Uncertainty analysis in hyperthermia treatment planning for head & neck cancer using polynomial Chaos expansion

Abstract Objective: Hyperthermia, the elevation of tumor temperature to 39 ∘ C–44 ∘ C, is an effective adjuvant treatment for head and neck (H&N) cancer, enhancing the effects of radiotherapy and chemotherapy. This study investigates the robustness of hyperthermia treatment planning (HTP) for H&N cancer using the HyperCollar3D applicator, focusing on uncertainties in patient positioning, tissue properties, and water bolus cooling efficacy. Approach: A retrospective analysis was conducted of 16 patients treated at the Erasmus medical center, utilizing polynomial chaos expansion to model the impact of uncertainties on temperature distributions and treatment quality metrics. Main results: Our findings indicate significant variability in target temperatures due to uncertainties in these tissue properties (2.1 ∘ C T 90 95% confidence interval), further exacerbated by patient positioning errors (2.3 ∘ C T 90 95% confidence interval for 5 mm positioning errors). Uncertainty in dielectric tissue properties causes the largest chunk of the variance (47%) in T 90 followed by positioning errors (22%). Significance: This study highlights the critical importance of accurate measurement of tissue properties and precise patient positioning to achieve effective hyperthermia treatment outcomes. Our findings strongly advocate the development of more robust and quantitative treatment planning and delivery approaches, aiming to enhance the precision and clinical efficacy of HTP protocols for H&N cancer treatments.