Connectome Constrained Graphical Models of MEG Coherence

Structural connectivity by axonal fiber bundles provides the backbone for communication between neural populations. Since axonal transmission occurs on a millisecond time scale, measures of M/EEG functional connectivity sensitive to phase synchronization in a frequency band, such as coherence, are expected to reflect structural connectivity. We develop a complex-valued Gaussian Graphical Model (cGGM) of MEG coherence whose edges are constrained by the structural connectome. The cGGMs edge strengths are summarized by partial coherence, a measure of conditional dependence. We made use of the adaptive graphical lasso (AGL) to fit the cGGMs which allows us to perform inference on the hypothesis that the structural connectome is reflected in MEG coherence in a frequency band. In simulations, we demonstrate that the structural connectivitys influence on the cGGM can be inferred using the AGL. Further, we show that fitting the cGGM is superior to alternative methods at recovering the structural connectome. Graphical modeling of MEG coherence is robust to the source localization estimates required to map MEG from sensors to the cortex. Finally, we show how cG-GMs can be used to explore how distinct parts of the structural connectome contribute to MEG coherence in different frequency bands. We think the cGGM is a useful tool that can improve interpretation of MEG coherence by making a direct link to the structural connectome.