A New Statistical Approach to Describe the Quality of Extra Virgin Olive Oils Using Near Infrared Spectroscopy (NIR) and Traditional Analytical Parameters

Abstract

NIR prediction models are developed for the determination of relevant parameters to evaluate olive oil quality such as acidity (FFA), peroxide value (PV), UV‐absorption at 232 and at 268/270 nm, p‐anisidine value (AnV), isomeric diacylglycerols (DG), and pyropheophytin A (PPP). In addition a new NIR method to estimate the age of olive oil is presented. The relevant wavenumbers are given for the calculation of the parameters and the precision data are presented in comparison to the chemical reference methods. The calibration and validation of the methods are executed with independent data sets (test in test instead of cross‐validation) to cover the wide range of variability of the analytical parameters including the corresponding accurate analytical results from the reference chemical methods. The correctness and accuracy of the developed NIR methods are verified by analyzing certified materials. Finally, a simple statistical approach has been developed to describe the quality of olive oils using the parameters FFA, PV, K232 and K270, DG and PPP. The probability of presence of a sensory defect (100% ≙ 1; 0% ≙ 0) is calculated using the following equation: Pred (BIN) = 1/(1 + exp(−(−9 + 37 × FFA‐0.9 × PV‐2.9 × K232 + 14 × K270 + 3.7 × PPP‐0.17 × DG))). Practical Applications: The use of NIR allows the analysis of different parameters relevant for the quality of olive oil in one run in comparison to the individual chemical reference methods. That allows saving time and money, and a skilled operator is not necessary, making the method interesting for routine analysis. The use of a simple equation developed from the logistic regression using FFA, PV, K‐values, DG, and PPP measured by NIR as variables enables for the first time to describe the sensory quality of olive oils without making sensory testing. This criterion without setting limits for the individual parameters estimates the probability of the presence of sensory defects on basis of chemical parameters. This tool can easily be used to differentiate the two categories extra virgin and virgin olive oils both using the traditional laboratory methods and the corresponding NIR‐methods. In the present study, NIR prediction models are developed for the determination of free fatty acids, peroxide value, UV‐absorption at 232 and at 268/270 nm, p‐anisidine value, isomeric di‐acyl‐glycerols, and pyropheophytin A to evaluate olive oil quality. On basis of NIR, a new method is developed to estimate the age of olive oil. The parameters FFA, PV, K232 and K270, DG and PPP have been used to describe the quality of olive oils by a simple statistical approach.

Importance of some minor compounds in olive oil authenticity and quality

“…As a relatively new approach some spectroscopic methods were also used in DAGs determination. Recently, DAG content of olive oils were predicted from Fourier transform near infrared (FT-NIR) spectroscopic data (Azizian, Wang, Li, & Kramer, 2018;Willenberg, Matthäus, & Gertz, 2019). In addition, a very recent study investigated the use of FTIR and UV-vis spectroscopic methods jointly and separately to estimate DAGs composition of olive oils .…”

QCM Sensor Arrays, Electroanalytical Techniques and NIR Spectroscopy Coupled to Multivariate Analysis for Quality Assessment of Food Products, Raw Materials, Ingredients and Foodborne Pathogen Detection: Challenges and Breakthroughs

“…For instance, NIR spectroscopy has been used for the determination of acidity (free fatty acid content), peroxide value (based on extinction coefficients for hydroperoxides at 232 and 270 nm which negatively correlate to NIR vibrational bands at 1440 nm), aldehyde level (via p-anisidine value), and level of refinement (via pyropheophytin A and isomeric diacylglycerol ratio). In particular, these parameters aid in the discernment of oil treatment, temperature of storage, and age [73], and can assist in quality assessment of waste cooking oils [74]. The latter application is of great importance to the budding biodiesel industry, where triglyceride degradation during routine cooking in open air can result in highly acidic cooking oils, requiring expensive esterification prior to use.…”

The German National Reference Centre for Authentic Food (NRZ-Authent)

“…Several NIRS models for the determination of parameters related to the oil quality have been described, e.g. for the analysis of the acidity, peroxide value, UV-absorption at 232 nm and 268/270 nm, isomeric diacylglycerols or pyropheophytin A (Willenberg et al, 2019).…”

“…The combination of these parameters in a logit regression model allows to calculate the probability of the presence of a sensory defect. As the percentage of accurately predicted samples was found to be 98.4% (Willenberg et al, 2019), this approach might be a helpful tool to identify potential fraudulent samples that have to be checked by a panel and thereby reducing the number of samples for the panels.…”