Rapid determination of chemical components and antioxidant activity of the fruit of Crataegus pinnatifida Bunge by NIRS and chemometrics.

OBJECTIVE: To establish a method for quality evaluation of the fruit of Crataegus pinnatifida Bunge, also known as Shanzha, by near-infrared spectroscopy combined with chemometrics. METHOD: Seventy-two batches of Shanzha samples were collected, and the content of total components (flavonoids, phenols and organic acids), monomer components (chlorogenic acid, hyperoside and isoquercitrin), as well as the antioxidant activity of 60% ethanol extract were determined by usual methods. Then, all measured values were correlated with the near infrared spectra of Shanzha, and the partial least squares regression models were established. As to improve the model performance, various methods for spectra pretreatment and wavelength selection were investigated. RESULTS: After optimization, the models obtained the coefficients of determination in both calibration and prediction >0.9, and the residual prediction deviations >3, indicating that the models had good prediction abilities. CONCLUSION: The present method can serve as an alternative to the methods for comprehensive and rapid quality evaluation of Shanzha.

Rapid quality evaluation of Plantaginis Semen by near infrared spectroscopy combined with chemometrics.

OBJECTIVE: To establish a rapid and comprehensive method for the quality evaluation of Plantaginis Semen (PS) by using near-infrared spectroscopy combined with chemometrics to determine the content of geniposidic acid (GEA), verbascoside (VE), isoverbascoside (IVE) and total flavonoids (TF) in PS, as well as the antioxidant activity (AOA). METHODS: The content of GEA, VE and IVE in PS were determined by HPLC, the content of TF in PS was determined by UV-Vis spectrophotometry, and the AOA of PS was characterized by the DPPH, ABTS and FRAP, respectively. Then, the measured values of each item were used as reference values and were correlated with the near infrared spectra of PS. Seven quantitative models were established by the partial least squares regression. A variety of spectral preprocessing, such as standard normal variation (SNV), multiplicative scatter correction (MSC), Savitzky-Golay smoothing (SG), derivative and their combination methods, were investigated. In addition, genetic algorithm (GA), particle swarm optimization (PSO), as well as competitive adaptive reweighted sampling (CARS) were also compared. All samples were divided into a calibration set and a prediction set at the ratio of 3:1 by the descending order of reference values. The coefficients of determination (R2), root mean square error (RMSE) and residual predictive deviation (RPD) were calculated to evaluate model performance. RESULTS: After optimization, the performance of each model was greatly improved, where the R2 for calibration and prediction were both greater than 0.8, the RPD were both greater than 2. Such satisfactory results indicated that the present models had good prediction accuracy. CONCLUSION: Quantitative models based on near infrared spectroscopy were herein established, which proved to be able to quickly and accurately determine the content of GEA, VE and IVE and TF, as well as AOA in PS, and which might provide a new method for rapid and comprehensive quality evaluation of PS.