Skip to content

MRI sequences

Jump to bottom
Decision, Action, and Neural Computation lab edited this page May 7, 2025 · 6 revisions

MRI sequences

This page outlines the MRI protocols required for laminar MEG studies. MRIs are used for two key purposes:

  1. Headcast Construction: to extract an accurate scalp surface and minimize co-registration error.
  2. Forward modelling: to define dipole location and orientation constraints for source reconstruction via accurate cortical surface extraction.

Overview of Required Sequences

1. Scalp Surface for Head-cast Construction

For accurate head-cast design, a short, motion-robust scan is needed to capture the scalp without skin distortion (Meyer et al., 2017). We recommend:

  • 1.0 mm isotropic T1-weighted MPRAGE
    • Acquisition time: ~3–4 minutes
    • No padding or foam inserts
    • Fat Saturation or Fast Water Excitation is strongly recommended to suppress chemical shift artifacts between lipid and water signals. These can lead to duplicated or displaced skin boundaries, degrading the precision of scalp surface detection (see Fig. 10 in Xiao et al., 2012).

See Headcast Construction for details.

2. Cortical Surface Extraction for Source Modelling

Accurate white and pial surface reconstructions are essential for laminar inference. This requires high-resolution structural images:

  • 0.8 mm T1-weighted MPRAGE
  • 0.8 mm T2-weighted SPACE (or equivalent)

These are combined in FreeSurfer to improve pial surface accuracy.

We previously used 0.8 mm Multi-Parameter Maps (MPMs) (Bonaiuto et al., 2018), but now typically use 0.8 mm T1w + T2w as this improves pial surface quality and reduces motion artifacts.

Note: Fat suppression is also critical here, but for a different reason — to reduce ringing artifacts caused by motion in high-intensity fat voxels, which can obscure GM/WM boundaries and interfere with surface extraction.

3. Multi-Echo FLASH (MEF) for BEM and Scalp Surfaces

If you want to generate Boundary Element Models (BEMs), we recommend:

  • Multi-echo FLASH (MEF) scans
    • 1.0 mm isotropic
    • Two flip angles (e.g., 5° and 30°)
    • 8 echo times (e.g., as in Fischl et al., 2004)
    • Acquisition time: ~9 minutes per flip angle

The MEF can be used to extract scalp/skull surfaces for BEMs and potentially for head-cast design as well.

Recommended Protocol Summary

Purpose Sequence Resolution Duration
Head-cast design T1w MPRAGE 1.0 mm ~3–4 min
Surface extraction T1w MPRAGE + T2w SPACE 0.8 mm ~11 min
BEM construction ME-FLASH (2 flips) 1.0 mm ~18 min

You can also consider MEMPRAGE or MP2RAGE for improved T1 contrast, but scan times are typically prohibitive for 0.8 mm resolution.

Resolution Comparison

Below is a comparison of surfaces extracted from 1.0 mm vs 0.8 mm data. Note the improved smoothness and cortical ribbon separation in the 0.8 mm surfaces.

1mm

surfs_1 0mm

0.8mm:

surfs_0 8mm

Notes

  • Use Fat Saturation or Fast Water Excitation to suppress chemical shift artifacts (for scalp detection) and ringing artifacts from motion (for GM/WM boundary detection).
  • We recommend acquiring T1 and T2 with similar bandwidths to minimize geometric distortions.
  • For Siemens scanners, GRAPPA and CAIPIRINHA can be used to accelerate acquisitions while maintaining SNR.

Additional Considerations

  • Some groups are exploring UTE-based sequences for faster or more accurate skull/scalp imaging. These may offer alternatives to MEF for BEM construction (Deininger-Czermak et al., 2022, Duc Vu et al., 2024).
  • If you are interested in integrating this pipeline with TMS planning or electric field modeling, consider evaluating SIMNIBS, which is designed to extract head models from T1w/T2w images.
  • Synthetic CT generation from DIXON-based MRI may offer another route for estimating skull conductivity distributions — useful in forward modelling and E-field prediction. This is more common in radiotherapy, but potentially adaptable.

References

Bonaiuto et al., eLife (2018): Lamina-specific cortical dynamics in human visual and sensorimotor cortices https://doi.org/10.7554/eLife.33977

Meyer et al., J Neurosci Methods (2017): Flexible head-casts for high spatial precision MEG https://doi.org/10.1016/j.jneumeth.2016.11.009

Xiao et al., Magnetic Resonance Imaging (2012): Multicontrast multiecho FLASH MRI for targeting the subthalamic nucleus https://doi.org/10.1016/j.mri.2012.02.006

Duc Vu et al., Magnetic Resonance in Medicine (2024): Three contrasts in 3 min: Rapid, high-resolution, and bone-selective UTE MRI for craniofacial imaging with automated deep-learning skull segmentation https://doi.org/10.1002/mrm.30275

Deininger-Czermak et al., Journal of Neuroradiology (2022): Evaluation of ultrashort echo-time (UTE) and fast-field-echo (FRACTURE) sequences for skull bone visualization and fracture detection – A postmortem study https://doi.org/10.1016/j.neurad.2021.11.001

With thanks to Franck Lamberton (CERMEP Lyon) for feedback and technical insight.

Clone this wiki locally