Validation of Magnetic Resonance Brain Thermometry and Bioheat Modeling of Brain Temperature Using Direct Measurements With an Implanted Fiber Optic Probe.

Abstract

Background and objectives: Brain temperature is a promising marker of prognosis after injury. Although magnetic resonance (MR) thermometry and biophysical modeling can approximate absolute temperature, limited validation has been performed with direct measurements in a large animal model. The goal of this pilot study was to compare brain temperature measured using MR chemical shift thermometry (CST) and biophysical modeling with that measured using a fiber optic probe in a porcine model.

Methods: MR data were acquired from a Yorkshire pig using a 3 T whole body MR scanner, with simultaneous monitoring of brain and body temperature using a fiber optic probe surgically implanted in the left thalamus and a rectal temperature probe, respectively. MR spectroscopy was acquired for CST. T1-weighted images and MR angiography and venography were acquired for biophysical modeling. Temperature was calculated using the chemical shift difference between water and N-acetylaspartate. Using MR input data of tissue and vessel structures, along with bioheat equations, whole brain temperatures were predicted. Brain temperatures measured using CST, biophysical modeling, and direct measurements from the fiber optic probe, along with rectal temperatures, were compared.

Results: Brain temperature measured using CST was 39.9°C, whereas that measured using the implanted probe was 40.6°C in the left thalamus. Modeled brain temperature, averaged across all voxels in the left thalamus, was 40.1 ± 0.2°C. The difference between brain temperatures measured using CST and modeled temperature was -0.2°C. Brain temperatures from all 3 methods were equal to or higher than the mean rectal temperature during CST (39.9°C, range = 39.6-40.2°C). The difference between brain temperatures measured using the implanted probe and CST was 0.7°C, whereas the difference between probe-measured and modeled temperature was 0.5°C, on the same order as CST (±0.8°C) and probe accuracy (±0.3°C).

Conclusion: Brain temperature quantification using CST and biophysical modeling showed strong agreement with ground truth measurements using a fiber optic probe, providing initial validation of both methods to quantify absolute brain temperature.