Motor imagery decoding from EEG data

curl https://bootstrap.pypa.io/get-pip.py -o get-pip.py
python get-pip.py --user

pip install mne matplotlib sklearn

MNE (https://mne.tools) is a library dedicated to the visualization and analysis of MEG, EEG, sEEG, ECoG signals.

matplotlib (https://matplotlib.org) is a scientific visualization library for producing quality images (especially for scientific publications).

sklearn (https://scikit-learn.org) is a library for machine learning.

The data set used contains EEG data acquired during different mental motor imagery (IMM) tasks. The 109 subjects performed 4 types of IMM tasks for 4 seconds, imagine:

• move the left hand
• move the right hand
• move both hands simultaneously
• move both feet simultaneously

64 EEG channels were recorded at 160Hz.

We will try to discriminate between the classes moving the right hand, and moving the feet. To do this, we use the algorithms of CSP (Common Spatial Patterns) and LDA (Linear Discriminant Analysis).

The different stages of treatment are:

• Filter the signal between 8 and 30 Hz
• Extract the windows linked to the events: move the right hand and the feet.
• Create the CSP + LDA model

• Evaluate the model by cross validation

We will identify a subject for whom cross-validation tends to show that it works and another for whom it does not work.

For each of these two subjects, display the patterns selected by the CSP.

Use the ICA to eliminate windows contaminated with blinking.