Publication: PLS-regression-model-assisted raman spectroscopy for vegetable oil classification and non-destructive analysis of alpha-tocopherol contents of vegetable oils
Issued Date
2021-10-01
Resource Type
ISSN
08891575
Other identifier(s)
2-s2.0-85112846377
Rights
Mahidol University
Rights Holder(s)
SCOPUS
Bibliographic Citation
Journal of Food Composition and Analysis. Vol.103, (2021)
Suggested Citation
Tar Tar Moe Htet, Jordi Cruz, Putthiporn Khongkaew, Chaweewan Suwanvecho, Leena Suntornsuk, Nantana Nuchtavorn, Waree Limwikrant, Chutima Phechkrajang PLS-regression-model-assisted raman spectroscopy for vegetable oil classification and non-destructive analysis of alpha-tocopherol contents of vegetable oils. Journal of Food Composition and Analysis. Vol.103, (2021). doi:10.1016/j.jfca.2021.104119 Retrieved from: https://repository.li.mahidol.ac.th/handle/20.500.14594/75575
Research Projects
Organizational Units
Authors
Journal Issue
Thesis
Title
PLS-regression-model-assisted raman spectroscopy for vegetable oil classification and non-destructive analysis of alpha-tocopherol contents of vegetable oils
Abstract
In this study, a method based on chemometrics and Raman spectroscopy was developed to classify vegetable oils and quantify alpha-tocopherol, the most common vitamin E source in vegetable oils. The Raman spectra of 108 oil samples, obtained from 18 commercial brands and six oil types, were recorded in the scattering mode. The results of classification models, partial least squares–discriminant analysis (PLS-DA), and soft independent modeling of class analogies (SIMCA) showed that all samples were accurately assigned the oil brands and vegetable types. Furthermore, the partial least squares regression (PLSR) model for the determination of the alpha-tocopherol content was established from the Raman spectra of 72 calibration samples modeled with reference values achieved from high-performance liquid chromatography (HPLC). Data from both methods were highly correlated (R2 > 0.95). For the optimum PLSR model, orthogonal signal correction was employed in the data (800–2000 cm−1). Thus, a highly efficient model with 2 latent factors and a good root mean square error of prediction (16.05), was obtained.