Abstract: Volatile odors of chilled tilapia fillets during different storage periods were collected using an electronic nose, and sensory, microbiological and chemical analyses were performed simultaneously. The results showed that the higher the odor concentration of chilled tilapia fillets, the stronger the sensor signal. Moreover, the sensor signals of the odors revealed good responses to their variations in degrees of freshness. Linear discriminant analysis (LDA) was more effective than principal component analysis (PCA) at distinguishing among tilapia fillets with different degrees of freshness, different total viable counts (TVC) and total volatile basic nitrogen (TVBN) contents, suggesting that electronic nose allows quick evaluation of quality deterioration of chilled tilapia fillets. Loading analysis and correlation analysis demonstrated that sensors W2W, W1W, W1S and W2S greatly could make more contribution to discriminating among tilapia fillets with different degrees of freshness and were well correlated with sensory acceptability, TVC or TVBN. Thus, these sensors are potentially useful for further development of fish-specific electronic nose.

Key words: electronic nose, chilled tilapia fillet, quality deterioration, principal component analysis, linear discriminant analysis, loading analysis