Effects of Inverse Solutions and Calibration Errors on Localization Performance of OPM-MEG.

Area of Science:

• Neuroimaging
• Biophysics
• Biomedical Engineering

Background:

• Optically pumped magnetometers (OPMs) enable portable magnetoencephalography (MEG) systems.
• Accurate source localization is vital for neuroscience and clinical applications of OPM-MEG.

Purpose of the Study:

• To evaluate the impact of calibration errors (gain, crosstalk, angular misalignment) on OPM-MEG source localization performance.
• To compare the robustness of different OPM array configurations (single-, dual-, tri-axis) and source inversion algorithms.

Main Methods:

• Simulated empirical OPM-MEG data using a biologically plausible model of brain activity.
• Investigated effects of simulated gain error, crosstalk, and angular misalignment.
• Compared localization performance across single-, dual-, and tri-axis OPM configurations and four source inversion algorithms (MSP, EBB, IID, LORETA).

Main Results:

• Gain error had minimal impact; crosstalk and angular error significantly degraded localization accuracy.
• Multiple Sparse Priors (MSP) algorithm showed the best robustness and overall performance.
• Deep neural sources were more susceptible to gain errors.

Conclusions:

• Accurate calibration of OPM arrays, especially dual- and tri-axis, is critical.
• MSP is the recommended inversion method for OPM-MEG due to its robustness.
• To achieve ideal OPM-MEG accuracy, crosstalk must be <2% and angular error <2°.