eeg-preprocessing 0.2.4

Semiautomatic framework for preprocessing EEG data

eeg-preprocessing

A semiautomatic framework for preprocessing EEG data.

Overview

The eeg-preprocessing package serves as a cookbook for preprocessing EEG signals in a semiautomatic and reproducible way. The general use-case of the package is to use it from a Jupyter notebook. The tutorials folder contains a sample notebook that demonstrates data operations such as loading and writing the data along with the transformation steps that are described in the Background section.

Installation

Install the latest version from PyPI into an existing environment:

$ pip install eeg_preprocessing

Since this project is under development, I would recommend installing it from source in editable mode with pip:

$ git clone https://github.com/weiglszonja/eeg-preprocessing.git
$ cd eeg-preprocessing
$ pip install -e .

Alternatively, you can install it in a new virtual environment with conda:

$ git clone https://github.com/weiglszonja/eeg-preprocessing.git
$ conda env create -f eeg-preprocessing/make_conda_env.yml
$ source activate eeg_preprocessing

Background

Electroencephalography (EEG) measures neural activity by recording electrical signals at the level of populations of neurons. The signals that are recorded from multiple sensors are inherently contaminated by noise. Preprocessing aims to attenuate noise in the EEG data without removing meaningful signals in the process.

The eeg-preprocessing package aims to serve as a semiautomatic and reproducible framework for preprocessing EEG signals prior to time-frequency-based analyses. It minimizes the manual steps required to clean the data based on visual inspection and reduces the number of choices that depend on the researcher for rejecting segments of data or interpolation of electrodes. This package utilizes modules from mne-Python (Gramfort et al., 2013), a popular open-source Python package for working with neurophysiological data. For automated rejection of bad data spans and interpolation of bad electrodes it uses the Autoreject (Jas et al., 2017) and the Random Sample Consensus (RANSAC) ( Bigdely-Shamlo et al., 2015) packages.

The general use-case of the package is to use it from a Jupyter notebook. The tutorials folder contains a sample notebook that demonstrates data operations such as loading and writing the data along with the transformation steps that are described below.

In order to remove high-frequency artefacts and low-frequency drifts, a zero-phase band-pass filter (0.5 - 45 Hz) is applied to the continuous data using mne-Python. This temporal filter adapts the filter length and transition band size based on the cutoff frequencies. The lower and upper cutoff frequencies can be changed in the configuration file (config.py) located at the utils folder.

Subsequently, the filtered data is segmented into nonoverlapping segments ( epochs) to facilitate analyses. The default duration of epochs is one seconds, however it can be changed in the configuration file.

The removal of bad data segments is done in three steps. First, epochs are rejected based on a global threshold on the z-score (> 3) of the epoch variance and amplitude range. To further facilitate the signal-to-noise ratio, independent components analysis (ICA) is applied to the epochs. ICA is a source-separation technique that decomposes the data into a set of components that are unique sources of variance in the data. The number of components and the algorithm to use can be specified in the configuration file. The default method is the infomax algorithm that finds independent signals by maximizing entropy as described by (Bell & Sejnowski, 1995), (Nadal & Parga, 1999). Components containing blink artefacts are automatically identified using mne-Python. The interactive visualization of ICA sources lets the user decide which components should be rejected based on their topographies, time-courses or frequency spectra. The number of components that were removed from the data are documented in the “description” field of the epochs instance “info” structure. The final step of epoch rejection is to apply Autoreject (Jas et al., 2017) on the ICA cleaned data. Autoreject uses unsupervised learning to estimate the rejection threshold for the epochs. In order to reduce computation time that increases with the number of segments and channels, autoreject can be fitted on a representative subset of epochs (25% of total epochs). Once the parameters are learned, the solution can be applied to any data that contains channels that were used during fit.

The final step of preprocessing is to find and interpolate outlier channels. The Random Sample Consensus (RANSAC) algorithm (Fischler & Bolles, 1981) selects a random subsample of good channels to make predictions of each channel in small non-overlapping 4 seconds long time windows. It uses a method of spherical splines (Perrin et al., 1989) to interpolate the bad sensors.

Additionally, the EEG reference can be changed to a “virtual reference” that is the average of all channels using mne-Python.

Usage

The tutorials folder contains a sample jupyter notebook that demonstrates the preprocessing pipeline. See this documentation about supported EEG file formats.

