Application of a potentiometric sensor array as a technique in sensory analysis

This paper reports on the application of a potentiometric sensor array used for monitoring changes in probiotic fermented milk during storage, classification of probiotic fermented milk according to flavor and to accurately predict the results from a human sensory panel. For that purpose the potentiometric electronic tongue consisting of seven sensors and an Ag/AgCl reference electrode was used. The samples of plain, strawberry, apple-pear and forest-fruit probiotic fermented milk were stored during 20 days on two different temperatures and monitored by the electronic tongue and the human sensory panel. Various pattern recognition techniques are adapted including multivariate data processing based on principal components analysis (PCA) for monitoring changes occurring in probiotic fermented milk, artificial neural networks (ANN) for the classification of probiotic fermented milk during storage, partial least square regression (PLS) and artificial neural networks (ANN) to estimate and predict the sensory panel evaluation results. The highest correct classification percentage (97%) was obtained for plain probiotic fermented milk and the lowest (87%) for apple-pear flavored probiotic fermented milk. The highest correlation between the sensor array and the human sensory panel was obtained for the forest-fruit flavored probiotic fermented milk both by using artificial neural networks (0.998) and partial least square regression (0.992). Results from these analyses demonstrate that the electronic tongue can be used to monitor changes in probiotic fermented milk during storage, to classify probiotic fermented milk according to flavor and to predict the sensory characteristics and their relationship to the quality of the probiotic fermented milk measured by consumer.