A machine learning surrogate model for fast approximation of simulated microwave ablation zones

Microwave ablation (MWA) is a minimally invasive therapy for liver, lung, and kidney tumors. Computational modeling with finite element methods (FEM) can simulate ablation zones but is too slow for interactive use. This study develops and evaluates a machine learning surrogate model to rapidly approximate 50 0 C isothermal contours from FEM simulations of MWA, as a first step toward faster computational exploration. A feedforward neural network was trained on 2,625 finite element method (FEM) simulations incorporating tumor geometry and dielectric properties. Predictions were compared with FEM ablation zones for accuracy. The network generated rapid predictions with high correlation (R > 0.95) to FEM contours. Error distributions were centered near zero, confirming robust performance. The FEM machine learning framework eliminates the need for runtime electrothermal simulations, providing a physics-informed framework for rapid computational exploration in MWA treatment planning. The method demonstrates the promising potential of ML-based surrogates to accelerate computational studies in ablation planning.