Multivariate Statistical Analysis for the Classification of Sausages Based on Physicochemical Attributes, Using Attenuated Total Reflectance-Fourier Transform Infrared (ATR-FTIR) and Inductively Coupled Plasma-Mass Spectrometry (ICP-MS).

Abstract

Sausage is a convenient food that is widely consumed in the world and in Vietnam. Due to the rapid development of this product, the authenticity of many famous brands has faded by the rise of adulteration. Therefore, in this study, principal component analysis (PCA) was combined with chemical analysis to identify 6 sausage brands. Sausage samples were dried and then ground to a fine powder for both instrumental analyses of attenuated total reflectance-Fourier transform infrared (ATR-FTIR) and inductively coupled plasma-mass spectrometry (ICP-MS). Dried measurements of ATR-FTIR was performed directly on the ZnSe crystal, while elemental data were obtained through microwave digestion before the ICP-MS analysis. Principal component analysis (PCA) within the framework software of XLSTAT and STATISTICA 12 was performed on the spectroscopy and elemental dataset of sausage samples. PCA visualized the distinction of 6 sausage brands on both datasets of ATR-FTIR and ICP-MS. The classification on the spectroscopy profile showed that although more than 90% variation of the dataset was explained on the first two PCs, the difference between several brands was not detected as the distribution of data overlapped with one another. The PCA observation of the elemental composition on PC1 and PC3 has separated the sausage brands into 6 distinctive groups. Besides, several key elements contributed to the brands' identification have been detected, and the most distinctive elements are Na, K, Ca, and Ba. PCA visualization showed the feasibility of the classification of sausage samples from different brands when combined with the results of FT-IR and ICP-MS methods. The experiment was able to differentiate the sausages from the 5 brands using multivariate statistics.