[Paper Review] A study of resting-state EEG biomarkers for depression recognition

Shuting Sun, Jianxiu Li+4 more|arXiv (Cornell University)|Feb 23, 2020

Functional Brain Connectivity Studies37 references20 citations

TL;DR

This study investigates resting-state EEG biomarkers for major depressive disorder (MDD) recognition using 128-channel EEG data from 24 MDD patients and 29 controls. Functional connectivity features based on Phase Lag Index (PLI) outperformed linear and nonlinear features, achieving 82.31% classification accuracy with ReliefF feature selection and logistic regression, particularly highlighting intrahemispheric PLI connections as key biomarkers for MDD.

ABSTRACT

Background: Depression has become a major health burden worldwide, and effective detection depression is a great public-health challenge. This Electroencephalography (EEG)-based research is to explore the effective biomarkers for depression recognition. Methods: Resting state EEG data was collected from 24 major depressive patients (MDD) and 29 normal controls using 128 channel HydroCel Geodesic Sensor Net (HCGSN). To better identify depression, we extracted different types of EEG features including linear features, nonlinear features and functional connectivity features phase lagging index (PLI) to comprehensively analyze the EEG signals in patients with MDD. And using different feature selection methods and classifiers to evaluate the optimal feature sets. Results: Functional connectivity feature PLI is superior to the linear features and nonlinear features. And when combining all the types of features to classify MDD patients, we can obtain the highest classification accuracy 82.31% using ReliefF feature selection method and logistic regression (LR) classifier. Analyzing the distribution of optimal feature set, it was found that intrahemispheric connection edges of PLI were much more than the interhemispheric connection edges, and the intrahemispheric connection edges had a significant differences between two groups. Conclusion: Functional connectivity feature PLI plays an important role in depression recognition. Especially, intrahemispheric connection edges of PLI might be an effective biomarker to identify depression. And statistic results suggested that MDD patients might exist functional dysfunction in left hemisphere.

Motivation & Objective

• To identify reliable EEG biomarkers for major depressive disorder (MDD) using resting-state EEG signals.
• To compare the discriminative power of linear, nonlinear, and functional connectivity EEG features in distinguishing MDD patients from healthy controls.
• To determine the optimal combination of features and machine learning methods for accurate MDD classification.
• To investigate hemispheric differences in functional connectivity patterns in MDD patients.

Proposed method

• Collected 128-channel resting-state EEG data from 24 MDD patients and 29 healthy controls using a HydroCel Geodesic Sensor Net.
• Extracted three types of EEG features: linear (e.g., power spectral density), nonlinear (e.g., sample entropy), and functional connectivity using Phase Lag Index (PLI).
• Applied multiple feature selection methods, including ReliefF, to identify the most informative features.
• Employed logistic regression (LR) as the primary classifier to evaluate performance across different feature sets.
• Analyzed spatial distribution of optimal features to identify significant connectivity patterns, especially intrahemispheric vs. interhemispheric connections.
• Conducted statistical comparisons between MDD and control groups to assess significance of observed differences in PLI features.

Experimental results

Research questions

• RQ1 Which type of EEG feature—linear, nonlinear, or functional connectivity—best discriminates MDD patients from healthy controls?
• RQ2 Can combining multiple EEG feature types improve classification accuracy for MDD detection?
• RQ3 Are intrahemispheric functional connectivity patterns in PLI more informative than interhemispheric patterns for identifying MDD?
• RQ4 Does the left hemisphere show more pronounced functional connectivity disruptions in MDD patients?

Key findings

• Functional connectivity features based on Phase Lag Index (PLI) significantly outperformed linear and nonlinear EEG features in classifying MDD.
• The highest classification accuracy of 82.31% was achieved using the ReliefF feature selection method combined with logistic regression on the full set of features.
• The optimal feature set contained significantly more intrahemispheric PLI connections than interhemispheric connections, indicating a stronger role for intrahemispheric networks in MDD.
• Intrahemispheric PLI connections showed statistically significant differences between MDD patients and controls, suggesting their potential as a biomarker.
• Statistical analysis indicated a possible functional dysfunction in the left hemisphere among MDD patients.