Infrared spectroscopy and chemometrics for predicting commercial categories of virgin olive oils and supporting the panel test.

The aim of this study was to create rapid and sustainable instrumental methods for screening virgin olive oils (VOOs) to support the Panel test. The Panel test is the official sensory method used in EU regulations to determine the commercial category of VOOs. The Panel test is based on a time-consuming and expensive approach, so reducing the number of samples to be analysed is crucial. Spectroscopy offers a potential solution for quickly determining VOOs composition and predicting their quality grade. In this context, three spectroscopic techniques were explored: Near-Infrared (NIR), Fourier-Transform Infrared (FT-IR), and Raman spectroscopy. A dataset of 100 VOOs samples, categorized into the three official grades (extra virgin, EVOO, virgin, VOO and lampante, LOO) established in EU, based on the Panel test results, was analysed. An initial analysis of all spectra revealed typical for triacylglycerols molecular vibrations and not good variability between types of samples, indicating low specificity. However, FT-IR data paired with two different Partial Least Squares-Discriminant Analysis (PLS-DA) models - one differentiating LOO from non-LOO (VOO and EVOO) and another distinguishing LOO from VOO - yielded promising results. Cross-validation indicated successful sample classification with percentages ranging from 81% to 96%, in which LOO vs. no-LOO model showed the highest performance. These findings suggest that FT-IR coupled with chemometric analysis holds promise, particularly for discriminating LOO (inedible) from the higher-quality grade VOO and EVOO categories. Further research efforts are needed to possibly make the herein developed models more robust and potentially extend their application to differentiate all three VOO quality grades.