epilepsy_eeg_classification is a python project that works with EEG data to classify epilepsy events.
In this repo you will find resources to:
- Preprocessing of raw EEG data
- Extration time domain and frequency domain features from EEG data
- Implementation of five popular ML classification models selected to tackle nonlinear and imbalanced data (MLP, KNN, Kernel SVM, Random Forest, AdaBoost)
This study is based on course project paper titled '
- Goal 1: Compare the performance of five popular nonlinear ML algorithms on patient-specific seizure classifications
- Goal 2: Compare patient-specific versus non-patient specific classification performance
The complete study can be found in report.pdf
[CHB-MIT Scalp EEG Database] (https://physionet.org/content/chbmit/1.0.0/) [1]
scripts/preprocessing.py Inputs raw EEG files, performs high and low pass bandwidth filters, epoch segmentation, and feature extraction
scripts/consolidation.py Combines multiple preprocessed datasets of same or different subjects into a combined dataset
scripts/eeg_classifcation.py Runs five seizure classification algorithms for a given dataset
scripts/pyeeg.py Useful functions to extra time-domain and frequency-domain features from raw EEG [2]
Raw EEG files from CHB-MIT database were preprocessed by preprocessing.py and saved into the data folder. Then, consolidation.py was used to combine 35-45 hour long files into the following datasets used for model training. These were not uploaded due to Github file size limits.
data/chb01.csv 145610 epochs, 505 with seizure
data/chb02.csv 125685 epochs, 207 with seizure
data/chb03.csv 136464 epochs, 465 with seizure
data/five_Subjects.csv 125685 epochs, 430 with seizure
[1] Ali Shoeb. Application of Machine Learning to Epileptic Seizure Onset Detection and Treatment. PhD Thesis, Massachusetts Institute of Technology, September 2009.
[2] Forrest Sheng Bao, Xin Liu, Christina Zhang, "PyEEG: An Open Source Python Module for EEG/MEG Feature Extraction", Computational Intelligence and Neuroscience, vol. 2011, Article ID 406391, 7 pages, 2011. https://doi.org/10.1155/2011/406391