Multivariate modelling of the microstructural quality of food emulsions based on NMR

Understanding the relations between sensory/physical parameters and the underlying microstructural features is an essential step for designing and manufacturing novel food products. 'Deductive' strategies to derive such structure-property relationships operate on time-scales, which do not match with the currently required pace of research and development. In this work an 'inductive' approach has been outlined that deploys benchtop spectroscopic NMR and NIR measurements and multi-variate data analysis in order to generate explorative models that relate microstructure and functional parameters. Using protein-stabilised oil-in-water model emulsions, the use of a partial least squares and a multi-linear regression approach for processing and analysis of time-domain NMR data is demonstrated, and benchmarked against the deployment of NIR spectroscopy.