Connectivity-based change point detection for large-size functional networks.

Area of Science:

• Neuroscience
• Network Science
• Data Science

Background:

• The brain exhibits dynamic functional network states, not static configurations.
• Characterizing these dynamic states requires methods applicable to large-scale brain networks and long time series.
• Existing methods may not be optimal for large, complex functional brain networks.

Purpose of the Study:

• To develop a fast and efficient method for detecting change points in large-size functional networks from resting-state fMRI data.
• To address the need for network-based change point detection in large networks with extended time series.
• To enhance the analysis of dynamic brain states in resting-state functional magnetic resonance imaging (fMRI).

Main Methods:

• A novel covariance change point detection statistic is proposed.
• The method utilizes a local false discovery rate (fdr) estimation based on the empirical null principle.
• A semiparametric mixture model is employed for statistical testing.
• The approach is validated using simulations and fMRI datasets (up to 300 nodes) from the Human Connectome Project.

Main Results:

• The proposed method demonstrates efficiency in detecting change points within large-scale functional brain networks.
• The technique shows reduced sensitivity to the selection of window size.
• The method successfully identifies changed edges within the networks.
• Effective analysis of both task-based and resting-state fMRI data is shown.

Conclusions:

• The developed covariance-based change point detection method is highly effective for large-size resting-state brain networks.
• This approach facilitates the exploration of dynamic brain states over long durations.
• The method offers a valuable tool for analyzing complex, large-scale neuroimaging data.