Qualitative evaluation of Tunisian olive oils using an electronic tongue and chemometric tools: a prospective study

Abstract

Olive oil commercialization has a great impact in the regional economy of several countries including Tunisia. It is a high-value food product, quite prone to frauds. So, it is important to establish analytical techniques that can ensure labelling correctness regarding olive oil quality as well as its origin, namely concerning the olive(s) cultivar(s) used in the production, which is of major importance for monovarietal olive oils. Traditional analytical techniques like those based on chromatography are quite expensive, time consuming, not portable and difficult to implement in-situ, considering the usual harsh environments of the olive industry. In this work, the feasibility of using an electronic tongue as a classification tool for discriminating Tunisian olive oils according to their quality level (i.e., extra virgin olive oil, virgin olive oil or lampante olive oil) and olive cultivar (i.e., Chetoui, Shali and others, according to the label information) was evaluated for the first time. Olive oil quality was assessed quality parameters (free acidity, peroxide values, K232 and K270 extinction coefficients) and on the organoleptic evaluation carried out by a sensory panel parameters (according to the International Olive Council directives). The potentiometric signal profiles recorded with the electrochemical multi-sensor device during the electrochemical analysis of olive oils hydroethanolioc extracts ( 5 min), coupled with linear discriminant models, established based on the most informative sub-sets of sensors, selected by a simulated annealing algorithm, were able to satisfactorily perform olive oils discrimination according to (i) Olive cultivar sensitivities of 88% for leave-one-out cross-validation and mean sensitivities of 79% for the repeated K-folds cross-validation (4 folds with 10 repeats), achieved with a multivariate model based on the information gathered by 20 sensors of the electronic tongue; and, (ii) Quality level: sensitivities of 91% for leave-one-out cross-validation and mean sensitivities of 84% for the repeated K-folds cross-validation (4 folds with 10 repeats), achieved with a multivariate model based on the information gathered by 26 sensors of the electronic tongue. Overall, the results show the satisfactory performance of a potentiometric electronic tongue containing cross-sensitivity lipid membranes as sensors, which may be tentatively a\ributed to the capacity of the electrochemical device in discriminating olive oils with different polar compounds contents, which are related to specific sensory attributes of olive oils such as bitterness and pungency. Furthermore, the present study, concerning Tunisian olive oils analysis using an electronic tongue, confirms the results previously reported in the literature for olive oils from other geographical origins.