EEG-Signal-Processing-MNE

An automated end-to-end EEG processing pipeline built with MNE-Python. Features data loading/concatenation, 10-20 montage standardization, bandpass filtering, FastICA physiological artifact subtraction, time-locked epoching, and frequency-domain feature extraction (PSD).

🚀 Key Features

• Data Ingestion & Concatenation: Seamless loading and merging of multiple contiguous European Data Format (.edf) files.
• Nomenclature Standardisation: Robust string-cleansing logic mapping physical channel configurations (e.g., trailing punctuation) dynamically to standard International 10-20 system templates.
• Ocular Artifact Subtraction: Implements Independent Component Analysis (FastICA) to isolate, visualize, and exclude eye blinks and motor movements.
• Time-Locked Epoching: Segments continuous signals into time-locked epochs centered around stimulus annotations with custom baseline correction.
• Spectral Feature Extraction: Computes Power Spectral Density (PSD) across standard frequency bands (Delta, Theta, Alpha, Beta).

🛠️ Pipeline Architecture

The pipeline enforces a logical, strict neuroscientific data workflow to preserve signal integrity:

1. Data Loading & Verification ➡️
2. Channel Renaming & Montage Alignment ➡️
3. Bandpass Filtering (1.0–40.0 Hz) ➡️
4. Epoching & Segmentation ➡️
5. ICA Decomposition & Artifact Removal ➡️
6. Feature Extraction (Power Spectral Density)

📊 Dataset

The data used in this project comes from the [PhysioNet EEG Motor Movement/Imagery Dataset]. To run this notebook locally:

1. Download files S001R07.edf and S001R11.edf.
2. Place them in your local working directory before executing the pipeline.

📦 Prerequisites

Ensure you have Python installed alongside the dependencies listed in requirements.txt:

pip install -r requirements.txt