Relating descriptive analysis and instrumental texture data of processed diced tomatoes

Abstract

The textural properties of two varieties of tomatoes processed in duplicate by three methods (aseptic, cold-fill and hot-fill) were assessed by descriptive analysis and instrumental measurements [Kramer shear press, back extrusion and texture profile analysis (TPA)]. Sensory and instrumental data were compared and correlated using correlation analysis, principal component analysis (PCA), cluster analysis (CA) and partial least squares (PLS) regression analysis. The biplot of PLS2 with all the samples and variables explained 89% of the X variables (instrumental measurements) and 58% of the Y variables (sensory attributes). Sensory firmness was mostly explained by the instrumental firmness measurements such as the area under the curve for the Kramer and back extrusion tests. Samples were separated according to the processing method, with cold-filled samples clustered in the region where firmness variables (sensory and instrumental) were located; and hot-filled and aseptic samples grouped in the region of cohesiveness (TPA measurement), chewy (sensory) and metallic variables. These were consistent with the results from the PCA of the matrix of mean sensory texture ratings and instrumental variables across samples, run on the correlation matrix. Cluster analysis of the sensory matrix clearly classified the samples according to the processing method, whereas cluster analysis of the instrumental matrix did not. We conclude that descriptive analysis provided a more accurate account of the textural properties of diced tomatoes than the instrumental measurements we used and that PLS is the technique of choice for relating sensory and instrumental variables.