Multi-session statistics on beamformed MEG data

Abstract

Beamforming has been widely adopted as a source reconstruction technique in the analysis of magnetoencephalography data. Most beamforming implementations incorporate a spatially-varying rescaling (which we term weights normalisation) to correct for the inherent depth bias in raw beamformer estimates. Here, we demonstrate that such rescaling can cause critical problems whenever analyses are performed over multiple sessions of separately beamformed data, for example when comparing effect sizes between different populations. Importantly, we show that the weights-normalised beamformer estimates of neural activity can even lead to a reversal in the inferred sign of the effect being measured. We instead recommend that no weights normalisation be carried out; any depth bias is instead accounted for in the calculation of multi-session (e.g. group) statistics. We demonstrate the severity of the weights normalisation confound with a 2-D simulation, and in real MEG data by performing a group statistical analysis to detect differences in alpha power in eyes-closed rest compared with continuous visual stimulation.

Modulation of hippocampal theta and hippocampal‐prefrontal cortex function by a schizophrenia risk gene

“…This has consistently been shown to be a robust way of detecting stationary functional connectivity [Brookes et al, 2011a[Brookes et al, , 2011bLuckhoo et al, 2012]. Both beamformer-weights-normalized and nonbeamformer-weights-normalized envelopes were estimated (these will be used later in the group-level analysis and see [Luckhoo et al, 2014] for a description of the beamformer weights).…”

“…For each subject, we performed a temporal regression of the component time course segment from the nonweights-normalized downsampled envelopes. This gave a spatial map for each RSN that is specific to each subject but critically has an unbiased estimate of the true variance of activity for that RSN, which is essential for all subsequent multisubject statistics [Luckhoo et al, 2014].…”

Electrophysiological measures of resting state functional connectivity and their relationship with working memory capacity in childhood

“…The envelope time series for every voxel was then effectively low‐pass filtered by dividing each envelope time course into 1s windows and averaging within those windows (Brookes et al ., , who also used the same frequencies of interest as we used here). Both beamformer‐weights‐normalized and non‐beamformer‐weights‐normalized envelopes were estimated for use in the subsequent group‐level (general linear model) analysis (Luckhoo, Brookes & Woolrich, ). Spatial smoothing was also applied to the down‐sampled envelope estimates (FWHM 5 mm).…”

GABAA receptor mapping in human using non-invasive electrophysiology

“…For each of the beamformer voxels on the 6x6x6mm reconstruction grid (5061 voxels), we estimated source activity for each voxel, frequency, person and session. Estimating raw power estimates from MEG/EEG data, in source-space, and comparing these across people and sessions is problematic because geometrical effects, which vary from session to session and across the brain, can lead to artefactual inter-session differences (Luckhoo et al, 2014). This is particularly important for beamformer reconstructions in which typical corrections for depth-biases in single-state filter weights can exacerbate these artefactual differences.…”