Predicting Human Brain Temperature

We demonstrate the first personalized predictions of human brain temperature using structural brain data from individuals and a biophysical model derived from first principles.
Predicting Human Brain Temperature
Like

Share this post

Choose a social network to share with, or copy the URL to share elsewhere

This is a representation of how your post may appear on social media. The actual post will vary between social networks

While brain temperature measurements were first reported as early as the mid-nineteenth century, our understanding of human brain temperature is still largely limited by a lack of non-invasive thermometry methods. Recent advances in magnetic resonance imaging (MRI) technology have facilitated whole brain temperature measurements via chemical shift thermometry, but several challenges limit routine clinical application. Given the importance of brain temperature regulation for recovery after stroke, heart attack, and head injury, there is an immediate need to characterize thermal dynamics in the individual brain.

For the past several years the Fleischer Lab, led by Dr. Candace Fleischer at the Emory University School of Medicine, has been developing non-invasive methods for quantitative and absolute brain thermometry. Together with clinical collaborators Dr. Jason Allen and Dr. Fadi Nahab, and biostatistician Dr. Benjamin Risk, our team previously demonstrated non-invasive MR brain thermometry in a primate stroke model and in patients with cerebrovascular disease (AJNR, 2017;38(11):2044; AJNR, 2017, 38(4):712). We observed small increases in temperature are associated with measurably worse outcomes, confirming the role of brain temperature as a potential prognostic marker after injury. While these findings were promising, a limitation of current brain thermometry is the lack of external validation. Implanted temperature probes used in surgery are impractical for research use, and there are currently no routine methods to measure and predict temperature in patients.

To address the challenges in experimental thermometry, predictive biophysical models have been developed to characterize human brain temperature. Most models rely on fundamental principles of bioheat transfer, e.g., Pennes’ bioheat transfer equation (J Appl Physiol, 1948;1(2):93), and progress has been made in generating realistic human anatomy (Phys Med Biol, 1996;41(5):865; Sci Rep, 2018;8(1):7877). Absent from current models, however, is 1) full conservation of energy, mass, and momentum when calculating temperature, and 2) personalized predictions of brain temperature accounting for differences in brain anatomy between individuals. As personalization, or tailoring treatment and care for each patient, is an important goal of contemporary medicine, individual predictions are crucial.

With the goal of developing an improved biophysical model of human brain temperature, our team collaborated with Dr. Andrei Fedorov, an expert in the field of heat transfer and first principles biophysical modeling. Our current manuscript addresses many of the deficiencies in previous models, including adherence to first principles of heat transfer physics as well as the use of individual brain data for personalized predictions. Biophysical modeling using a 3D formulation of thermal physics in the brain by Dongsuk Sung, a graduate student in Dr. Fleischer’s laboratory, and Dr. Peter Kottke, a research engineer in Dr. Fedorov’s laboratory, combined with measured brain anatomy from each subject (brain tissue and vessel structure), enabled detailed predictions of individual brain temperature maps.

Figure 1. Simulated human brain temperature maps. Individual brain tissue and vessel structure acquired with MRI in a healthy human volunteer was used as input data for our fully conserved biophysical model. From left to right: sagittal, axial, and coronal views.

Our model accounts for heat transfer within the brain via conduction, convection, and advection, and was calculated in three physical domains: arteries, tissue, and veins. Individual vessel and tissue structure was obtained from MRI data collected from each subject, and vasculature was augmented using a randomly exploring rapid tree algorithm. Vessel structure maps were used to generate flow rates throughout the blood vessel network, as cerebral blood flow is the primary mechanism for heat dissipation in the brain. Heat generation is the result of cerebral metabolism, and tissue structure maps were used in a similar manner to calculate local metabolic rates. Finally, steady-state equations for each domain were formulated to ensure local and global energy conservation and solved to calculate 3D temperature distributions. The primary finding of this study was the observation that temperature magnitude and spatial patterns differ between individuals, further supporting the need for a personalized approach to brain temperature management. Our modeling efforts were accompanied by experimental MR brain thermometry in healthy humans, and temperature measurements were similar to model predictions. 

Building on these foundational results, future work will simulate the thermal response of the brain in patients after injury such as ischemic stroke. We anticipate brain temperature predictions and measurements will facilitate improved prognostication and patient stratification for emerging treatments, particularly after cerebrovascular injury. Our full manuscript published in Communications Physics is available here.

Please sign in or register for FREE

If you are a registered user on Research Communities by Springer Nature, please sign in

Follow the Topic

Electrical and Electronic Engineering
Technology and Engineering > Electrical and Electronic Engineering

Related Collections

With collections, you can get published faster and increase your visibility.

AI and machine learning in the design and synthesis of crystalline functional materials

This collection hosted by Communications Physics, Nature Communications and Scientific Reports aims to highlight recent advances and applications of AI and machine learning methods in solid-state materials science with a focus on crystalline systems.

Publishing Model: Open Access

Deadline: Dec 31, 2024

The superconducting diode effect

This collection aims to showcase the fundamental physics & current theoretical understanding of the superconducting diode effect and other nonreciprocal phenomena in superconductors.

Publishing Model: Open Access

Deadline: Jan 26, 2025