Meat emulsion moisture content monitoring using SWIR hyperspectral imaging system

Abstract The quality of meat emulsions during mixing has not been precisely controlled due to the absence of rapid, objective, and non-destructive measurement techniques. This study aimed to investigate the effect of meat formulation and mixing time on the moisture content of meat emulsions using a short-wave infrared (SWIR, 1000–1700 nm) hyperspectral camera system combined with chemometric modeling. Meat emulsions were prepared with three different fat-to-lean meat ratios(2:6, 3:7, and 4:4) and mixed for up to 15 min(0, 1, 2, 5, 7, 10, and 15 min). Moisture contents differed significantly among formulations, with high meat emulsions (Group A) retaining higher values (63.4%) compared to high fat formulations (Group B, 57.2%; Group C, 52.4%). Mixing time also influenced moisture: initial samples (0 min) showed the highest values (61.8%), while moisture stabilized between 56–58% after 1–15 min, suggesting structural homogenization during processing. Two-way ANOVA indicated significant main effects of formulation ratio and mixing time (p<0.001) but no significant interaction. Spectral analysis identified important wavelengths at 1000–1124, 1190–1230, ~1400, and 1600–1700 nm, corresponding to O–H overtones of water and C–H/N–H absorptions of fat and protein. PLSR and ElasticNet highlighted water-related lower wavelengths, whereas RF emphasized mid- and high-wavelengths reflecting fat and protein signals. The Ensemble model integrated both, achieving the most reliable predictions (R²_P = 0.73–0.75). Hyperspectral imaging further visualized moisture distribution, showing reduced heterogeneity with mixing progression. These findings demonstrate that SWIR spectroscopy, combined with ensemble regression, provides a powerful non-destructive method for predicting moisture content and monitoring mixing uniformity in meat emulsions. This approach highlights the potential of SWIR-based systems for real-time, online quality control in meat processing industries.