Denoising EEG Signals using ICA

Justin Nguyen, Mingxiao An, Siddharth Mehta, Yi-Chia Wu

Semester Project for 18-797 with Dr. Bhiksha Raj

Project Structure

Our project is centered around the BuildingBlocks directory which contains our single subject experiments and "building blocks" for putting together MI denoising and classification trials across the entire dataset.

The directories: BaselineClassification, BlinkingDenoiseClassification, EyeMovementLRDenoiseClassification, EyeMovementUDDenoiseClassification, HeadMovementDenoiseClassification, JawMovementDenoiseClassification contain our denoising and classification trails for each noise type.

Cho2017.py defines the class for processing the dataset and plotting.ipynb is a notebook which generates the plots for our presentation and reports.

Contributions

Justin Nguyen: Cho2017.py, EEG experiments in BuildingBlocks, and the starter notebooks for the Corrmap denoising notebooks for the entire dataset trials.

Mingxiao An: plotting.ipynb and all the bug fixes relating to getting the ICA de-noised signal and the classifier (CSP+LDA) working.

Siddarth Mehta: MI-Classifier notebooks and setting up the classification pipeline and setting up the dataset hold-out, crossvalidation, and testing.

Yi-Chia Wu: MI-Classifier notebooks and transferred starter notebooks for use on just blink artifacts to eye-movement, head-movement, and jaw-movement noise types.