A thermography-based bioheat time dependent model for non-invasive measurement of face microvascular perfusion

This study presents a non-invasive thermography-based model for the time-resolved estimation of skin microvascular perfusion. Knowledge of microvascular perfusion offers valuable insights into several physiological processes governed by the autonomic nervous system, and particularly by its sympathetic branch. Dysfunctions of the autonomic system are associated with a range of clinical disorders, including cardiac arrhythmias, dizziness, and gastrointestinal disturbances. Consequently, the investigation of autonomic and sympathetic nervous system activity plays a crucial role in the diagnosis and understanding of cardiovascular diseases. The method was experimentally evaluated using resting-condition measurements from 14 participants and dynamic measurements from two additional participants during breath-holding and Cold Pressor Test tasks. The analysis relies on an indirect estimation of microvascular perfusion from skin surface temperature measured by infrared thermography. To model this relationship, an inverse, surface-equivalent formulation of Pennes’ bioheat equation was employed. In the proposed formulation, the transient thermal storage term is retained and estimated from the temporal evolution of the measured skin temperature. Through this inverse bioheat model, the equivalent superficial blood-perfusion velocity is reconstructed from the temperature data acquired by an infrared thermal camera and then converted into a superficial blood mass perfusion flux for map visualization. Thermography provides distinct advantages compared to alternative sensing techniques, as it enables real-time, full-field acquisition of facial temperature distributions. This capability allows, via the modified bioheat model, the continuous estimation of facial blood perfusion over time. Each thermographic frame is converted in real time into a corresponding perfusion map, thereby generating a dynamic representation of microvascular activity. The proposed imaging technique thus delivers an instantaneous and comprehensive overview of facial perfusion dynamics, suitable for subsequent quantitative analyses. Under resting conditions, the mean skin surface temperature across the 14 participants was 33.163 ±0.419 °C, and the corresponding mean model-derived perfusion estimate was 0.24 8± 0.077kgm-2s-1. In the dynamic evaluation, the temperature and perfusion-estimate changes had concordant signs in all nine task-rest comparisons, supporting the responsiveness of the model-derived output to small within-participant thermal variations. Finally, a GUM-based uncertainty analysis has been performed to identify the main uncertainty contributions and evaluate their propagation to the model-derived perfusion estimate.