Microwave hyperthermia is an emerging non-invasive therapeutic modality for cancer treatment that relies on controlled heating of tumor tissue. Accurate prediction of internal temperature is critical to ensure therapeutic efficacy while avoiding damage to surrounding healthy tissues. This paper proposes a hybrid system identification framework that integrates the Sooty Tern Optimization Algorithm (STOA) with a Forgetting Factor Recursive Least Squares (FFRLS) algorithm to estimate the internal temperature of microwave-heated tissues based on surface temperature measurements. A linear-in-parameters model is developed to capture the dynamic thermal behavior of tissue heating and heat conduction. The STOA is employed to perform global exploration of the parameter space and effectively mitigate premature convergence to local optima, while the FFRLS algorithm provides online refinement and real-time adaptive updating of model parameters. Experimental results demonstrate that the proposed STOA-FFRLS method significantly outperforms the conventional RLS approach in terms of estimation accuracy and numerical stability, yielding substantially reduced modeling errors and residual fluctuations. The validated model facilitates the development of a non-invasive, high-precision thermal monitoring framework for safe and effective microwave hyperthermia treatment.
Sun Mengqing (Wed,) studied this question.