This repository is an example of creating individual event-related des/synchronization ERDS maps and plots for EEG Motor Imagery data

The repository contains code and instructions to read the "EEG datasets for motor imagery brain-computer interface" which is in .mat format and transforms it to MNE-python classes. Also demonstrates how to apply Artifact Subspace Reconstruction (ASR) to clean the data. Contains code to create epochs and also includes custom functions that uses MNE-python tfr (time-frequency) class to create ERDS maps and plots, and a report in .html format

Example plots from analysis of subject 41 of the dataset:

ERDS Map - Motor Imagery Rigth - s41 - ASR - Wavelet ERDS Map - Motor Imagery Left - s41 - ASR - Wavelet ERDS Map - Motor Imagery Rigth - s41 - ASR - Multitapers ERDS Map - Motor Imagery Left - s41 - ASR - Multitapers ERDS Plot - Motor Imagery - s41 - ASR - Wavelet ERDS Plot - Motor Imagery - s41 - ASR - Wavelet

Dataset

The "EEG datasets for motor imagery brain-computer interface" includes a 64-channels EEG recordings from 52 subjects obtained during motor imagery and execution tasks.

An example of how to read from the dataset and convert to .fif files for MNE-python can be found in the notebook read_raw_data.

Artifact Subspace Reconstruction (ASR)

ASR is a technique used to remove artifacts from EEG data. This repository uses the ASR functions from asrpy.

An example of how to apply ASR to the dataset is presented in the notebook ASR_processing.

ERDS (Event-Related Desynchronization/Synchronization) Maps

ERDS maps are used to visualize changes in brain activity during specific events or tasks. They show the relative power of different frequency bands at different time points. ERDS maps are commonly used in brain-computer interface research to analyze motor imagery tasks.

To generate ERDS maps from the EEG data, follow these steps:

1. Preprocess the EEG data and create epochs. In this case, epochs are created from the ASR pre-processed data. Check the notebook epochs_ASR on epoching.
2. Compute the power spectrum for of the epochs and create a tfr (time-frequency) MNE-python object. Check the notebook ERDS_plots.
3. Calculate the ERDS values by comparing the power spectrum to a baseline period and average across epochs. Check the notebook ERDS_plots.
4. Visualize the ERDS maps and plot using the custom functions in utils. Check the notebook ERDS_plots.

Usage

1. Download the "EEG datasets for motor imagery brain-computer interface" dataset from [https://doi.org/10.1093/gigascience/gix034].
2. Run the provided scripts or modify them according to your needs.

Credits

The subsequent repositories/data were utilized, either closely adhering to their original form or serving as a source of inspiration:

• EEG datasets for motor imagery brain-computer interface [https://doi.org/10.1093/gigascience/gix034]
• ASR repository [https://github.com/DiGyt/asrpy]
• MNE-python [https://mne.tools/stable/index.html]
• MNE-python Time-Frequency ERDS Tutorial [https://mne.tools/stable/auto_examples/time_frequency/time_frequency_erds.html]