Estimating functional connectivity using 2D tangential components in MEG sensor space

Connectivity analysis in magnetoencephalography (MEG) sensor space is commonly used for the preliminary estimation of the functional relationship between cortical areas, but the results are difficult to interpret due to the field spread effect. To improve the interpretability of sensor-level connectivity analysis, we introduce and test a new connectivity measure based on imaginary coherence in this study.

MEG signals from axial gradiometers are subjected to a wavelet transform at a frequency of interest, and are used to reconstruct 2D tangential magnetometer signals. The Euclidean norm of imaginary coherence values obtained from four available pairs between 2D tangential components at two locations is then used to estimate functional connectivity between sensor locations. The use of this new connectivity measure can be extended to 2D planar gradiometer signals or 3D source signals, where the functional relationship between multi-dimensional signals at different locations needs to be quantified as scalar variables.

The proposed method was applied to measured and simulated auditory evoked MEG data. The Euclidean norm of imaginary coherence reliably eliminated the field spread effect and showed increased inter-hemispheric coherence between sensors above the left and the right auditory cortex. The significance of the results was tested by introducing variability in spontaneous brain activities in multi-trial evoked data simulations.

The distribution of imaginary coherence among axial gradiometer signals shows peaks not at the sensors directly above the neuronal current sources, but at sensors with field extreme.