Abstract
OBJECTIVE: Focused ultrasound (FUS) is a noninvasive therapy that can ablate tissues with precision. Computational simulations using the Pennes Bioheat Transfer Equation (PBTE) can aid FUS treatments by predicting temperature distributions. However, traditional models assume constant thermal and acoustic properties, potentially oversimplifying tissue behavior during treatments. METHODS: This study integrates temperature-dependent thermal properties (thermal conductivity, specific heat capacity, and perfusion) into finite-difference time-domain FUS simulations. Three scenarios were simulated: (1) homogeneous liver with both high- and low-power sonications, (2) rabbit thigh muscle validated against preclinical magnetic resonance temperature imaging (MRTI), and (3) an extended simulation of the rabbit thigh with temperature-dependent acoustic properties. Comparisons were made against models using constant-properties collected at 25 °C, 20 °C, and/or 37 °C. RESULTS: For high-power liver sonications, temperature-dependent properties increased necrosis (tissue volume exceeding 240 CEM43) by 17.6% and 13% compared to constant-property models at 25 °C and 37 °C, respectively. Low-power sonications had 17-20% lower temperature rises with temperature-dependent properties. In rabbit muscle, temperature-dependent models showed up to 18% difference in necrosis volume, with temperature curves following the trends observed in MRTI. Incorporating temperature-dependent acoustic properties increased the predicted necrosis volume by up to 30%. Updating thermal properties every 2.5 s maintained accuracy (<1% difference in peak temperature) while reducing computational cost by 70%. CONCLUSION: For FUS simulations involving perfusion shutdown in highly perfused tissues (high-power ablations) or involving hyperemia (low-power hyperthermia), incorporating temperature-dependent properties substantially impacts temperature profiles and necrosis predictions. Properties need not be updated every time step to balance simulation accuracy and computational efficiency.