Control of cod-liver oil composition with laser scattering imaging combined with machine learning procedures: The cases of adulteration and oxidation

Abstract

Chemical changes in cod-liver oil produced by oxidation and adulterations with other oils were modelled using RGB-laser scattering imaging. Two types of composition-altered cod-liver oil were: oxidised oil at three different temperatures (4, 20, and 40 °C) and cod-liver oil adultered with wheat-germ, soybeana, sesame and corn. Both types of altered oils and control samples were analysed by the imaging technique and chemical measurements were also carried out to know the oxidation status. The capacity of the technique for detecting pure cod-liver oil was tested by applying multivariate regression and discriminant procedures based on PLS using the information captured from each type of laser (650 nm, 550 nm and 450 nm). The results showed the capacity of the technique to capture the variability provided by the cod-liver oil against the other vegetable oils used as adulterants. It was discriminate from all vegetable oils and all them were detected when were added as adulterants. Changes in the oxidation status were also modelled by predicting the oxidation parameters with R2 > 0.90, independently of the temperature of storage. Those capacities made it possible to discriminate pure cod-liver from all tested adulterated-oxidised samples performing a common model. The prediction capacity was synergic when models were performed including the three lasers, as opposed to the results obtained with singly-laser models.