Using near-infrared reflectance spectroscopy to predict physical parameters of beef

The objective of this study was to develop near-infrared reflectance spectroscopy models for predicting the physical parameters of beef from commercial cuts. Four hundred and forty-two minced beef samples from various commercial cuts of two cattle breeds were used for the modeling process and randomly divided into two subsets: a calibration set and an independent prediction set (75 vs. 25%). Reflectance spectra (1000–1800?nm) were collected from both subsets of samples, and calibration models were built using partial least square regression on the calibration set of samples. Different mathematical pretreatments were tested and mean centering or multiplicative scatter correction combined with first-derivative preprocessing gave the best calibration models on all the beef physical traits. Based on the selected calibration equations, the coefficients of determination in calibration and prediction for Warner-Bratzler shear force, pH, color L*, and a* were higher than 0.60, which means models in the calibration were acceptable. However, the ratio performance deviation for all parameters was less than 2.0, indicating that the prediction abilities on independent validation set were not adequate for routine analysis. Further studies are required to establish more robust models for practical application of using near-infrared reflectance spectroscopy to predict physical traits of beef.