Exploring Phase Recurrence in Decomposed EEG Signals for Major Depressive Disorder Detection

Major Depressive Disorder (MDD) is a complex mental health condition that impacts millions globally, requiring robust diagnostic solutions. Evidence shows that EEG signals from MDD patients differ significantly in their neural oscillations and connectivity patterns compared to healthy individuals. This study proposes an EEG-based detection method for MDD by integrating Variational Mode Decomposition (VMD), phase analysis, and Recurrence Quantification Analysis (RQA). EEG signals were decomposed into intrinsic mode functions using VMD, and phase information was extracted through the Hilbert Transform. RQA metrics, including Recurrence Rate, Determinism, and Laminarity, were calculated to capture non-linear dynamics in the phase space and further given as input to a machine learning classifier to differentiate MDD patients from healthy controls. The proposed framework achieved a classification accuracy of 97.66% with the AdaBoost classifier, outperforming traditional methods. The region-wise analysis highlighted higher recurrence and reduced variability in frontal and temporal regions in MDD patients, indicating altered connectivity patterns. This integration offers a reliable framework for accurate MDD detection from EEG signals, advancing the development of objective clinical biomarkers.