Authentication of grated hard cheeses and quantification of adulteration by FT-NIR spectroscopy and multivariate analysis

This work proposes the use of Near Infrared Spectroscopy (NIR) and multivariate analysis to identify and quantify potential adulterants in grated hard cheeses. The study includes permitted additives (microcellulose and silicon dioxide) at levels higher than those regulated, as well as non-permitted substances like wheat flour, wheat semolina, and sawdust. An authentication approach was developed using One-Class Partial Least Squares (OC-PLS) and Data-Driven Soft Independent Modeling of Class Analogy (DD-SIMCA). Excellent results were obtained by DD-SIMCA using multiplicative scattering correction (MSC), projecting all unadulterated grated hard cheese samples within the acceptance area (i.e., 100% of sensitivity in both the training and test sets) and 77 out 80 adulterated samples outside the acceptance area (i.e., 96% of specificity in the test set). Additionally, satisfactory quantification results were obtained by Partial Least Squares (PLS) for the study of microcellulose, wheat flour, and sawdust, with root mean square error of prediction (RMSEP) values of 1.217, 1.317, and 0.562% w/w, R 2pred values of 0.945, 0.945, and 0.957, and relative error of prediction (REP) values of 4.285, 4.271, and 4.836%, respectively. The proposed methodology provides a simple, rapid, and low-cost analytical tool to identify and quantify adulterations in grated hard cheeses, with the aim to protect consumers from deceptive practices that compromise the nutritional quality of this widely consumed food.