Assessing the Repeatability of Multi-Frequency Multi-Layer Brain Network Topologies Across Alternative Researcher's...

Area of Science:

• Neuroscience
• Network Science
• Data Science

Background:

• Multilayer functional brain networks integrate information across frequencies, offering advantages over single-frequency analyses.
• Resting-state fMRI (rs-fMRI) is a key tool for studying brain connectivity.
• Reproducibility in multilayer network construction is crucial for reliable findings.

Purpose of the Study:

• To assess the reproducibility of multi-frequency multilayer functional connectivity topologies derived from rs-fMRI data.
• To investigate the impact of various preprocessing choices on brain network topology.
• To recommend best practices for generating consistent and reproducible brain network analyses.

Main Methods:

• Analysis of rs-fMRI datasets from a single subject (longitudinal) and multiple subjects (cross-sectional).
• Exploration of different frequency extraction filtering methods, connectivity estimators, topological filtering schemes, and spatial scales.
• Systematic evaluation of preprocessing parameter combinations for network reproducibility.

Main Results:

• Identified specific combinations of preprocessing steps that yield consistently reproducible multi-frequency multilayer brain network topologies.
• Demonstrated that choices in filtering, connectivity estimation, and spatial scale significantly influence network reproducibility.
• Provided empirical evidence on the robustness of multilayer network analysis under varying methodological choices.

Conclusions:

• Specific preprocessing pipelines can ensure reproducible multilayer functional brain network topologies from rs-fMRI.
• Understanding the impact of methodological choices is essential for advancing multilayer network analysis in neuroscience.
• This work offers guidance for researchers aiming for reliable and reproducible brain network studies.