Multiple subject analysis of functional brain network communities through co-regularized spectral clustering

Area of Science:

• Neuroscience
• Network Science
• Computational Neuroscience

Background:

• The human brain functions as a complex network with segregated modules and short path lengths.
• Understanding these modules is crucial for comprehending brain network organization.
• Existing methods like data averaging or consensus clustering have limitations in handling multi-subject neurophysiological data and subject variability.

Purpose of the Study:

• To adapt a co-regularized multiview spectral clustering approach for analyzing functional brain networks.
• To address challenges in multi-subject data analysis, including subject variability and outliers.
• To investigate the functional networks involved in cognitive control by analyzing electroencephalography (EEG) data during error-related negativity (ERN) responses.

Main Methods:

• Adaptation of a co-regularized multiview spectral clustering framework.
• Application to electroencephalography (EEG) data from a study on error-related negativity (ERN).
• Comparison of functional network structures between error and correct response conditions.

Main Results:

• The proposed method effectively identifies underlying modules in functional brain networks.
• The framework accommodates multi-subject data while accounting for individual variability.
• Distinct network structures were observed between error and correct response conditions, shedding light on cognitive control processes.

Conclusions:

• The co-regularized multiview spectral clustering approach offers a robust solution for analyzing complex brain networks from multi-subject EEG data.
• This method enhances the understanding of functional brain organization and cognitive control mechanisms.
• The findings highlight the utility of advanced clustering techniques in neurophysiological research.