Benchmarking methods for mapping functional connectivity in the brain
Poster No:
1386
Submission Type:
Abstract Submission
Authors:
Zhen-Qi Liu1, Andrea Luppi1, Justine Hansen1, Ye Tian2, Andrew Zalesky2, Thomas Yeo3, Ben Fulcher4, Bratislav Misic1
Institutions:
1Montreal Neurological Institute, Montreal, Canada, 2Melbourne Neuropsychiatric Centre, The University of Melbourne, Melbourne, Australia, 3Yong Loo Lin School of Medicine, National University of Singapore, Singapore, Singapore, 4School of Physics, The University of Sydney, Sydney, Australia
First Author:
Co-Author(s):
Introduction:
The orchestrated dynamics of the brain at the macroscale are often mapped with noninvasive functional imaging and studied as 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, functional dynamics in the brain extends far beyond simple linear effects. There exist hundreds of pairwise interaction statistics in the broader scientific literature that can be used to measure FC. How to estimate and compare FC networks is a fundamental question that affects all studies in this field. Here we systematically benchmark how canonical features of FC networks vary across a large library of 239 pairwise interaction statistics. We comprehensively investigate the topological and geometric organization, neurobiological associations, and cognitive-behavioral relevance of FC matrices computed, including their (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.
Methods:
Pairwise statistics were derived for N = 326 unrelated healthy young adults using functional time series from the Human Connectome Project (HCP). We used the pyspi package (Cliff et al., 2023) to estimate 239 pairwise statistics from 49 pairwise interaction measures in 6 families of statistics, yielding 239 FC matrices for each participant. Structure-function relationship was evaluated using structural connectivity estimated from diffusion tractography and network communication measures (Liu et al., 2023). Multimodal neurophysiological networks were used to contextualize the FC networks (Hansen et al., 2023). Individual differences were quantified by (1) fingerprinting using identifiability index, and (2) brain-behavior predictions using kernel ridge regression. Information flow patterns were estimated with the Integrated Information Decomposition (ΦID) framework (Luppi et al., 2024), and relative contributions of information-dynamic atoms were derived from dominance analysis. To evaluate generalizability across different acquisition and preprocessing pipelines, we also replicated the analyses in 6 additional fMRI datasets with >1000 subjects in total.
Results:
We found substantial quantitative and qualitative variation across FC methods, even for well-studies phenomena. The location of hubs were localized in unimodal cortex for some methods, but are more widespread across the unimodal–transmodal axis for others. The weight-distance decay and structure-function coupling were captured by most methods, but at considerable variation. Different FC methods often align with different forms of inter-regional biological similarity, and have varying capabilities for capturing individual differences. Using information flow decomposition, we found that the FC methods are differentially sensitive to the redundant and synergistic information flow, suggesting a promising mechanistic explanation. 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.
Conclusions:
In summary, our study comprehensively benchmarks the network architecture of resting state BOLD FC using a large library of pairwise statistics. We observe substantial variation across FC methods and across a wide array of analyses, reflecting differential sensitivity to biological features and to types of information flow. Our results highlight the importance of tailoring a pairwise statistic to a specific neurophysiological mechanism and research question, setting the foundation for future studies to optimize their choice of FC method.
Modeling and Analysis Methods:
Connectivity (eg. functional, effective, structural) 2
fMRI Connectivity and Network Modeling 1
Methods Development
Keywords:
Other - functional connectivity; network neuroscience; structure-function relationship
1|2Indicates the priority used for review
·Figure 1
·Figure 2
Abstract Information
By submitting your proposal, you grant permission for the Organization for Human Brain Mapping (OHBM) to distribute your work in any format, including video, audio print and electronic text through OHBM OnDemand, social media channels, the OHBM website, or other electronic publications and media.
The Open Science Special Interest Group (OSSIG) is introducing a reproducibility challenge for OHBM 2025. This new initiative aims to enhance the reproducibility of scientific results and foster collaborations between labs. Teams will consist of a “source” party and a “reproducing” party, and will be evaluated on the success of their replication, the openness of the source work, and additional deliverables. Click here for more information. Propose your OHBM abstract(s) as source work for future OHBM meetings by selecting one of the following options:
Please indicate below if your study was a "resting state" or "task-activation” study.
Healthy subjects only or patients (note that patient studies may also involve healthy subjects):
Was this research conducted in the United States?
Were any human subjects research approved by the relevant Institutional Review Board or ethics panel? NOTE: Any human subjects studies without IRB approval will be automatically rejected.
Were any animal research approved by the relevant IACUC or other animal research panel? NOTE: Any animal studies without IACUC approval will be automatically rejected.
Please indicate which methods were used in your research:
Provide references using APA citation style.
Hansen, J. Y., Shafiei, G., Voigt, K., Liang, E. X., Cox, S. M., Leyton, M., ... & Misic, B. (2023). Integrating multimodal and multiscale connectivity blueprints of the human cerebral cortex in health and disease. PLoS biology, 21(9), e3002314.
Liu, Z. Q., Shafiei, G., Baillet, S., & Misic, B. (2023). Spatially heterogeneous structure-function coupling in haemodynamic and electromagnetic brain networks. NeuroImage, 278, 120276.
Luppi, A. I., Rosas, F. E., Mediano, P. A., Menon, D. K., & Stamatakis, E. A. (2024). Information decomposition and the informational architecture of the brain. Trends in Cognitive Sciences.