Determining the hydration kinetics of food soils using hyperspectral imaging

Effective cleaning of food deposits on surfaces often begins with the absorption of aqueous cleaning solutions into the foulant layer, a process that is conventionally monitored with gravimetric approaches. In this study, we developed a novel method using Short-Wave Infrared (SWIR) Hyperspectral Imaging (HSI) and multivariate analysis to investigate the absorption kinetics of water into food foulants, complemented by the gravimetric method. The food foulant samples, consisting of starch, protein, and fat, were exposed to aqueous cleaning solutions as a function of foulant treatment, surfactant type and concentration, and solution pH. The Peleg model was fitted to characterise the water absorption in each system, with R² in the range of 0.961 to 0.998, depending on system conditions. The degree of water absorption was increased by the presence of an anionic surfactant. A Partial Least Squares Regression (PLSR) model was trained on the gravimetric and hyperspectral data and showed strong predictive performance (R²_Val = 0.975, RMSE_Val = 0.032 g/g_fouant). Pixel-level prediction maps and spatial heterogeneity metrics showed that water uptake was spatially non-uniform, with the greatest within-sample variation occurring at intermediate soaking times. Time series predictions from HSI were fitted to the Peleg equation, which showed a satisfactory agreement with the equations fitted to the experimental data. This work demonstrates that SWIR HSI and PLSR can quantify both bulk water absorption kinetics and local moisture heterogeneity in food soils, highlighting the benefits of non-invasive, spatially resolved water-content prediction compared with gravimetric methods.