Effects of head modeling errors on the spatial frequency representation of MEG

Area of Science:

• Biophysics
• Neuroimaging
• Biomedical Engineering

Background:

• Next-generation optically-pumped magnetometers (OPMs) in magnetoencephalography (MEG) offer closer proximity to the head compared to traditional sensors.
• Accurate head modeling is crucial for precise source localization in MEG.
• The impact of head model inaccuracies on OPM-based MEG is not fully understood.

Purpose of the Study:

• To investigate how head model inaccuracies affect signal and source reconstruction accuracy in MEG.
• To assess the influence of sensor array distance from the head on these accuracies.
• To evaluate the importance of head modeling for OPMs.

Main Methods:

• A spherical head model using the boundary element method (BEM) was created and radially perturbed.
• Forward signals were calculated for dipolar sources at various depths using perturbed and unperturbed models.
• Equivalent current dipole (ECD) source localization was performed, and signal/ECD errors were quantified.

Main Results:

• In high signal-to-noise ratio (SNR) conditions, head model inaccuracies increase signal and ECD errors for sensors closer to the head.
• These inaccuracies particularly affect deep and superficial source localization.
• In noisy conditions, the higher SNR from closer sensors outweighs head geometry inaccuracies, improving ECD fit.

Conclusions:

• Head model accuracy is critical for OPM-based MEG, especially in low-noise environments.
• While closer sensor placement enhances SNR, precise head modeling is necessary to leverage OPMs' full potential for accurate source localization.
• Further research should focus on developing robust head modeling techniques for OPM-based MEG systems.