The impact of channel density, inverse solutions, connectivity metrics and calibration errors on OPM-MEG connectivity analysis: A simulation study.

Magnetoencephalography (MEG) systems based on optically pumped magnetometers (OPMs) have rapidly developed in the fields of brain function, health, and disease. Functional connectivity analysis related to the resting-state has gained popularity as a field of research in recent years. Several studies have attempted to use OPM-based MEG (OPM-MEG) for brain network estimation research; however, the choice of source connectivity analysis pipeline may lead to outcome variability. Several methods and related parameters must be selected carefully at each step of the analysis. Therefore, this study assessed the effect of such analytical variability on the OPM-MEG connectivity analysis by conducting simulations. Synthetic MEG data corresponding to two default mode networks (DMN) with six or ten DMN regions were generated using the Gaussian Graphical Spectral (GGS) model. Six intersensor spacings were constructed, and six inverse algorithms and six functional connectivity measures were selected to assess their impact on the network reconstruction accuracy. Three potential sources of error - errors in the sensor gain, crosstalk, and angular errors of the sensitive axis of the OPM - were also assessed. Analytical variability with regard to the tested intersensor spacings, inverse solutions, and functional connectivity measures led to high result variability. Crosstalk exerted a significant impact on the accuracy, which may lead to network reconstruction failure. The accuracy improvement caused by an increase in the sensor density may be reduced by gain and angular errors. The minimum norm estimate (MNE) and weighted minimum norm estimate (wMNE) exhibited low robustness to sensor noise and calibration errors. Hence, a calibration workflow for accurate sensor parameters, such as the gain and direction of the sensitive axis, before commencing OPM-MEG measurement and a careful choice of different method combinations play crucial roles in ensuring that OPMs yield optimal results for functional connectivity analysis. A thorough framework for analyzing brain connectivity networks was provided herein.