Rhubarb (Rhei Radix et Rhizoma) is a classic herbal laxative medicine in Europe and a very famous natural medicine in Asia, especially in China. In this study, near-infrared spectroscopy (NIRS) was first used for rapid and simultaneous analysis of five main active components (chrysophanol, aloe-emodin, rhein, emodin and physcion) in rhubarb of 6 geographical origins, processing and spurious samples. A total of 124 samples (73 raw, 40 processed and 11 spurious samples) were collected. With the reference values determined by HPLC, two calibration strategies, partial least squares (PLS) as a linear regression method and artificial neural networks (ANN) as a non-linear regression method, were studied. For the PLS strategy, 11 spectral pre-processing methods, 5 spectral regions and different latent variables (LVs) were systematically compared, while 3 spectral pre-processing methods and 5 ANN algorithms were studied for the ANN strategy. The results indicated that PLS was more suitable for the analysis of chrysophanol, aloe-emodin, emodin and physcion, whereas ANN was better for rhein. For the optimal NIR models of chrysophanol, aloe-emodin, rhein, emodin and physcion, the correlation coefficients of the calibration set (Rcal) were 0.9916, 0.9762, 0.9839, 0.9794 and 0.9800, respectively; the correlation coefficients of the prediction set (Rpre) were 0.9888, 0.9359, 0.9410, 0.9805 and 0.9785, respectively; the root mean square error of validation (RMSEP) were 0.0402, 0.0197, 0.0593, 0.0133 and 0.0192, respectively. Subsequently, the optimal NIR models were used to study the effects of geographical regions and processing, and identify the spurious rhubarb. Collectively, NIRS may be a well-acceptable method for quality evaluation of rhubarb and other traditional Chinese medicine (TCM).
Keywords: Artificial neural networks; Geographical region; Near-infrared spectroscopy; Partial least squares; Processing; Rhubarb.
Copyright © 2018 Elsevier B.V. All rights reserved.