INNOVATIVE APPROACHES IN THE INTERPRETATION OF DATA FROM ELECTRONIC NOSES APPLIED TO FEED ANALYSIS

Recently, food and also feed research focused their attention on the use of electronic nose (EN) devices as a support for decision-making in the area of product quality and safety. Given that olfactometric techniques have largely been used as fast and cost-effective approach for food authenticity, quality and safety assessment, the applicability of EN to feed analysis, especially to mycotoxins detection in corn and durum wheat (Triticum durum), was investigated in recent studies. The present effort aimed at exploring innovative approaches in interpreting multivariate data coming from a AIRSENSE PEN2 model Electronic Nose. Data were collected in two different trials: a pilot study carried out in a medium-sized Italian feed factory for the identification of aflatoxin contamination in raw material (corn) and the support of quality controls in pelleted feed, and a comprehensive study focused on EN potential as a screening tool to classify durum wheat samples on the basis of their deoxynivalenol (DON) contamination level. The two studies are accurately reported here and the results are detailed and diffusely commented by the author. Advantages and disadvantages of each different model in the interpretation of data are highlighted and discussed. Also a comparison of various pattern recognition approaches for the same experimental protocol is presented, in order to suggest the appropriate algorithm for the application at hand.