EEG-Based Major Depressive Disorder (MDD) Detection using Federated Learning

A privacy-preserving web-based federated learning system for detecting Major Depressive Disorder (MDD) using EEG signals. This framework enables multiple healthcare institutions to collaboratively train machine learning models without sharing sensitive patient data.

📋 Table of Contents

• Overview
• Key Features
• Dataset
• Contact

🌟 Overview

This repository implements a federated learning framework for EEG-based detection of Major Depressive Disorder (MDD). The system addresses critical challenges in healthcare AI:

• Data Privacy: Raw EEG data never leaves the client institutions
• Collaborative Learning: Multiple hospitals can improve models together
• Clinical Interpretability: Spectral entropy analysis provides biomarkers
• Real-World Deployment: Web-based interface for practical use

Novel Contributions:

• First federated learning implementation for EEG-based MDD detection
• Novel loss-weighted aggregation for non-IID data distributions
• Four FL aggregation strategies for comprehensive comparison
• Web-based Django platform with client-server architecture
• Integration of spectral entropy for clinical decision support

🚀 Key Features

Privacy Protection

• Raw EEG data remains on client devices
• Only model updates (weights) are shared
• Encrypted client-server communication
• No central data storage

Multiple Aggregation Strategies

1. Proposed Method: Loss-weighted aggregation
2. FedAvg: Sample-weighted standard federated averaging
3. Simple Averaging: Equal weights for all clients
4. FedNova: Normalized by local computation steps

EEG Processing Pipeline

• Gamma band (30-100 Hz) extraction
• 15-second epoch segmentation
• Spectral entropy calculation
• Z-score normalization
• 20-channel selection (10-20 system)

Technical Stack

• Backend: Django 4.x, TensorFlow 2.x, MNE-Python
• Frontend: HTML5, CSS3, JavaScript
• ML Framework: Keras with 1D CNN architecture
• Data Processing: NumPy, SciPy, scikit-learn

Dataset

Public EEG Dataset This project uses the publicly available EEG dataset from:

Mumtaz, W., Xia, L., Ali, S. S. A., Yasin, M. A. M., Husaini, M., & Malik, A. S. (2017). Electroencephalogram (EEG)-based computer-aided technique to diagnose major depressive disorder (MDD). Biomedical Signal Processing and Control, 31, 108-115.

Dataset Specifications:

Subjects: 30 Healthy Controls, 34 MDD Patients

Age: HC = 40.33 ± 12.86, MDD = 38.23 ± 15.64

Recording: 5-minute EEG, 256 Hz sampling rate

Electrodes: 19 channels (10-20 system)

Conditions: Eyes closed, eyes open, P300 task

Format: EDF files