Graph theory analysis based on cross frequency coupling methods in major depressive disorder: A resting state EEG...

Area of Science:

• Neuroscience
• Computational Psychiatry
• Biomarkers

Background:

• Major depressive disorder (MDD) is a prevalent and disabling condition impacting personal and social functioning.
• Current diagnostic methods rely on subjective patient reports and clinical expertise, potentially limiting diagnostic precision.
• Objective biomarkers are needed for accurate and reliable MDD diagnosis and severity assessment.

Purpose of the Study:

• To identify electroencephalography (EEG)-based biomarkers for diagnosing depression severity.
• To investigate the utility of cross-frequency coupling (CFC) and graph theory metrics in differentiating depression levels.
• To develop a machine learning model for multi-level depression severity classification using EEG data.

Main Methods:

• Resting-state EEG signals were acquired from 37 participants (15 healthy, 10 moderately depressed, 12 severely depressed).
• EEG data were analyzed using cross-frequency coupling (CFC) measures and graph theory metrics (degree, K-coreness centrality).
• A Support Vector Machine (SVM) classifier was trained using selected CFC features to predict depression severity.

Main Results:

• Depression impacts the entire cerebral cortex, with notable effects in frontal and occipital regions.
• Significant differences in degree and K-coreness centrality were observed across most brain regions.
• The SVM classifier achieved 94.25% accuracy in classifying depression severity using four CFC features (low α and low γ bands).

Conclusions:

• The combination of CFC and graph theory analysis provides effective EEG-based biomarkers for multi-level depression severity classification.
• This novel approach shows promise for objective diagnosis and treatment monitoring of MDD.
• Integrating this method with psychological scales may enhance MDD diagnosis and management.