Rapid Measurement of Antioxidant Properties of Dendrobium officinale Using Near-Infrared Spectroscopy and Chemometrics

Abstract

Dendrobium officinale (D. officinale), often used as a dual-use plant with herbal medicine and food applications, has attracted considerable attention for health-benefiting components and wide economic value. The antioxidant ability of D. officinale is of great significance to ensure its health care value and safeguard consumers' interests. However, the common analytical methods for evaluating the antioxidant ability of D. officinale are time-consuming, laborious, and costly. In this study, near-infrared (NIR) spectroscopy and chemometrics were employed to establish a rapid and accurate method for the determination of 2,2'-azinobis-3-ethylbenzothiazoline-6-sulfonic acid (ABTS) scavenging capacity, 2,2-diphenyl-1-picrylhydrazyl (DPPH) scavenging capacity, and ferric reducing antioxidant power (FRAP) in D. officinale. The quantitative models were developed based on the partial least squares (PLS) algorithm. Two wavelength selection methods, namely the genetic algorithm (GA) and competitive adaptive reweighted sampling (CARS) method, were used for model optimization. The CARS-PLS models exhibited superior predictive performance compared to other PLS models. The root mean square errors of cross-validation (RMSECVs) for ABTS, FRAP, and DPPH were 0.44%, 2.64 μmol/L, and 2.06%, respectively. The results demonstrated the potential application of NIR spectroscopy combined with the CARS-PLS model for the rapid prediction of antioxidant activity in D. officinale. This method can serve as an alternative to conventional analytical methods for efficiently quantifying the antioxidant properties in D. officinale.

Keywords: Dendrobium officinale; antioxidant activity; chemometrics; near-infrared spectroscopy.

Figure 1

Raw near-infrared spectra of all Dendrobium officinale samples. Each line represents the near-infrared spectrum of each sample.

Figure 2

Plots of spectral outlier detection of ABTS.

Figure 3

Histogram of selection for each wavelength after 100 runs by genetic algorithm for ABTS. The blue dashed line indicates the boundary.

Figure 4

Plots of CARS wavelength selection on spectra for ABTS (A), FRAP (C), and DPPH (E). Plots (a–c) show the changing trend in the number of sampled wavelengths, RMSECV values, and the regression coefficient path of each wavelength with increase in sampling runs, respectively. Each line with different color is composed of the regression coefficient values of each wavelength under all sampling runs. Plots of wavelength distribution by CARS for ABTS (B), FRAP (D), and DPPH (F).

Figure 4

Plots of CARS wavelength selection on spectra for ABTS (A), FRAP (C), and DPPH (E). Plots (a–c) show the changing trend in the number of sampled wavelengths, RMSECV values, and the regression coefficient path of each wavelength with increase in sampling runs, respectively. Each line with different color is composed of the regression coefficient values of each wavelength under all sampling runs. Plots of wavelength distribution by CARS for ABTS (B), FRAP (D), and DPPH (F).

Figure 5

Scatter plots of reference values and prediction values for ABTS (A,B), FRAP (C,D), and DPPH (E,F) and using Full-PLS models (A,C,E) and CARS-PLS models (B,D,F).

Figure 5

Scatter plots of reference values and prediction values for ABTS (A,B), FRAP (C,D), and DPPH (E,F) and using Full-PLS models (A,C,E) and CARS-PLS models (B,D,F).