Research Article: High-resolution directed human connectomes and the Consensus Connectome Dynamics

Date Published: April 16, 2019

Publisher: Public Library of Science

Author(s): Balázs Szalkai, Csaba Kerepesi, Bálint Varga, Vince Grolmusz, Yuanquan Wang.


Here we show a method of directing the edges of the connectomes, prepared from HARDI datasets from the human brain. Before the present work, no high-definition directed braingraphs were published, because the tractography methods in use are not capable of assigning directions to the neural tracts discovered. Previous work on the functional connectomes applied low-resolution functional MRI-detected statistical causality for the assignment of directions of connectomes of typically several dozens of vertices. Our method is based on the phenomenon of the “Consensus Connectome Dynamics”, described earlier by our research group. In this contribution, we apply the method to the 423 braingraphs, each with 1015 vertices, computed from the public release of the Human Connectome Project, and we also made the directed connectomes publicly available at the site We also show the robustness of our edge directing method in four independently chosen connectome datasets: we have found that 86% of the edges, which were present in all four datasets, get the same directions in all datasets; therefore the direction method is robust. While our new edge-directing method still needs more empirical validation, we think that our present contribution opens up new possibilities in the analysis of the high-definition human connectome.

Partial Text

High-angular resolution diffusion imaging (HARDI) datasets are widely used today for the construction of connectomes or braingraphs. The nodes (or vertices) of these graphs correspond to anatomically identified [1, 2], small (1-1.5 cm2) areas of the gray matter, and two nodes are connected by an undirected edge if neural fiber tracts are discovered, connecting the gray matter areas, corresponding to the two nodes [3, 4].

In the present work, we are assigning directions to some of the edges of braingraphs, depending on the appearance of that edge in the CCD.

The braingraphs in this study were constructed from the data of the Human Connectome Project [5], using the workflow described in [36]. The undirected braingraphs have 1015 vertices each; the whole set of 423 braingraphs can be downloaded from the site




Leave a Reply

Your email address will not be published.