Efficient and rapid assessment of starch gelatinization through intelligent methodologies

Starch transformation during gelatinization significantly influences the water binding capacity, viscosity, and stability of the matrix, making its monitoring crucial for food quality control. In this study, B-Type potato starch was used as the research subject. To address the limitations of traditional microscopy in real-time monitoring, an intelligent system based on Mask R-CNN was developed for the analysis of starch granules. The model, trained on a specific dataset, enables precise segmentation, accurate identification of the grains, and efficient calculation of the degree of gelatinization (DG). The results reveal distinct phase transitions: at approximately 55 °C, the proportion of granules in the first stage decreases by about 5.5 %, with most transitioning to the second stage, while some advance to the third and fourth stages. At 60 °C, the number of grains in the fourth stage increases rapidly, and by 72 °C, all granules have completely transitioned to the final stage. Notably, between 57 °C and 68 °C, grains from all four stages coexist. This automated method demonstrates exceptional accuracy in identifying and segmenting potato starch granules, completing the assessments in less than a second. By integrating smart technologies, this approach provides deeper insights into the gelatinization process of potato starch under varying conditions and paves the way for innovations in food formulation, process optimization, and quality assurance.