Optimising experimental design for MEG resting state functional connectivity measurement

Abstract

The study of functional connectivity using magnetoencephalography (MEG) is an expanding area of neuroimaging, and adds an extra dimension to the more common assessments made using fMRI. The importance of such metrics is growing, with recent demonstrations of their utility in clinical research, however previous reports suggest that whilst group level resting state connectivity is robust, single session recordings lack repeatability. Such robustness is critical if MEG measures in individual subjects are to prove clinically valuable. In the present paper, we test how practical aspects of experimental design affect the intra-subject repeatability of MEG findings; specifically we assess the effect of coregistration method and data recording duration. We show that the use of a foam head-cast, which is known to improve coregistration accuracy, increased significantly the between session repeatability of both beamformer reconstruction and connectivity estimation. We also show that recording duration is a critical parameter, with large improvements in repeatability apparent when using ten minute, compared to five minute recordings. Further analyses suggest that the origin of this latter effect is not underpinned by technical aspects of source reconstruction, but rather by a genuine effect of brain state; short recordings are simply inefficient at capturing the canonical MEG network in a single subject. Our results provide important insights on experimental design and will prove valuable for future MEG connectivity studies.3

Threat of shock increases excitability and connectivity of the intraparietal sulcus

“…In the current study, we chose to address this limitation by extracting two-second time windows prior to each startle presentation to minimize the effect of the blink artifact on our power estimates. However, it has been shown that reliability of the MEG connectivity estimates increases as the duration of the recording increases, and durations of ~10 min or greater may be needed to maximize reliability (Liuzzi et al, 2016). Therefore, using such short intervals did not allow for the ability to obtain reliable estimates of MEG connectivity.…”

Consistency of magnetoencephalographic functional connectivity and network reconstruction using a template versus native MRI for co‐registration

“…The lowest consistency for global parameters averaged across subjects was observed for the MST leaf fraction and diameter, although these still yielded moderate to very good consistency in this relatively small sample. Owing to the higher order nature of these network measures, we expected their consistency to decrease nonlinearly: a basic, robust measure such as relative power will only be slightly affected by reconstruction errors [Beg et al, 2009;L opez et al, 2014;Steinstraeter et al, 2009]. However, these inconsistencies could be augmented in the next step of determining phase correlations.…”

“…However, placing vitamin E capsules at anatomical landmarks may lead to larger co-registration errors than for instance coregistration with surface matching [Adjamian et al, 2004]. As a result, one might argue that differences between template and native MRI co-registrations may have been imperfectly assessed, which may have led to higher consistency between the two approaches in comparison to the use of surface matching or head-casts as a co-registration method [Liuzzi et al, 2016;Troebinger et al, 2014]. Moreover, in the current study we did not use digitized head shapes to warp the template MRI to the subject's anatomy.…”

Measuring robust functional connectivity from resting-state MEG using amplitude and entropy correlation across frequency-bands and temporal scales

“…Several recent studies have found that many commonly used techniques for measuring functional connectivity in MEG lack repeatability between healthy subjects, and even show inconsistency over repeated scans of the same subject [45] [8] [25]. [8] found that the method that gave the most consistent connectivity was oscillatory amplitude envelope correlation (AEC), using symmetric orthogonalisation to remove spurious zero-lag correlation between timecourses due to signal leakage [7].…”