321
The road map to correlating instrumental measurements to sensory evaluation of food texture. J.-F.
Meullenet, Assistant Professor. University of Arkansas, Department of Food Science, Fayetteville, AR
72704.
Texture of foodstuff has long been recognized as a multi-dimensional
sensory quality that manifests itself in many ways. Sensory and instrumental tests are often carried out
simultaneously in order to obtain correlations between the two methods. The evaluation of the relationship
between sensory attributes and instrumental parameters often assumes a linear relationship between a single
instrumental measure and a single sensory attribute. However, the failure to examine the possibility of a
non-linear relationship often results in poor statistical correlations. Furthermore, it has been suggested that
the use of several instrumental parameters for the prediction of a single sensory attribute would improve the
accuracy of the predictive models. In this presentation, these issues will be discussed. In particular, the use
of alternative instrumental data treatments, such as Spectral Stress Strain Analysis, will be presented. In
addition, particular emphasis will be placed on the use of multivariate analysis techniques, such Partial
Least Squares Regression and Artificial Neural Networks, for correlating instrumental measurements to the
sensory perception of texture. Specific applications for rice, and dairy and meat products using instruments
such as a Texture Analyzer, a stress control rheometer and a near-infrared spectrophotometer will be
presented.
|
