A fast and non-destructive method to discriminate beef samples using TD-NMR

Abstract

Carr-Purcell-Meiboom-Gill (CPMG) and Continuous Wave Free Precision (CWFP) pulse sequences were used to obtain time-domain 1H Nuclear Magnetic Resonance (TD-NMR) data of beef samples. A total of 99 beef samples from different bull race (Angus, Bonsmara and Canchim) and sex (heifers and steer) were used in this study. The CPMG and CWFP data sets were processed using univariate and multivariate analysis in order to develop classification models to discriminate the samples according to the animal sex and bull race. A paired Student's t test indicated that the univariate models obtained with CWFP data set (using Mz/Mo) had superior performance than those obtained with CPMG (using T2 values). Multivariate models showed similar predictability for both pulse sequences, with an average of correct classification higher than 80% for most of models developed. In general, the results showed that the TD-NMR spectroscopy when combined with univariate and multivariate analysis could provide a valuable tool for tracing the sex and bull race of beef samples, providing a reliable identification in a rapid, relatively cheap and non-invasive way, even in packages.