$ jupyter notebook tutorials/run_preprocessing_tutorial.ipynb

Or you can import the methods into your own project and setup the pipeline with a Python script:

import os

from eeg_preprocessing.preprocessing import prepare_epochs_for_ica, run_ica, \
    run_autoreject, run_ransac
from eeg_preprocessing.utils.raw import read_raw_measurement
from eeg_preprocessing.utils.epochs import create_epochs


def run_pipeline(source: str):
    target_path = os.path.join(source, 'preprocessed')
    if not os.path.exists(target_path):
        os.makedirs(target_path)
    files = [file for file in os.listdir(source) if
             file.endswith(('.edf', '.vhdr', '.fif.gz'))]

    for file in files:
        raw = read_raw_measurement(raw_file_path=os.path.join(source, file))
        print(raw.info)

        # create epochs from filtered continuous data
        epochs = create_epochs(raw=raw)

        # initial rejection of bad epochs
        epochs_faster = prepare_epochs_for_ica(epochs=epochs)

        ica = run_ica(epochs=epochs_faster)
        # uncomment the line below to visualize ICA sources
        # block=True halts the execution of code until the plot is closed
        #ica.plot_sources(epochs_faster, start=0, stop=10, block=True)
        ica.apply(epochs_faster)
        epochs_faster.info['description'] = f'n_components: {len(ica.exclude)}'

        reject_log = run_autoreject(epochs_faster, n_jobs=11, subset=False)
        epochs_autoreject = epochs_faster.copy().drop(reject_log.bad_epochs,
                                                      reason='AUTOREJECT')

        epochs_ransac = run_ransac(epochs_autoreject)

        # set average reference
        epochs_ransac.set_eeg_reference('average')

        # save clean epochs
        fid = epochs_autoreject.info['fid']
        epochs_clean_fname = f'{fid}_ICA_autoreject_ransac'
        postfix = '-epo.fif.gz'
        epochs_ransac.save(
            os.path.join(target_path, f'{epochs_clean_fname}{postfix}'),
            overwrite=False)


if __name__ == '__main__':
    run_pipeline(source='/Volumes/crnl-memo-hd/EEG')

Tests

Run unittests from the terminal with:

$ python -m unittest

Contribution

This project is under development; comments are all welcome and encouraged! Suggestions related to this project can be made with opening an issue at the issue tracker of the project. Contributions and enhancements to the code can be made by forking the project first; committing changes to the forked project and then opening a pull request from the forked branch to the master branch of eeg-preprocessing.

References

Bell, A. J., & Sejnowski, T. J. (1995). An information-maximization approach to blind separation and blind deconvolution. Neural Computation, 7(6), 1129–1159. Bigdely-Shamlo, N., Mullen, T., Kothe, C.,

Su, K.-M., & Robbins, K. A. (2015). The PREP pipeline: standardized preprocessing for large-scale EEG analysis. In Frontiers in Neuroinformatics ( Vol. 9). https://doi.org/10.3389/fninf.2015.00016

Fischler, M. A., & Bolles, R. C. (1981). Random sample consensus. In Communications of the ACM (Vol. 24, Issue 6, pp. 381–395) . https://doi.org/10.1145/358669.358692

Gramfort, A., Luessi, M., Larson, E., Engemann, D. A., Strohmeier, D., Brodbeck, C., Goj, R., Jas, M., Brooks, T., Parkkonen, L., & Hämäläinen, M. ( 2013). MEG and EEG data analysis with MNE-Python. Frontiers in Neuroscience, 7, 267.

Jas, M., Engemann, D. A., Bekhti, Y., Raimondo, F., & Gramfort, A. (2017). Autoreject: Automated artifact rejection for MEG and EEG data. NeuroImage, 159, 417–429.

Nadal, J.-P., & Parga, N. (1999). SENSORY CODING: INFORMATION MAXIMIZATION AND REDUNDANCY REDUCTION. In Neuronal Information Processing (pp. 164–171) . https://doi.org/10.1142/9789812818041_0008

Perrin, F., Pernier, J., Bertrand, O., & Echallier, J. F. (1989). Spherical splines for scalp potential and current density mapping. Electroencephalography and Clinical Neurophysiology, 72(2), 184–187.