Pairwise maximum entropy model explains the role of white matter structure in shaping emergent co-activation states

Arian Ashourvan,Shi Gu,Adam Pines,Christopher W Lynn,Kathryn A Davis,Preya Shah,Brian Litt,Danielle S Bassett

doi:10.1038/s42003-021-01700-6

Abstract

A major challenge in neuroscience is determining a quantitative relationship between the brain’s white matter structural connectivity and emergent activity. We seek to uncover the intrinsic relationship among brain regions fundamental to their functional activity by constructing a pairwise maximum entropy model (MEM) of the inter-ictal activation patterns of five patients with medically refractory epilepsy over an average of ~14 hours of band-passed intracranial EEG (iEEG) recordings per patient. We find that the pairwise MEM accurately predicts iEEG electrodes’ activation patterns’ probability and their pairwise correlations. We demonstrate that the estimated pairwise MEM’s interaction weights predict structural connectivity and its strength over several frequencies significantly beyond what is expected based solely on sampled regions’ distance in most patients. Together, the pairwise MEM offers a framework for explaining iEEG functional connectivity and provides insight into how the brain’s structural connectome gives rise to large-scale activation patterns by promoting co-activation between connected structures.

Highlights

A major challenge in neuroscience is determining a quantitative relationship between the brain’s white matter structural connectivity and emergent activity
We compared the receiver operator characteristic (ROC) results for the detection of empirical and null structural connectivity matrices using different functional methods and recording montages in Supplementary Fig. 11. These results show that pairwise maximum entropy model (MEM) and partial correlation of intracranial EEG (iEEG) power provide the most accurate and significant identification of anatomical connectivity from functional estimates in 3 out of 5 patients across several frequency bands
We provide the aforementioned results from a representative subject (#5). These results suggest that, the presence of structural connectivity is highly confounded with inter-regional distance, the pairwise MEM reveals high structure-function coupling across a broad range of frequencies, beyond those anticipated by inter-regional distance alone

Summary

Introduction

A major challenge in neuroscience is determining a quantitative relationship between the brain’s white matter structural connectivity and emergent activity. To distinguish between direct and indirect communication, one must begin with a model of how patterns of activity are generated in the brain and infer the network of underlying interactions that best describes correlations in the data Biophysical microcircuit modeling such as dynamical causal modeling[23,24] aims to address these limitations using neural mass models of synaptic dynamics, informed by empirical ion channel and structural priors. Fitting a pairwise MEM entails iteratively adjusting the strength of individual region activation and all region pair interactions until the estimated correlations match the correlations observed in the data In this way, the MEM makes quantitative predictions about the frequencies of global activity patterns, rather than quantifying the similarities between regions, as is common in studies of functional connectivity. Prior studies using resting-state fMRI data have demonstrated that the pairwise MEM accurately predicts the observed patterns of regional activations, and provides a more accurate map of the underlying structural connectivity than conventional functional connectivity methods[29]

Objectives

Methods

Conclusion