Graph theoretical analysis of resting-state MEG data: Identifying interhemispheric connectivity and the default mode

Abstract

Interhemispheric connectivity with resting state MEG has been elusive, and demonstration of the default mode network (DMN) yet more challenging. Recent seed-based MEG analyses have shown interhemispheric connectivity using power envelope correlations. The purpose of this study is to compare graph theoretic maps of brain connectivity generated using MEG with and without signal leakage correction to evaluate for the presence of interhemispheric connectivity. Eight minutes of resting state eyes-open MEG data were obtained in 22 normal male subjects enrolled in an IRB-approved study (ages 16–18). Data were processed using an in-house automated MEG processing pipeline and projected into standard (MNI) source space at 7 mm resolution using a scalar beamformer. Mean beta-band amplitude was sampled at 2.5 second epochs from the source space time series. Leakage correction was performed in the time domain of the source space beam-formed signal prior to amplitude transformation. Graph theoretic voxel-wise source space correlation connectivity analysis was performed for leakage-corrected and uncorrected data. Degree maps were thresholded across subjects for the top 20% of connected nodes to identify hubs. Additional degree maps for sensory, visual, motor, and temporal regions were generated to identify interhemispheric connectivity using laterality indices. Hubs for the uncorrected MEG networks were predominantly symmetric and midline, bearing some resemblance to fMRI networks. These included the cingulate cortex, bilateral inferior frontal lobes, bilateral hippocampal formations and bilateral cerebellar hemispheres. These uncorrected networks however, demonstrated little to no interhemispheric connectivity using the ROI-based degree maps. Leakage corrected MEG data identified the DMN, with hubs in the posterior cingulate and biparietal areas. These corrected networks demonstrated robust interhemispheric connectivity for the ROI-based degree maps. Graph theoretic analysis of MEG resting state data without signal leakage correction can demonstrate symmetric networks with some resemblance to fMRI networks. These networks however, are an artifact of high local correlation from signal leakage and lack interhemispheric connectivity. Following signal leakage correction, MEG hubs emerge in the DMN, with strong interhemispheric connectivity.

Directed functional connections underlying spontaneous brain activity

“…Power in the delta, theta, alpha, beta and gamma frequency bands measured with M/EEG has been shown to be correlated with BOLD signals or with the RSNs depicted by fMRI (Brookes, Hale, et al, ; de Pasquale et al, ; Laufs et al, ; Mantini et al, ). In addition, the networks derived from MEG (Brookes, Woolrich, et al, ; Maldjian et al, ), low‐density EEG data (19 electrodes) (Chen et al, ), or high‐density EEG (Britz et al, ; Liu et al, ) were similar to RSNs derived from fMRI.…”

Multimodal description of whole brain connectivity: A comparison of resting state MEG, fMRI, and DWI

“…We employed beamforming because it has been successfully applied and recommended for the analysis of resting‐state MEG‐FC in several previous studies [Brookes et al, ,b; Hillebrand et al, ; Hipp et al, ; Schoffelen and Gross, ], despite the method itself assumes that source activations are uncorrelated. Besides, we selected amplitude correlation as the connectivity metric to assess FC, which has been found to reproduce decently fMRI resting state networks [Brookes et al, ,b; Maldjian et al, ], and included recent recommendations for leakage correction [Brookes et al, ; Maldjian et al, ]. Thirdly, it has been recently argued that fMRI time‐series lasting about thirty minutes are required to achieve stability in resting‐state FC [Laumann et al, ].…”

“…Before computing the FC values, a leakage correction was performed in the time domain to control for volume conduction, as suggested in [Brookes et al, ; Maldjian et al, ]. Then, the amplitude of the source time series was extracted with a Hilbert transform, and smoothed with a boxcar filter of 1 second width.…”

Optimising experimental design for MEG resting state functional connectivity measurement

“…̂( ) and ̂( )) may exhibit significant correlation/coherence, purely due to 'signal leakage' between regions. This manifests as a zero-phase lag correlation between the seed and test timecourses and, without careful control, it artifactually inflates estimated connectivity between regions (Maldjian et al, 2014). Signal leakage has been well studied, and a number of methods, including PLI and ICoh, are robust to it since zerophase-lag effects are inherently removed.…”