Personalized predictions and non-invasive imaging of human brain temperature

Abstract

Brain temperature is an important yet understudied medical parameter, and increased brain temperature after injury is associated with worse patient outcomes. The scarcity of methods for measuring brain temperature non-invasively motivates the need for computational models enabling predictions when clinical measurements are challenging. Here, we develop a biophysical model based on the first principles of energy and mass conservation that uses data from magnetic resonance imaging of individual brain tissue and vessel structure to facilitate personalized brain temperature predictions. We compare model-predicted 3D thermal distributions with experimental temperature measured using whole brain magnetic resonance-based thermometry. We find brain thermometry maps predicted by the model capture unique spatial variations for each subject, which are in agreement with experimentally-measured temperatures. As medicine becomes more personalized, this foundational study provides a framework to develop an individualized approach for brain temperature predictions.