Urban heat island assessment is increasingly important for rapidly urbanizing delta cities where built-up expansion, water networks and fragmented vegetation interact to shape surface thermal conditions. This study investigates the spatiotemporal variation of land surface temperature and urban heat islands in Can Tho City, Viet Nam, from 2015 to 2025 using Landsat 8 and Landsat 9 Collection 2 Level 2 imagery, Google Earth Engine and Random Forest regression. Annual median composites were generated after cloud and saturation masking, and three land-cover indicators were derived to represent vegetation, built-up surfaces and water bodies. A standardized urban heat island index was calculated from land surface temperature, and 33,542 sample points were extracted for machine-learning analysis. Results show that built-up surfaces expanded from the central urban core toward Cai Rang, Binh Thuy and major transport corridors, while vegetation became more fragmented and the cooling function of minor canals was locally weakened. The citywide mean land surface temperature reached about 35.06 C. Random Forest explained approximately one third of the observed thermal variability, with an error of 2.4 C. Built-up intensity was the strongest driver, accounting for 46.5 percent of feature importance, followed by water and vegetation. The results provide evidence for heat-sensitive urban planning in riverine delta cities.