Optical system for tablet variety discrimination using visible/near-infrared spectroscopy

Area of Science:

• Spectroscopy
• Cheminformatics
• Botanical analysis

Background:

• Ginkgo biloba L. tablets are widely used in traditional medicine.
• Accurate variety discrimination is crucial for ensuring product quality and efficacy.
• Existing methods for ginkgo tablet identification can be time-consuming and labor-intensive.

Purpose of the Study:

• To develop and validate a visible/near-infrared spectroscopy (Vis/NIRS) based optical system for discriminating between different varieties of ginkgo tablets.
• To establish chemometric models for accurate classification of ginkgo tablet varieties.
• To provide a novel, rapid, and non-destructive method for ginkgo tablet variety identification.

Main Methods:

• Development of an optical system incorporating a light source, beam splitter, sample chamber, and diffuse reflection detector.
• Acquisition of Vis/NIRS spectra (325-1075 nm) from 270 ginkgo tablet samples across nine varieties.
• Application of principal component artificial neural network (PC-ANN) and stepwise multiple linear regression (SMLR) for chemometric analysis and model development.

Main Results:

• The PC-ANN model achieved a high discrimination rate of 91.1% for ginkgo tablet varieties.
• Principal component analysis identified optimal wavelengths (e.g., 481, 458, 466, 570, 1000, 662, 400 nm).
• The SMLR model, using selected wavelengths, yielded a discrimination rate of 84.4%.

Conclusions:

• The developed Vis/NIRS optical system is a viable tool for discriminating ginkgo (Ginkgo biloba L.) tablet varieties.
• This spectroscopic approach offers a new, fast, and non-destructive method for ginkgo tablet quality control.
• The study demonstrates the potential of chemometrics combined with spectroscopy for authenticating herbal products.