SHIELD-Dataset-for-EEG-based-MDD-detection

Description

This dataset contains anonymised EEG-derived features extracted from resting-state EEG recordings of 91 participants (50 MDD, 41 HC) recruited from the Department of Psychiatry, PGIMER Chandigarh, India (IEC Approval No. PGI/IEC/2023/EIC001111). EEG signals were recorded using a standard 19-channel international 10-20 electrode placement system under eyes-closed resting-state conditions for healthy and major depressive disorder cohorts. Following ICA-based artifact removal, a multi-domain feature extraction pipeline was applied yielding 314 features per subject — comprising spectral power, wavelet energy, Hjorth parameters, Higuchi fractal dimension, spectral entropy, frontal alpha asymmetry, and theta-to-alpha ratio across all 19 electrodes. The dataset is provided as an anonymised numerical feature matrix (91 × 314) with binary class labels (MDD=1, HC=0). No raw EEG signals, patient identifiers, or demographic details are included. This dataset supports the SHIELD framework for EEG-based MDD detection. Complete source code is available at https://github.com/Gaganbedi2009/SHIELD-framework-for-EEG-based-MDD-detection.

Institutions

• Panjab University

  Chandigarh, Chandigarh
• Punjabi University

  Punjab, Patiāla
• Post Graduate Institute of Medical Education and Research

  Chandigarh, Chandigarh

Categories

Licence