Discrimination of Beef Samples by Electronic Nose and Pattern Recognition Techniques Preliminary Results

In this paper a study about the possibility of beef characterization with electronic nose is presented. Three beef classes were compared: Piemontese (PIE), Limousin (FRA) and meat from Argentine (ARG). 150 meat samples were put in glass vials and analysed with a commercial electronic nose instrument based on 10 metal oxide semiconductor sensors. Sensors response of beef classes seemed to be different. Different supervised and unsupervised pattern recognition procedures were applied to sensors signal: principal component analysis (PCA) and partial least square discriminant analysis (PLS-DA). Multivariate analysis pointed out promising classification and prediction results. Three clusters (according to the beef classes) can be clearly discriminated in PCA score plot. Statistical parameters from calibration, validation and prediction of PLS-DA model revealed themselves to be indices of good model. These results demonstrate that electronic nose technology with multivariate analysis models is promising for the rapid determination of differences in meat aroma.