Schizophrenia EEG Classification Using EEG Brain Signals

A novel deep learning approach to classify Schizophrenia using EEG (electroencephalography) brain signals with LSTM and Dense neural networks.

📋 Status

Finished - Schizophrenia EEG Classification Project

🎯 Project Overview

This project implements a sequential neural network model using TimeDistributed LSTM layers combined with Dense layers to classify EEG signals into two categories:

• Class 0: Control (Non-Schizophrenia)
• Class 1: Schizophrenia

The model uses temporal feature extraction from EEG data for binary classification with high accuracy.

📁 Project Structure

Schizophrenia_EEG/
├── notebooks/
│   └── schizophrenia-v2.ipynb                  # Main project notebook
├── modules/
│   ├── augmentation.py                         # ✅ Data augmentation module
│   ├── data_loading.py                         # ✅ Data loading utilities
│   ├── evaluate.py                             # ✅ Model evaluation functions
│   ├── feature_extraction.py                   # ✅ Feature extraction module
│   └── training.py                             # ✅ Model training module
├── references/
│   ├── Input Data Representation.ipynb         # Reference notebook
│   ├── Training.ipynb                          # Reference notebook
│   ├── eegemotion/                             # Reference materials
│   └── graphics/
│       └── brainmaps.png                       # EEG channel layout reference
├── sample_images/
│   ├── de/                                     # Differential Entropy samples
│   │   ├── band_0/patient_0.png
│   │   ├── band_1/patient_0.png
│   │   ├── band_2/patient_0.png
│   │   └── band_3/patient_0.png
│   ├── psd/                                    # Power Spectral Density samples
│   │   ├── band_0/patient_0.png
│   │   ├── band_1/patient_0.png
│   │   ├── band_2/patient_0.png
│   │   └── band_3/patient_0.png
│   └── se/                                     # Spectral Entropy samples
│       ├── band_0/patient_0.png
│       ├── band_1/patient_0.png
│       ├── band_2/patient_0.png
│       └── band_3/patient_0.png
├── results/
│   └── confusion_matrix_schizophrenia.png      # Model performance results
├── graphics/
│   ├── brainmaps.png                           # EEG electrode positions
│   └── model_architecture_timedistributed.png  # Model architecture diagram
├── README.md
├── LICENSE
├── requirements.txt
└── setup.py

🧠 EEG Channel Reference

Reference visualization of EEG electrode positions and channel arrangement

🏗️ Model Architecture

Deep learning model using TimeDistributed LSTM layers with temporal feature extraction

📊 Sample Visualizations

Differential Entropy (DE) Features

These heatmaps show the Differential Entropy values across EEG channels for each frequency band:

Band 0 (Delta: 1-4 Hz)

Band 1 (Theta: 4-8 Hz)

Band 2 (Alpha: 8-13 Hz)

Band 3 (Beta: 13-30 Hz)

Power Spectral Density (PSD) Features

These heatmaps represent the power spectrum across different frequency bands:

Band 0 (Delta: 1-4 Hz)

Band 1 (Theta: 4-8 Hz)

Band 2 (Alpha: 8-13 Hz)

Band 3 (Beta: 13-30 Hz)

🎯 Model Results

Confusion Matrix

Classification performance metrics showing true positives, true negatives, false positives, and false negatives

🚀 Getting Started

Prerequisites

• Python 3.7+
• TensorFlow/Keras
• NumPy
• Pandas
• Scikit-learn
• Matplotlib
• Scipy

Installation

# Clone the repository
git clone https://github.com/ans036/Schizophrenia_EEG.git
cd Schizophrenia_EEG

# Install dependencies
pip install -r requirements.txt

Usage

1. Data Preparation:

   from modules.data_loading import load_eeg_data
   data = load_eeg_data('path_to_eeg_files')

2. Feature Extraction:

   from modules.feature_extraction import extract_features
   features = extract_features(data)

3. Data Augmentation (optional):

   from modules.augmentation import augment_data
   augmented_features = augment_data(features)

4. Model Training:

   from modules.training import train_model
   model = train_model(features, labels)

5. Model Evaluation:

   from modules.evaluate import evaluate_model
   results = evaluate_model(model, test_data, test_labels)

📚 Key Features

✅ TimeDistributed LSTM: Processes temporal patterns in EEG data
✅ Multiple Feature Extraction: Differential Entropy, Power Spectral Density, Spectral Entropy
✅ Data Augmentation: Increases training dataset diversity
✅ Modular Architecture: Easy to understand and extend
✅ Comprehensive Evaluation: Detailed performance metrics and visualizations

📖 References

• EEG Emotion Recognition Dataset: eegemotion/
• Input Data Representation: Input Data Representation.ipynb
• Training Details: Training.ipynb

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

👤 Author

Anish - GitHub Profile

🤝 Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

📧 Contact

For questions or suggestions, please email ani.tubai022@gmail.com or open an issue on the GitHub repository.

---

Last Updated: 2026
Status: Development Finished