Application of Sensory Evaluation, HS‐SPME GC‐MS, E‐Nose, and E‐Tongue for Quality Detection in Citrus Fruits

Abstract In this study, electronic tongue (E‐tongue), headspace solid‐phase microextraction gas chromatography‐mass spectrometer (GC‐MS), electronic nose (E‐nose), and quantitative describe analysis (QDA) were applied to describe the 2 types of citrus fruits (Satsuma mandarins [ Citrus unshiu Marc.] and sweet oranges [ Citrus sinensis {L.} Osbeck]) and their mixing juices systematically and comprehensively. As some aroma components or some flavor molecules interacted with the whole juice matrix, the changes of most components in the fruit juice were not in proportion to the mixing ratio of the 2 citrus fruits. The potential correlations among the signals of E‐tongue and E‐nose, volatile components, and sensory attributes were analyzed by using analysis of variance partial least squares regression. The result showed that the variables from the sensor signals (E‐tongue system and E‐nose system) had significant and positive (or negative) correlations to the most variables of volatile components (GC‐MS) and sensory attributes (QDA). The simultaneous utilization of E‐tongue and E‐nose obtained a perfect classification result with 100% accuracy rate based on linear discriminant analysis and also attained a satisfying prediction with high coefficient association for the sensory attributes ( R 2 > 0.994 for training sets and R 2 > 0.983 for testing sets) and for the volatile components ( R 2 > 0.992 for training sets and R 2 > 0.990 for testing sets) based on random forest. Being easy‐to‐use, cost‐effective, robust, and capable of providing a fast analysis procedure, E‐nose and E‐tongue could be used as an alternative detection system to traditional analysis methods, such as GC‐MS and sensory evaluation by human panel in the fruit industry.