Abstract: In the present paper, the qualitative and quantitative detection of water molecules in water-injected pork were implemented by using low-field nuclear magnetic resonance (NMR) combined with chemometrics. Longissimus dorsi muscles were adulterated with 0%, 5%, 10%, 15%, 20% and 25% of deionized water by intramuscular injection. The peak area proportion of immobilized water (P21) in water-injected pork showed a significantly decreasing trend (P < 0.05), while the relaxation time (T22) and peak area proportion (P22) of free water increased significantly (P < 0.05). Subsequently, a single-variable linear regression (LR) was developed to quantitatively detect the moisture content in pork with rp (correlation coefficient of prediction set) = 0.31, and RMSEp (root mean square error of prediction set) = 4.02%). The optimized rp and RMSEp were 0.45 and 1.79%, respectively. On this basis, a multiple linear regression (MLR) model and a partial least squares regression (PLSR) model were used to quantitatively predict the moisture content in pork. It turned out that the MLR model had better prediction performance (rp = 0.90, RMSEp = 0.57%) than the PLSR model. To sum up, low field nuclear magnetic resonance combined with chemometrics can be an effective method for qualitative and quantitative detection of water molecules in water-injected pork.

Key words: water-injected pork, moisture, low-field nuclear magnetic resonance, qualitative and quantitative detection