meg-qc 1.0.6

Tool for automated MEG and EEG data quality control

MEEGqc

Automated quality assessment and quality control for MEG and EEG data.
Open-source, BIDS-aligned, schema-driven, ships interactive HTML reports and machine-readable BIDS derivatives.

---

What MEEGqc is

MEEGqc turns raw MEG and EEG recordings into auditable quality information at three nested scopes: subject-level (one recording / one subject), dataset-level (across all subjects in one dataset), and multi-dataset-level (across acquisition sites or studies). It supports both detailed per-recording inspection and large-scale dataset screening.

It separates two layers that are usually conflated:

• QA (quality assessment) is the continuous, unthresholded measurement of the signal: per-channel and per-epoch distributions, power spectral density (PSD), ECG / EOG correlations, muscle-band noise, head-movement traces.
• QC (quality control) is the criterion-based decisions made on top of QA, in two sub-layers: (1) per-metric flags from thresholds on QA outputs, and (2) the Global Quality Index (GQI), a single 0–100 composite score per recording with a transparent penalty breakdown across four families.

Both layers ship two output streams: BIDS derivatives (TSV + JSON sidecars) for programmatic re-use, and interactive HTML reports for human review.

---

Documentation

The complete documentation, including a hands-on tutorial with two real downloadable BIDS datasets, a full report reference, the metric and GQI math, and inline videos of every interactive view, lives at:

https://ancplaboldenburg.github.io/megqc_documentation/

This README is the short version. For anything beyond install + launch, start there.

---

Supported modalities and formats

Modality	Vendors / formats	Notes
MEG	FIF (.fif: MEGIN / Elekta and OPM exports), CTF (.ds)	Yokogawa / KIT, Ricoh, BTi / 4D not yet supported
EEG	EDF (.edf), BDF (.bdf), BrainVision (.vhdr), EEGLAB (.set), Neuroscan (.cnt), EGI (.mff), FIF (.fif)	CTF EEG (*_eeg.ds) is not auto-discovered

Inputs must be organised as a BIDS dataset. MEEGqc reads dataset_description.json and the per-modality file hierarchy (sub-XX/ses-YY/{meg,eeg}/), writes derivatives under derivatives/MEEGqc/, and uses the BIDS schema to validate downstream output.

Install

Two paths. Pick the one that fits.

Option A. One-click installer (recommended for end users)

Download the platform installer bundle and double-click. The installer creates an isolated Python 3.10 environment at ~/MEEGqc/, installs meg-qc into it, and registers a native app launcher (MEEGqc.app on macOS, MEEGqc.desktop on Linux, MEEGqc.exe on Windows).

Download the installer bundle (zip, ~12 KB)

Platform	Script inside the zip	Default install dir
macOS (arm64)	install_MEEGqc.command	~/MEEGqc/
Linux (x86_64)	install_MEEGqc.sh	~/MEEGqc/
Windows (x86_64)	install_MEEGqc.bat	~/MEEGqc/

After install, launch with the native app (no terminal needed) or via the CLI commands described below.

Option B. Manual install with pip (for users in their own Python env)

python3 -m venv ~/venvs/meegqc
source ~/venvs/meegqc/bin/activate
pip install meeg-qc       # rebrand-aligned distribution (wraps meg-qc)
# OR equivalently:
pip install meg-qc        # canonical PyPI name, same code, same release cadence

meeg-qc is a thin wrapper around meg-qc. Both ship from this repository in lockstep. Existing users keep pip install meg-qc and nothing changes.

Supported Python: 3.10, 3.11, 3.12, 3.13, 3.14.

Launch the GUI

meegqc      # rebrand-aligned alias
megqc       # legacy name, same entry point

The GUI has two top tabs: QA/QC calculation (run the engine on a BIDS dataset) and QA/QC plotting (build the HTML reports). Inputs, Output folder, and Log live in collapsible sections. App preferences (theme, CPU cores, RAM) are in the Settings dialog; the analysis parameters that drive the metrics are in the settings.ini editor.

