Evaluation of different pressing systems for extraction of edible seed oils using NIR Spectroscopy and multivariate data analysis

Edible oils can be considered a valuable component of a healthy diet, due to the high content of unsaturated fatty acids, the presence of important quality micronutrients and their antioxidant properties. However, these valuable parameters, which can be successfully predicted using Near Infrared (NIR) Spectroscopy [1], may be highly affected by the method employed for oil extraction, resulting in different product yields and quality [2]. In this study, NIR spectroscopy coupled to multivariate data analysis was applied as a rapid and non-invasive method to compare edible seed oils extracted using two different kinds of presses, i.e. screw and hydraulic. For both the pressing systems, five kinds of vegetable seeds (hemp, linseed, sunflower, pumpkin, walnut) and two working conditions, related to degree of applied pressure and pressure time, were considered. A whole dataset of 336 NIR spectra was collected on the extracted oil samples in transmittance mode using a FT-NIR benchtop spectrometer (MPA, Bruker Optics). After an initial pre-processing of the NIR spectra using Standard Normal Variate (SNV) algorithm, Principal Component Analysis (PCA) models were built to get an overview of the samples’ distribution in the multivariate space. As a result, the oil samples grouped in well-defined clusters it the PCA score space, according to the type of seed. The PCA models, separately developed for each type of seed, highlighted that both the walnut and the linseed oils samples grouped in well-defined clusters according to the pressing system, suggesting a great effect of this operating parameter on the chemical composition of the extracted oils. Conversely, this effect was not found for the sunflower, pumpkin and hemp oils. Effects related to the degree of applied pressure and pressure time were not observed in any type of seed oils considered.