to develop non-destructive and rapid testing methods using hyperspectral imaging combined with chemometrics and information fusion strategies. Spectral data from the visible and near-infrared (VNIR) and short-wave infrared (SWIR) regions were collected from AS powder and slices. Hyperspectral data from an individual side effectively identified AS origins, with the partial least squares discriminant analysis (PLS-DA) model constructed by second derivative (SD) and data-level fusion, achieving 100 % accuracy in both training and testing sets. For quantitative regression models, back propagation neural network (BPNN) models demonstrated strong performance. Optimal predictions for FeA, LEA, and CCA were achieved using VNIR data (R_p² = 0.9287, RMSE_p = 0.2244, RPD = 3.7285), data-level fusion (R_p² = 0.8302, RMSE_p = 0.7983, RPD = 2.2463), and feature-level fusion using the competitive adaptive reweighted sampling (CARS) algorithm (R_p² = 0.8762, RMSE_p = 0.7241, RPD = 2.6288), respectively. These findings demonstrate the effectiveness of hyperspectral imaging combined with chemometrics and fusion strategies for AS quality assessment, offering good prospects for real-world applications.