Editorial: From Raw MEG/EEG to Publication: How to Perform MEG/EEG Group Analysis With Free Academic Software

Arnaud Delorme, Robert Oostenveld, François Tadel, Alexandre Gramfort, Srikantan S. Nagarajan, Vladimir LitvakFrontiers in Neuroscience 2022 · DOI · OpenAlex

Abstract

International audience. This editorial introduces a special collection of 25 research articles designed to bridge the gap between "single-subject" tutorials and the complex group-level analyses required for publication. By leveraging open-source toolboxes (like EEGLAB, FieldTrip, and MNE) and standardized data formats (BIDS), the collection provides a roadmap for reproducible, end-to-end neuroimaging workflows.

Introduction: The Challenge of Group-Level Analysis

In the fields of Magnetoencephalography (MEG) and Electroencephalography (EEG), most educational resources are "idealized." They typically demonstrate how to analyze a single, high-quality subject to showcase a specific tool's functionality. However, actual research involves "noisy" groups of participants, intricate statistical comparisons, and a labyrinth of preprocessing choices.

Researchers often struggle to transition from these simplified tutorials to the robust, automated pipelines required for peer-reviewed publications. This lack of standardized, well-documented group-level workflows contributes significantly to the "reproducibility crisis" in neuroscience. The authors of this Research Topic address this gap by curating a set of articles that provide not just scientific results, but the full code and data necessary for complete replication.

Methodological Framework

The editorial summarizes a "Research Topic" where contributors were tasked with providing detailed descriptions of complete group analyses. The methodological approach rests on several key pillars:

  • Open-Source Ecosystem: The papers utilize a variety of free academic software, including EEGLAB, FieldTrip, MNE-Python, SPM, Brainstorm, and NUTMEG. While MATLAB remains the dominant language (used in 21 of 25 papers), there is a growing presence of Python and integration with R.
  • Standardized Data (BIDS): To foster comparability, the editors encouraged the use of the Brain Imaging Data Structure (BIDS). This standard forces tool developers to use a common format, significantly increasing interoperability between different software packages.
  • Benchmark Datasets: 10 of the 25 papers utilized the HW Dataset (Henson and Wakeman, 2015), a multi-modal dataset containing MEG, EEG, and fMRI responses to face stimuli. The remaining papers used other publicly available data, ensuring that the pipelines are accessible to any researcher with an internet connection.

Summary of the Research Collection

The 25 accepted articles cover a wide spectrum of M/EEG research, including automated processing, source localization, and connectivity analysis. The following tables provide a comprehensive overview of the papers included in this special collection, detailing the tools, languages, and primary outcomes of each study.

Summary of research articles (Part 1) detailing authors, tools, and methodologies.
Summary of research articles (Part 2) detailing authors, tools, and methodologies.

Key Insights and Scientific Impact

The collection highlights several critical shifts in the field:

  • Interoperability: The adoption of BIDS allows researchers to, for example, preprocess data in EEGLAB and perform source localization in MNE-Python seamlessly.
  • Cross-Tool Validation: By processing the same benchmark dataset (HW) through different toolboxes, the community can perform "sanity checks." Consistent results across different algorithms increase confidence in the underlying neuroscientific findings.
  • Transparency: The focus on shareable scripts moves neuroimaging away from "black box" button-pressing toward transparent, version-controlled science.
  • Multimodal Integration: Several papers demonstrate joint EEG/MEG or even EEG/MEG/fMRI integration, reflecting a move toward a more holistic view of brain dynamics.

Limitations and Future Outlook

Despite the progress, the authors identified areas for improvement. Notably, 11 of the 25 papers did not specify a clear software license for their scripts, creating potential legal hurdles for adaptation. Additionally, open-source tools often lack the extensive testing budgets of commercial software, necessitating continued community-driven validation. The ultimate goal is a "plug-and-play" ecosystem where researchers can pick the best module from any toolbox without compatibility headaches, centered around the BIDS framework.

Summary updated

Abstract

International audience
BibTeX
@inproceedings{delorme2022editorial,
  author = {Arnaud Delorme and Robert Oostenveld and François Tadel and Alexandre Gramfort and Srikantan S. Nagarajan and Vladimir Litvak},
  title = {{Editorial: From Raw MEG/EEG to Publication: How to Perform MEG/EEG Group Analysis With Free Academic Software}},
  booktitle = {Frontiers in Neuroscience},
  year = {2022},
  doi = {10.3389/fnins.2022.854471},
}
Figure