Thermometric inversion of dynamic specific absorption rate for bioheat temperature prediction in microwave ablation

In the bioheat model, the specific absorption rate (SAR) quantifies mass-specific power deposition and is a key factor influencing temperature simulation accuracy in microwave ablation (MWA). Conventional thermometric SAR estimation is based on early-time temperature slopes, which can miss the temporal evolution of effective deposition over the entire heating interval. We propose a physics-constrained thermometric inversion framework to identify an effective dynamic SAR using a fast semi-analytical forward solver for stable full-duration prediction. The recovered SAR adopts a compact, mechanism-informed form that combines a baseline spatial envelope with a constrained spatiotemporal modulation to capture heating dynamics. Phantom experiments demonstrate substantially lower global errors and markedly improved cross-sensor consistency than the early-time slope baseline and two electromagnetic-based SAR models. A model-component ablation study further indicates that spatiotemporal modulation is necessary to reproduce the observed build-up and late-stage heating behavior. The dynamic SAR provides a practical and compact heat-source description for rapid temperature prediction in MWA and related thermal therapies.