Automated process for detecting and rejecting EEG artifacts.

This ongoing process explores supervised learning methods to detect artifacts in EEG data and possibly other time series.

• Current models are often task specific
• Feature engineering & selection
• High dimensionality
• High variability between subjects
• Low signal-to-noise ratio
• Non-stationary signal

• Machine Learning

  • Supervised:

    • Random Forest Classifier/Regressor
    • Support Vector Classifier/Machine

  • Unsupervised:

    • Isolation Forest

• Deep Learning

  • Supervised:
    • Convolutional Neural Network (CNN)*
    • Recurrent Neural Network (RNN)

*CNN will be our goal for the final model.

• F1 Score
• Precision
• Recall
• AUC-ROC Curve

• Anaconda: within the tbear directory containing the file environment.yml perform:
  • Problems may arise with Windows users.

conda env create -f environment.yml

• pip

pip install numpy scipy matplotlib pandas scikit-learn jupyter mne tensorflow

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

Inspiration, code snippets, etc.

• Gramfort, M. Luessi, E. Larson, D. Engemann, D. Strohmeier, C. Brodbeck, L. Parkkonen, M. Hämäläinen, MNE software for processing MEG and EEG data, NeuroImage, Volume 86, 1 February 2014, Pages 446-460, ISSN 1053-811
• Roy, Yannick & Banville, Hubert & Albuquerque, Isabela & Gramfort, Alexandre & Faubert, Jocelyn. (2019). Deep learning-based electroencephalography analysis: a systematic review.