Subgraph entropy based network approaches for classifying bipolar disorder from resting-state magnetoencephalography.

Area of Science:

• Neuroscience
• Network Science
• Biomarker Discovery

Background:

• Bipolar disorder diagnosis relies on clinical interviews, lacking objective biomarkers.
• Magnetoencephalography (MEG) measures brain activity, showing potential for identifying neurological disorders.
• Current network science methods struggle to capture the full complexity of brain networks in bipolar disorder.

Purpose of the Study:

• To introduce subgraph entropy as a novel metric for analyzing brain network complexity in bipolar disorder.
• To identify reliable biomarkers for bipolar disorder using magnetoencephalography and network science.
• To improve the accuracy and objectivity of bipolar disorder diagnosis.

Main Methods:

• Utilized magnetoencephalography (MEG) signals to measure brain activity in individuals with and without bipolar disorder.
• Applied subgraph entropy, an information-theoretic metric, to characterize the complexity of resting-state brain networks.
• Examined node entropy and edge entropy, specific forms of subgraph entropy, across multiple frequency bands.

Main Results:

• Subgraph entropy features significantly improved the classification of bipolar disorder, especially in the beta frequency band.
• Edge entropy in the beta frequency band achieved high accuracy (0.8462), specificity (0.7308), and sensitivity (0.9231) in distinguishing patients from controls.
• Identified specific brain regions and functional connectivity patterns associated with bipolar disorder.

Conclusions:

• Subgraph entropy is a powerful tool for characterizing brain network complexity and identifying biomarkers for bipolar disorder.
• MEG combined with subgraph entropy offers a more objective and accurate approach to diagnosing bipolar disorder.
• This novel method facilitates a deeper understanding of the pathological mechanisms underlying bipolar disorder.