Functional alignments in brain connectivity networks

Alzheimer’s disease (AD) is a brain disconnection syndrome, where functional connectivity analysis can detect changes in neural activity in pre-dementia stages [8]. Functional connectivity networks from functional magnetic resonance imaging (fMRI) and electroencephalogram (EEG) are susceptible to signal noise from biologic artefacts (e.g. cardiac artefacts) and environmental sources (e.g. electrical interference). A particular challenge for EEG is volume conduction, whereby a signal from a single source propagates through biological tissue to be detected simultaneously by multiple sensors (channels). The imaginary part of coherency (iCOH) provides a measure for connectivity that avoids this signal contamination, by ignoring correlation between signals with zero or π-phase lag. This removes false instantaneous activity with connectivity denoting synchronised signals at a given time lag, but it does come at the cost of erasing true instantaneous activity. We propose eigenvector alignment (EA) as a method for evaluating pairwise relationships from network eigenvectors; revealing noise robust, structural, insights from functional connectivity networks.