Rapid Determination of Total Phenol in Leaf Extracts of a Medicinal Plant using Infrared Spectroscopy and Chemometric Methods

Abstract

Infrared spectroscopy combined with chemometrics has been applied for simple and rapid determination of total phenol in leaf extracts of medicinal plants. Phenols were extracted from leaves of medicinal plants by ultrasonication and maceration methods. Fourier transform infrared and near infrared spectra of selected leaf extracts of medicinal plants were correlated with total phenolic content using chemometrics. The chemometric methods used for calibration analysis were partial least square (PLS), principal component regression, and support vector machines regression and for classification analysis linear discriminant analysis (LDA), soft independent modelling of class analogies, and support vector machines classification were applied. The results have shown that the PLS method gives the best calibration model with the coefficient of determination (R2) of 0.9920, whereas the root mean square error of calibration is 2.0049. In addition, R2 of leave-one-out cross validation and 2-fold-cross-validation were 0.9939 and 0.9834, respectively. Linear discriminant analysis showed good results of classification analysis with 100% accuracy. The PLS and LDA models were then applicated to real samples. The paired samples t-test showed that total phenol content measured by NIR and UV-Vis spectroscopy methods gave no significant differences (p > 0.05).