Fast detection and quantification of pork meat in other meats by reflectance FT-NIR spectroscopy and multivariate analysis

Abstract

This study aimed to develop a fast analytical method, combining near infrared reflectance spectroscopy and multivariate analysis, for detection and quantification of pork meat in other meat samples. A total of 5952 mixture samples from 39 types of meat were prepared in triplicate, with the inclusion of pork at 0%, 1%, 5%, 10%, 30%, 50%, 70%, 90% and 100%. Each sample was scanned using an FT-NIR spectrophotometer in the reflection mode. Spectra were collected in the wavenumber range from 10,000 to 4000 cm^-1, at a resolution of 2 cm^-1 and a total path length of 0.5 mm. Principal Component Analysis (PCA) revealed the similarities and differences among the various types of meat samples and Partial Least-Squares Discriminant Analysis (PLS-DA) showed a good discrimination between pure and pork-spiked meat samples. A Partial Least-Squares Regression (PLSR) model was built to predict the pork meat contents in other meats, which provided the R² value of 0.9774 and RMSECV value of 1.08%. Additionally, an external validation was carried out using a test set, providing a rather good prediction error, with an RMSEP value of 1.84%.

Keywords: Near infrared reflectance spectroscopy; PCA; PLS-DA; PLSR; Pork meat.