Potential of NIR hyperspectral imaging for fast fraud detection in minced beef products with oat flour and cornstarch

Minced beef products are widely consumed due to their high protein content and sensory attributes. However, these products are susceptible to adulteration motivated by economic reasons, which is often difficult to detect using conventional analytical methods. In this study, a near-infrared (NIR) hyperspectral imaging system coupled with chemometrics was evaluated for fast and non-destructive detection of fraud in minced beef products containing meat extenders (oat flour and cornstarch). Partial least squares regression (PLSR) models were developed using the full spectral range and a subset of key wavelengths selected through variable importance in projection (VIP) and competitive adaptive reweighted sampling (CARS) variable selection methods. Among the evaluated preprocessing strategies, the PLSR model based on second-derivative spectra exhibited the best predictive performance. Although the simplified CARS-PLSR model (R 2 P = 0.96, RMSEP = 0.94 % w/w, and RPD P = 4.97) was better than the VIP-PLSR model (R 2 P = 0.94, RMSEP = 1.15 % w/w, and RPD P = 4.34), both models demonstrated good performance in quantifying the total content of meat extenders. The use of a reduced number of wavelengths reduces model complexity and increases computational efficiency, facilitating real-time industrial implementation without compromising predictive accuracy, even in real-world fraud scenarios involving various adulteration levels. These results highlight the effectiveness and practical value of combining hyperspectral imaging with chemometrics for rapid authentication and quality control of minced beef products.