シソの香気評価に関する研究 シソ葉香気の機器分析データと官能評価データの統計的解析

Abstract

The relationship between gas chromatic (GC) data and sensory evaluation was analyzed in 36 perilla leaf samples by multivariate analysis. Using the external vector model, the quantities of GC peaks were fitted as vectors with the sample configuration obtained by principal component analysis of sensory data. The degree of fitness of the concentration represented by the area under the GC peak with the sample configuration was generally better using the absolute values rather than the relative values. This analysis indicates that terpenoids are important contributors to the sensory attribute of perilla aroma. The finding also reveals that changes in relative quantities of lypoxygenase derivatives, such as (Z)-3-hexen-1-ol and (E)-2-hexenal, following changes in absolute quantities of terpenoids, are important. The GC peaks whose degrees of fitness were statistically significant (P<0.05), show that the directions of vectors generally corresponded to their respective odor characters. Hence, we can conjecture on the influence that the constituents, represented by the GC peaks, have on the sensory properties of the perilla samples by this method. Based on stepwise multiple regression analysis, the absolute values yielded higher accuracy than did the relative values in the 1st principal component (PC), whereas the relative values gave higher accuracy in 2nd PC. The reasonable accuracy of estimation was observed in every multiple regression formula. These results indicate that there is a possibility for estimating the aroma property of fresh perilla leaves objectively by applying multiple regression formula and GC data without using a sensory evaluation.