Determination of saffron adulteration thorough the package using Vis-NIR hyperspectral imaging and chemometric techniques

Food authenticity is a very important field of research which enables consumers to get what they pay for and increases their overall level of trust of food. The occurrence of multiple widespread incidents of food fraud over the past decade has increased attention to issues of food authenticity as well as focus by both the food industry and regulatory agencies. Saffron is one of the most important spices with different properties ranging from food coloring and flavoring properties to health benefits. However, because of its limited production and high price, it is always a good candidate for adulteration [1,2]. Hyperspectral Imaging (HSI) has been proposed in recent decade as a reliable tool for food authenticity. This technique combines spectroscopy and imaging techniques, enabling the acquisition of hyperspectral images. These images are composed of reflectance pixel maps containing hundreds of channels, one per wavelength [3]. To the best of our knowledge, there is no study in the literature for saffron authenticity and adulteration detection with HSI and chemometric techniques. Therefore, the objective of the present study was combination of Vis-NIR HSI with chemometric techniques for saffron authentication and adulteration detection (safflower, calendula and style) in intact form thorough the package. Three different packages of thin and thick plastic envelope and poly-crystal were tested. For this purpose, HSI images of 38 intact authentic saffron samples were recorded in spectral range 400-950 nm. Then, multivariate curve resolution-alternating least squares (MCR-ALS) was used for obtaining pure spatial and spectral profiles of the components and then, principal component analysis (PCA) was applied to the resolved spatial profiles of saffron to find possible patterns of authentic samples. According to the PCA results, six authentic saffron samples were selected to make pooled sample for further analysis. Binary mixture of pooled saffron sample and safflower, calendula and style adulterants were prepared in five levels (5, 10, 15, 25, 35% (w/w)). Partial least squares-discriminant analysis (PLS-DA) algorithm was used for the classification of MCR-ALS resolved spatial profiles of saffron. The PLS-DA performance was evaluated using sensitivity (sen), specificity (spe) and accuracy (acc) which were 100 % for calibration, cross-validation and prediction sets using three different saffron packages. Finally, a dataset composed of pooled saffron samples and three adulterants in different levels was designed using D-optimal design and analyzed by MCR-ALS, PCA and PLS-DA methods. Again, the good values of sen, spe and acc for PLS-DA were obtained which confirmed the validity of the proposed method. It is concluded that the developed model can be used as a routine method to check the authenticity of intact saffron through the package.