Determination of the Moisture Content in Beef Without Weighing Using Benchtop Time-Domain Nuclear Magnetic Resonance Spectrometer and Chemometrics

The models presented in this study were developed by combining time-domain nuclear magnetic resonance (TD-NMR) and chemometrics to predict the moisture content in beef samples from ten different cattle breeds. The results were very promising because the root mean square error for calibration (RMSEC) was 0.6 g/100 g, according to the partial least squares (PLS) or principal component regression (PCR) models. The correlation coefficients between the reference method and the predicted values were 0.96 for both models. No significant differences were found for the predicted values of the chemometric models when a paired Student’s t-test with confidence levels of 95% or 90% was applied. From these results, the moisture content in meat samples of different cattle breeds can be determined in a through-package mode with high analytical frequency, without pre-treatment of the samples or weighing.