Benchmarking methods for mapping functional connectivity in the brain
Zhen-Qi Liu,Andrea I. Luppi,Justine Y. Hansen,Ye Ella Tian,Andrew Zalesky,B.T. Thomas Yeo,Ben D. Fulcher,Bratislav Misic
DOI: https://doi.org/10.1101/2024.05.07.593018
2024-05-08
Abstract:The networked architecture of the brain promotes synchrony among neuronal populations and the emergence of coherent dynamics. These communication patterns can be comprehensively mapped using noninvasive functional imaging, resulting in functional connectivity (FC) networks. Despite its popularity, FC is a statistical construct and its operational definition is arbitrary. While most studies use zero-lag Pearson’s correlations by default, there exist hundreds of pairwise interaction statistics in the broader scientific literature that can be used to estimate FC. How the organization of the FC matrix varies with the choice of pairwise statistic is a fundamental methodological question that affects all studies in this rapidly growing field. Here we comprehensively benchmark the topological and geometric organization, neurobiological associations, and cognitive-behavioral relevance of FC matrices computed using a large library of 239 pairwise interaction statistics. We comprehensively investigate how canonical features of FC networks vary with the choice of pairwise statistic, including (1) hub mapping, (2) weight-distance trade-offs, (3) structure–function coupling, (4) correspondence with other neurophysiological networks, (5) individual fingerprinting, and (6) brain–behavior prediction. We find substantial quantitative and qualitative variation across FC methods. Throughout, we observe that measures such as covariance (full correlation), precision (partial correlation) and distance display multiple desirable properties, including close correspondence with structural connectivity, the capacity to differentiate individuals and to predict individual differences in behavior. Using information flow decomposition, we find that differences among FC methods may arise from differential sensitivity to the underlying mechanisms of inter-regional communication, with some more sensitive to redundant and some to synergistic information flow. In summary, our report highlights the importance of tailoring a pairwise statistic to a specific neurophysiological mechanism and research question, providing a blueprint for future studies to optimize their choice of FC method.
Neuroscience