Towards a Digital Twin for Microvascular Environment in Head and Neck Cancer
Please login to view abstract download link
We present a computational model designed to simulate oxygen delivery in the microenvironment of head and neck cancer patients and its impact on radiotherapy (RT) outcomes. We first built and calibrated the model using data collected with a sublingual microscope from a cohort of 63 patients. The model comprises a first step of microvascular network generation, a second regarding the microvascular flow simulation, and a final step for oxygen delivery and RT simulation. As a result of these operations, we created a synthetic population of microvascular networks matching the morphological (e.g., capillary density - CD) and functional (e.g., red blood cell velocity) features of the real microvasculature. The results showed good alignment between the synthetic model and clinical data. Next, the model was personalized for nine patients, simulating their specific microvascular networks based on their CD. Overall, the synthetic data successfully reproduced both morphological and functional aspects of microvasculature. We further simulated the oxygen delivery and RT for these cases, considering also the surrounding 3D microenvironment. This step allows us to estimate the patient-specific hypoxic fraction in the microenvironment. The variations in oxygen delivery affected RT outcomes, with more vascularized patients showing better treatment responses. We also evaluated the treatment response, varying the tumor radiosensitivity and the treatment (photon RT vs carbon ions hadron therapy). The proposed workflow enables the development of microvascular digital twins for RT, leveraging data for sublingual microvasculature. Furthermore, these findings emphasize the importance of vascularization and radiosensitivity in replicating the RT outcomes correctly.
