Thermometry-derived quasi-dynamic source modeling for full-course temperature prediction in microwave ablation: A phantom study

Accurate temperature prediction in microwave ablation requires a source model that can reflect the stage-dependent nature of heating. In this study, we developed a thermometry-derived quasi-dynamic effective-SAR framework for full-course temperature prediction. The method estimates an effective source in successive time windows, incorporates diffusion correction, and constructs a compact source model within a physics-based heat-transfer formulation. The framework was evaluated in phantom experiments at 50 W and 60 W using multi-point temperature measurements acquired over 600 s. Compared with six alternative models, including thermometry-based and electromagnetic simulation-based baselines, the proposed approach achieved the best agreement with the measured temperatures at both power levels, with RMSE values of 2.732 °C and 2.587 °C, respectively, while more faithfully reproducing the temperature evolution throughout the heating process. These results support the feasibility of the proposed framework for thermometry-guided full-course temperature prediction in microwave ablation.