Hierarchical Spectral Consensus Clustering for Group Analysis of Functional Brain Networks

Area of Science:

• Cognitive Neuroscience
• Neuroimaging Analysis
• Network Science

Background:

• Understanding cognitive functions requires examining brain region integration.
• Graph theory is increasingly used to map brain functional connectivity.
• Identifying community structure in functional networks is crucial but challenging with multi-subject data.

Purpose of the Study:

• To develop a method for identifying representative community structures in multi-subject functional brain networks.
• To address limitations of current methods like data averaging and voting for cross-subject analysis.
• To introduce novel information-theoretic criteria for optimal community structure selection.

Main Methods:

• A hierarchical consensus spectral clustering approach is proposed.
• Information-theoretic criteria are used for selecting the optimal community structure.
• The framework is applied to electroencephalogram (EEG) data.

Main Results:

• The method successfully identifies community structures in functional brain networks.
• It accounts for inter-subject variability in functional connectivity.
• Applied to error-related negativity data, it enhances understanding of cognitive control networks.

Conclusions:

• The proposed hierarchical consensus spectral clustering offers a robust solution for analyzing multi-subject brain network data.
• This approach improves the identification of functional communities relevant to cognitive processes.
• It provides a valuable tool for cognitive neuroscience research, particularly in studying cognitive control.