Application of near infrared spectroscopy to predict chemical composition of meat and meat products

Abstract

Near infrared spectroscopy was evaluated as a tool for predicting main chemical constituents (intramuscular fat, protein, water content, water-to-protein ratio) of different raw meats and meat products. Muscle and meat product samples (n=294) were divided into four groups: 1) pig longissimus dorsi muscle, 2) different pork muscles, 3) different muscles of different species and 4) meat and meat products. The quality of the developed models was evaluated using coefficient of determination in calibration (R 2 C) and prediction (R 2 P), standard error in calibration (seC) and prediction (se) and RPD (ratio between standard deviation of the reference data and se). We prepared separate model for pig longissimus dorsi muscle samples and several combined models for various meats and meat products. Best prediction
results were obtained for intramuscular fat content (R 2 P P =0.94-0.99; RPD=4.1-10.1), followed by water content (R =0.67-0.96; RPD=1.2-5.0). Prediction of protein content was also very good (R 2 P 2 P =0.87-0.96; RPD=2.7-4.5), except in a separate sample set of pig longissimus dorsi muscles, which was probably due to narrow variation range. Water-to-protein ratio was also predicted satisfactory accuracy (R
=0.500.91; RPD=1.4-3.1). Developed models proved remarkable ability of near infrared spectroscopy for the prediction of chemical composition of raw meats and meat products and thus the potential to replace existing wet chemistry.