Improved functional connectivity network estimation for brain networks using multivariate partial coherence

摘要

Objective. Graphical networks and network metrics are widely used to understand and characterisebrain networks and brain function. These methods can be applied to a range of electrophysiologicaldata including electroencephalography, local field potential and single unit recordings. Functionalnetworks are often constructed using pair-wise correlation between variables. The objective of thisstudy is to demonstrate that functional networks can be more accurately estimated using partialcorrelation than with pair-wise correlation. Approach. We compared network metrics derived fromunconditional and conditional graphical networks, obtained using coherence and multivariatepartial coherence (MVPC), respectively. Graphical networks were constructed using coherence andMVPC estimates, and binary and weighted network metrics derived from these: node degree, pathlength, clustering coefficients and small-world index. Main results. Network metrics were appliedto simulated and experimental single unit spike train data. Simulated data used a 10x10 grid ofsimulated cortical neurons with centre-surround connectivity. Conditional network metrics gave amore accurate representation of the known connectivity: Numbers of excitatory connections hadrange 3–11, unconditional binary node degree had range 6–80, conditional node degree had range2–13. Experimental data used multi-electrode array recording with 19 single-units from left andright hippocampal brain areas in a rat model for epilepsy. Conditional network analysis showedsimilar trends to simulated data, with lower binary node degree and longer binary path lengthscompared to unconditional networks. Significance. We conclude that conditional networks, wherecommon dependencies are removed through partial coherence analysis, give a more accuraterepresentation of the interactions in a graphical network model. These results have importantimplications for graphical network analyses of brain networks and suggest that functionalnetworks should be derived using partial correlation, based on MVPC estimates, as opposed to thecommon approach of pair-wise correlation.