Network alignment and similarity reveal atlas-based topological differences in structural connectomes.

Matteo Frigo,Rachid Deriche,Emanuele Natale,David Coudert,Emilio Cruciani,Samuel Deslauriers-Gauthier

doi:10.1162/netn_a_00199

Abstract

The interactions between different brain regions can be modeled as a graph, called connectome, whose nodes correspond to parcels from a predefined brain atlas. The edges of the graph encode the strength of the axonal connectivity between regions of the atlas that can be estimated via diffusion magnetic resonance imaging (MRI) tractography. Herein, we aim to provide a novel perspective on the problem of choosing a suitable atlas for structural connectivity studies by assessing how robustly an atlas captures the network topology across different subjects in a homogeneous cohort. We measure this robustness by assessing the alignability of the connectomes, namely the possibility to retrieve graph matchings that provide highly similar graphs. We introduce two novel concepts. First, the graph Jaccard index (GJI), a graph similarity measure based on the well-established Jaccard index between sets; the GJI exhibits natural mathematical properties that are not satisfied by previous approaches. Second, we devise WL-align, a new technique for aligning connectomes obtained by adapting the Weisfeiler-Leman (WL) graph-isomorphism test. We validated the GJI and WL-align on data from the Human Connectome Project database, inferring a strategy for choosing a suitable parcellation for structural connectivity studies. Code and data are publicly available.

Highlights

Due to the immense complexity of the brain, it is impossible to gain any insight into its global operation without simplifying assumptions
We processed the data of 100 unrelated subjects from the Human Connectome Project (HCP) database and obtained the structural brain networks via dMRI-based tractography
The fiber orientation distribution functions (fODFs) were used as input for probabilistic anatomically constrained tractography performed with the iFOD2 algorithm [41] seeding from the gray matter - white matter interface and obtaining a total of five million streamlines per subject

Summary

Introduction

Due to the immense complexity of the brain, it is impossible to gain any insight into its global operation without simplifying assumptions. In addition to studying the characteristics of specific brain regions defined by a parcellation, there has been a growing interest in their relationship and interactions, an emerging field known as connectomics. In this context, the focus is shifted from understanding how information is segregated in the brain to how it is integrated. An underlying assumption is that a correspondence exists between nodes of the network across subjects, a condition which is usually satisfied by using a group parcellation [34, 18] The drawback of this strategy is that it ignores any subject specific changes in cortical organization and reduces the specificity of the results. This approach has never been investigated in the field of network neuroscience

Methods

Results

Discussion

Conclusion