Patient-specific microvascular computational modeling for estimating radiotherapy outcomes

This study presents a personalized computational framework for modeling the vascular microenvironment in head-and-neck cancer patients and evaluating the impact of microvasculature on radiotherapy outcomes. We first perform a population-based calibration of a microvascular model using data collected with a sublingual microscope from 62 patients, creating synthetic networks that capture microvascular features with a population-based approach. The calibrated models accurately reproduce key physiological parameters, such as red blood cells velocity, aligning with clinical data. Next, we personalize the model for nine patients, demonstrating that digital patient-specific microvascular networks can replicate individual vascular beds' structural and functional characteristics. Simulations highlight that, while morphological features improve with vascularization, red blood cells velocity is less predictable, revealing the limitations of using capillary density alone to describe microvascular complexity. We then integrate these microvascular models into a 3D virtual microenvironment to simulate oxygen delivery and radiotherapy response. Our results show that higher vascularization enhances oxygenation and reduces hypoxic regions, which correlates with improved tumor control probability. Additionally, our findings demonstrate how the properties of microvascular networks, radiosensitivity, and treatment parameters affect predicted radiotherapy outcomes. Our workflow supports the creation of microvascular digital twins, initialized using patient data from sublingual microscopy.