An Introduction to MEG Connectivity Measurements

NetOnZeroDXC: A package for the identification of networks out of multivariate time series via zero-delay cross-correlation

“…The delay τ , which has to be assigned as a positive integer number, and is therefore expressed in sampling time units, must be much smaller than the time scale to investigate. An example is provided by MEG analysis, in which the source leakage phenomenon spoils direct evaluations of r(k, w) [9]. It has to be noted that the use of a delay changes the value of the time series length to be used in the m max definition above from Λ to Λ − τ .…”

Brain‐wide network analysis of resting‐state neuromagnetic data

“…However, the magnetic field recorded by each sensor may be a mixed signal generated in multiple functional regions, whereas multiple sensors may include a signal from a common source. Since the computation of typical graph metrics requires all‐to‐all connectivity in the network of interest, a graph‐theory analysis in the sensor space may overestimate connectivity, called spurious connectivity, by changes in the amplitude of sources (Brookes, Woolrich, & Price, 2014 ; Hipp et al, 2012 ). A source‐space analysis may reduce the estimation bias in connectivity and graph metrics by separating overlapping source signals, but still has source leakage due to the ill‐posed nature of the inverse problem, inaccuracies in the forward solution, and incorrect assumptions caused by the inverse localization algorithm used (Brookes, Woolrich, & Price, 2014 ), which may lead to spurious connectivity in the source space.…”

“…Since the computation of typical graph metrics requires all‐to‐all connectivity in the network of interest, a graph‐theory analysis in the sensor space may overestimate connectivity, called spurious connectivity, by changes in the amplitude of sources (Brookes, Woolrich, & Price, 2014 ; Hipp et al, 2012 ). A source‐space analysis may reduce the estimation bias in connectivity and graph metrics by separating overlapping source signals, but still has source leakage due to the ill‐posed nature of the inverse problem, inaccuracies in the forward solution, and incorrect assumptions caused by the inverse localization algorithm used (Brookes, Woolrich, & Price, 2014 ), which may lead to spurious connectivity in the source space. Sophisticated methods for estimating reliable functional connectivity with a corrected bias in the source space have been developed by utilizing amplitude, phase, or other variables (Bastos & Schoffelen, 2015 ; Colclough et al, 2015 ; Hipp et al, 2012 ; Kim & Davis, 2021 ; Marzetti et al, 2013 ; Nolte et al, 2004 ; Palva et al, 2018 ; Sanchez‐Bornot et al, 2021 ; Stam, Nolte, & Daffertshofer, 2007 ; Vinck et al, 2011 ; Wang et al, 2018 ; Wens et al, 2015 ) as well as effective connectivity (Baccala & Sameshima, 2001 ; He et al, 2014 ; Kaminski & Blinowska, 1991 ; Lobier et al, 2014 ; Nolte, Ziehe, Kramer, et al, 2008 ; Nolte, Ziehe, Nikulin, et al, 2008 ; Schreiber, 2000 ).…”

“…Therefore, we employed a strategy using arbitrary spacing by a relatively large number of dipoles (approximately 1000 vertices within the brain) to compute connectivity and graph metrics. Another aim was to examine graph metrics derived from phase‐ and amplitude‐based connectivity, both of which are useful for suppressing spurious connectivity caused by source leakage (Brookes et al, 2012 ; Brookes, Woolrich, & Price, 2014 ; Hipp et al, 2012 ; O'Neill et al, 2015 ; Vinck et al, 2011 ). These two metrics partly reflect distinct neuronal mechanisms (Siems & Siegel, 2020 ), implying that they provide complementary information on network features.…”