🧠 EEG Signal Enhancement and EMG Noise Reduction

🌟 Introduction

Electroencephalography (EEG) is a widely used method for studying brain activity. However, EMG noise often contaminates EEG signals, making accurate analysis challenging.
This project focuses on:
✔️ Enhancing EEG signal quality
✔️ Reducing EMG noise through advanced signal processing techniques

🎯 The results of this work can be applied in brain-computer interfaces (BCI), neurofeedback, and medical diagnosis systems.

---

✨ Features

• 🛠️ Bandpass Filtering: Focuses on the critical EEG frequency range (0.5–50 Hz).
• 🌊 Wavelet Transform: Reduces EMG artifacts in both time and frequency domains.
• 🧹 Savitzky-Golay Smoothing: Smooths the waveform for enhanced clarity.
• 📊 Quantitative Evaluation: Improves Signal-to-Noise Ratio (SNR).

---

📂 Dataset

The dataset used is the EEG Motor Movement/Imagery Dataset from PhysioNet, including:

• 🧑‍🤝‍🧑 109 participants
• 📁 Over 1500 EEG recordings

📌 Dataset Source

---

🛠️ Methodology

The signal enhancement pipeline consists of the following steps:

1️⃣ Bandpass Filtering

A Butterworth filter isolates the EEG frequency range, removing unwanted noise while preserving essential signal components.

---

2️⃣ Wavelet Transform

Wavelet analysis decomposes the EEG signal, applying thresholding to suppress EMG artifacts and reconstruct the denoised signal.

---

3️⃣ Savitzky-Golay Smoothing

This technique applies polynomial fitting to smooth minor fluctuations while maintaining the signal's core features.

---

🚀 Results

The pipeline significantly improves signal quality:

📈 Before vs After Noise Reduction

• Raw Signal: Heavily contaminated by EMG noise.
• Filtered Signal: Unwanted noise removed.
• Enhanced Signal: Cleaned and smoothed for analysis.
  
  

🔢 Key Metrics

• Initial SNR: 4.19 dB
• Enhanced SNR: 2.38 dB

🔮 Challenges and Limitations

Challenges

• Signal Variability: EEG signals differ significantly between participants.
• Threshold Tuning: Noise reduction thresholds required iterative optimization.

Limitations

• Offline processing only; no real-time capabilities.
• Limited dataset scope.

---

📅 Future Work

• 🕒 Real-Time Processing: Implement the pipeline for real-time EEG analysis.
• 📈 Expand Dataset: Test with more diverse populations for broader applicability.
• 🤖 AI-Based Noise Reduction: Use machine learning for adaptive thresholding.

---

🛠️ Getting Started

Prerequisites

📌 Install Python (3.8 or above).
📌 Required libraries: NumPy, SciPy, PyWavelets, Pyedflib.

---

Installation

1️⃣ Clone the repository:

git clone https://github.com/habibkhan099/EEG-Signal-Enhancement-and-EMG-Noise-Reduction.git
cd EEG-Signal-Enhancement-and-EMG-Noise-Reduction

2️⃣ Install dependencies:

pip install -r requirements.txt

---

Usage

1️⃣ Prepare EEG data in .csv or .mat format.
2️⃣ Run the main script for signal processing:

python main.py --input your_data.csv --output enhanced_signal.csv

3️⃣ Visualize results:

main.py

---

📷 Visuals

Here are some visual aids for the project:

1️⃣ Block Diagram

2️⃣ GUI

---

📚 References

1️⃣ Goldberger AL, et al. PhysioNet: EEG Motor Imagery Dataset. PhysioNet.
2️⃣ Mallat, S. A Wavelet Tour of Signal Processing. Academic Press, 1999.
3️⃣ Tawhid, M.N.A. et al., Time-Frequency EEG Analysis Using Wavelet Transform.

---

🤝 Contributing

We welcome contributions! To contribute:
1️⃣ Fork the repository.
2️⃣ Create a new branch (feature-name).
3️⃣ Submit a pull request.

---

📄 License

This project is licensed under the MIT License. See the LICENSE file for more details.

---

About

Developed as a Semester Project of Digital Signal Processing during the B.Sc. Computer Engineering program at UET Taxila, supervised by Dr. Muhammad Majid.

---

📌 Repository

🔗 Visit Repository

---