RESUMO
The aim of the present study was to develop a near-infrared reflectance (NIR) spectroscopy rapid method for evaluation of beef quality. Partial least squares (PLS) prediction model for the physic-chemical characteristics such as moisture, fat, protein, pH, color and WBSF in beef was established with good veracity. One hundred fourteen samples from five different parts of beef carcass (tenderloin, ribeye, topside, shin, striploin) were collected from meat packer after 48 h aging. Spectra were obtained by scanning sample from 950 to 1 650 nm and pretreated the model by MSC, SNV and first derivative. Predictive correlation coefficients of physic-chemical parameters in beef were 0.947 2 (moisture), 0.924 5 (fat), 0.934 6 (protein), 0.620 2 (pH), 0.820 3 (L), 0.864 6 (a*), 0.753 0 (b*) and 0.475 9 (WBSF) respectively. Root mean square errors of calibration (RMSEC) were 0.313 3 (moisture), 0.221 0 (fat), 1.243 2 (protein), 0.744 6 (pH), 1.778 3 (L*), 1.394 2 (a*), 1.763 9 (b*) and 1.0743 (WBSF). They were externally validated with additional 30 beef samples. Statistics showed that there was no significant difference between predicted value and those obtained with conventional laboratory methods. The results showed that NIRS is a rapid, effective technique for evaluating beef quality. The predictions for chemical characteristics gave higher accuracy than prediction for physical characteristics.