Abstract:
Counterfeit beef is one of the most common issue affecting the Islam society in Thailand. Recently, counterfeit beef made from pork marinated in cow-blood has been used in Thai buffet restaurants in order to reduce cost. This study presents a feasible technique for rapid determination of counterfeit beef by using near-infrared spectroscopy (NIR) combined with multivariate data analysis. Prior to acquiring NIR spectra, the thermal phenomena of meat samples (pork, beef, and marinated pork) were investigated by thermogravimetric analysis (TGA). The thermal profiles of beef and pork show significant differences especially at degradation temperatures of 300°C – 400°C which correspond to the chemical compositions of free amino acids, long chains of peptides and lipids. Molecular information of the meat samples was collected by NIR spectrometer with reflection mode. The spectral pretreatments including Savitzky-Golay polynomial, standard normal variate (SNV) and outlier removal by interquartile range (IQR) were performed in order to remove the irrelevant information. By using spectral variance with 95% significance level, the significant NIR regions to discriminate the type of meat were selected at 5002-5503 cm⁻¹ (the first overtone of O-H stretching of water or moisture content), 5981-6479 cm⁻¹ (the first overtone of C-H stretching of fatty acids/fat), 6819-8207 cm⁻¹ (the second N-H overtone of protein), and 8485-8798 cm⁻¹ (the second overtone of O-H of water). Principal component analysis (PCA) with David-Bouldin index (DBI) was used to visualize the sample cluster. Score plots of PC4-PC5-PC11 were chosen as the best PCs to obtain high separation of meat samples. The correctly classified percentages (%CC) obtained from the Linear Discriminant Analysis was 79.48% and 88.64% for prediction of pork and beef sample, respectively. From overall study, the developed NIR spectrometer with powerful data analysis is a feasible technique for determination of the counterfeit beef with acceptable accuracy.
