Identification of herbal medicines using pattern recognition techniques with near-infrared reflectance spectra

초록

The objective of this study was to develop a nondestructive, rapid, and accurate analytical method to identify some herbal medicines that are very similar to the naked eye even though they are different species. Ginseng Radix, Austragali Radix, and Smilacis Rhizoma were used for development of the identification method in this study. Near-infrared/(NIR) reflectance spectra were collected over the region 400 to 2500 nm. With the use of derivatization of spectra the spectral features unique to each species are easier to differentiate. It was shown that NIR reflectance spectra of samples were moderately classified in principal-component space. To select the best identification method for application, pattern recognition techniques using wavelength distance, residual variance, Mahalanobis distance, and soft independent modeling of class analogy (SIMCA) were compared. This study shows that the residual variance method and SIMCA provide the most accurate prediction results. Overall, NIR reflectance spectroscopy using pattern recognition is shown to have significant potential as a rapid and accurate method for identification of herbal medicines. (C) 1999 Academic Press.