Assessing suitability of ultra-low-field MRI for TES electric field models

Objective Recent ultra-low-field (ULF) MRI systems provide an option for more accessible and cost-effective imaging to widely used higher field (HF, usually 3T) systems. However, it remains uncertain whether ULF scans can be used for reliable TES current-flow modeling as opposed to using 3T MRI.

We used raw anatomical scans of 23 healthy adults from 64 mT and 3T systems. We enhanced the 64 mT scans using SynthSR. Using our established current-flow modeling pipeline we determined the electric fields (EF) for the ULF and 3T scans induced by the classic M1-SO electrode montage. We quantified the differences in EF at three regions of interest (ROI): left motor cortex (LMC), anterior commissure (AC), and left hippocampus (LHC).

At the cortical ROI (LMC), the individual mean EF based on a 3T scan was found to be 27% higher than the corresponding ULF scan. The average mean relative error was found to be 40%. At AC, the individual mean EF based on a 3T scan continued to be elevated by about 13% in magnitude with respect to the corresponding ULF scan with a mean error range of 31-34%. At LHC, comparison of the induced individual mean EF magnitudes indicated no clear bias between 3T-based and ULF-based models. The average mean relative error at this level dropped to ∼20%.

The substantial decrease in cortical EF when using an ULF scan is predominantly explained by the overestimation of cerebrospinal fluid (CSF) noted by prior morphological comparison studies. Scaling up EF magnitude to match predicted cortical values to a 3T scan may be considered depending on the modeling question being addressed. While preliminary, our results suggest an “equalizing” effect between ULF and 3T predictions with increasing depth within the brain. Competing Interest Statement The authors have declared no competing interest.