A novel kinetic model is proposed for individual starch granule swelling based on swelling kinetics time series data. The data was acquired from time-lapse microscopy observations of starch granules in a dilute suspension and subsequent image analysis. Chemically modified waxy-maize starch was used. The size evolution of many granules when subjected to various hydro-thermal treatments on a temperature-controlled stage was determined. Observations showed that swelling is not an instantaneous process and varies from a granule to another. A model has been developed which takes into account the effect of temperature evolution and variability between the granules on their swelling. This model can contribute to robust multi-scale simulation of industrial processes involving starch suspensions.