For the full step-by-step GUI walkthrough (with paired light / dark screenshots and a video of the analysis-settings dialog), see the tutorial page.

Use the CLI

# 1. Export the default settings.ini into your config dir
get-meegqc-config --target_directory ./config

# 2. Run QA / QC calculation
run-meegqc --inputdata /path/to/bids_dataset --config ./config/settings.ini

# 3. Build interactive HTML reports
run-meegqc-plotting --inputdata /path/to/bids_dataset

# 4. (optional) Recompute the Global Quality Index summaries
globalqualityindex --inputdata /path/to/bids_dataset

# Pipeline in one command (calculation + plotting)
run-meegqc --inputdata /path/to/bids_dataset --config ./config/settings.ini --run-all

Both naming families work, you can mix and match. Every meeg* command is an alias for the same Python function as its meg* counterpart, so existing scripts using run-megqc continue to work unchanged.

Rebrand-aligned	Legacy (kept indefinitely)	Function it dispatches to
meegqc	megqc	GUI entry point
run-meegqc	run-megqc	QA / QC calculation pipeline
run-meegqc-plotting	run-megqc-plotting	HTML report builder
get-meegqc-config	get-megqc-config	Write the default settings.ini to a target directory
globalqualityindex	(no alias)	Recompute the GQI summaries on existing derivatives

Note the two flag spellings: run-megqc uses --n_jobs (with underscore); run-megqc-plotting uses --njobs (no underscore). They are not interchangeable.

Typical outputs

Default output tree inside the BIDS dataset's derivatives/:

derivatives/MEEGqc/
  calculation/        per-metric TSV + JSON summaries (the QA layer)
  reports/            interactive HTML reports (one per recording, plus group reports)
  summary_reports/    GQI artefacts (TSV + JSON), versioned via attempt files

The GQI is written per-modality: group_metrics/meg/Global_Quality_Index_attempt_<n>_meg.tsv and group_metrics/eeg/Global_Quality_Index_attempt_<n>_eeg.tsv.

The four report families (subject-level, dataset-level QA, dataset-level QC, multi-dataset-level) are documented in full with screenshots and videos on the docs site.

A note on the brand

The tool started as MEG-only and has grown first-class EEG support. The extra E in MEEGqc reflects that. A few legacy artefacts keep their old names so existing scripts and pipelines continue to work indefinitely:

• The Python import: import meg_qc (no meeg_qc package).
• The canonical PyPI distribution: meg-qc (the new meeg-qc is a thin meta-package that pulls in meg-qc).
• The legacy CLI commands (megqc, run-megqc, run-megqc-plotting, get-megqc-config) keep working alongside their meeg* siblings; both dispatch to the same Python function.

User-visible labels (GUI title, About dialog, log prefixes, version banner, installer apps and folders) all read MEEGqc.

For the complete rebrand changelog (what changed, what intentionally did not), see docs/REBRANDING.md. For EEG-specific implementation detail (the two glob fallbacks, the reference / montage / lobe handling, what each metric does on EEG vs MEG), see docs/EEG_Support_in_MEGqc.md.

Source code

https://github.com/ANCPLabOldenburg/MEEGqc

Citing

If you use MEEGqc in published work, please cite the software. The easiest way is the Cite this repository button on the right sidebar of the GitHub page, which generates APA / BibTeX from CITATION.cff.

For convenience, the BibTeX entry:

@software{meegqc_2026,
  author    = {Lopez Vilaret, Karel and Gapontseva, Evgeniia and Reer, Aaron and Bosch, Jorge and Karaca, Erdal},
  title     = {{MEEGqc}: automated quality assessment and quality control for {MEG} and {EEG} data},
  version   = {1.0.5},
  year      = {2026},
  publisher = {ANCP Lab, University of Oldenburg},
  url       = {https://github.com/ANCPLabOldenburg/MEEGqc},
  license   = {MIT},
}

License

MIT. See LICENSE.