Article ID Journal Published Year Pages File Type
8891467 LWT - Food Science and Technology 2018 8 Pages PDF
Abstract
In this study FT-NIR spectra of 73 extra virgin olive oil (EVOO) samples obtained from 21 different cultivars and 4 geographic regions were used to develop partial least squares regression (PLS-R) models for the rapid quantification of sterols for the first time in the literature. The results of the model validation showed that the total sterol content of the EVOO samples could be predicted with good prediction ability (Rp2 = 0.839, RMSEP = 192 mg/kg, RPD = 2.64). However, the prediction models for the individual sterol forms performed poorly. Except for heptadecanoic and eicosenic acids, models with good prediction ability could be established for the quantification of the major fatty acids found in EVOOs with Rp2 and RPD values ranging between 0.716-0.997 and 2.02-17.6, respectively. Even better model performances were obtained when fatty acids were grouped according to their unsaturation degree as SFA (Rp2 = 0.998, RMSEP = 0.102%, RPD = 21.8), MUFA (Rp2 = 0.997, RMSEP = 0.255%, RPD = 18.7) and PUFA (Rp2 = 0.998, RMSEP = 0.147%, RPD = 25.1).
Related Topics
Life Sciences Agricultural and Biological Sciences Food Science
Authors
, , , , , ,