A heat flow approach to functional infrared thermography

Functional infrared thermography is a noncontact approach for intraoperative functional mapping which leverages neurovascular coupling-driven heating of activated cortical areas. Conventional analysis of thermography data relies on demonstrating changes in absolute temperature, which can be an inconsistent marker of brain activity.

This work compares analyzing thermography data through the time derivative of temperature (heat flow) instead of absolute temperature (local heating). Functional maps were created for each patient using both the local heating and local heat flow approaches by calculating a group thermal response function and then correlating this signal to parcellated ROIs on the cortical surface.

The validity of each map was assessed by comparison with direct electrical stimulation, the clinical gold-standard. Maps created using heat flow were marginally improved for hand motor mapping (ROC-AUC 0.96 vs 0.93), but significantly improved for face motor mapping (ROC-AUC 0.89 vs 0.68). Both approaches were comparatively inconsistent in identifying hand sensory areas (ROC-AUC 0.69 vs 0.68).

Functional maps created with the heat flow approach have better overall correspondence to direct electrical stimulation results. In addition, the temperature-based approach is susceptible to activation artefacts outside of positive stimulation sites, which was markedly decreased in the heat flow approach.

Our results demonstrate the importance of applying the temporal derivative when analyzing thermography experiments with long-block designs. Identifying reliable indicators of brain activity is essential for establishing infrared thermography as a brain mapping technique.