facetpy 2.0.2

FACETpy - EEG-Data Correction Framework

A Python toolbox for correcting EEG artifacts using Averaged Artifact Subtraction (AAS) and other advanced methods. Built on MNE-Python.
Explore FACETpy docs »

Report bug · Request feature · Documentation

Built on MNE-Python, FACETpy provides a modular pipeline architecture that lets researchers process, evaluate, and compare correction results with minimal code.

Key features

• Load EEG from EDF, GDF, and BIDS formats
• Artifact correction: AAS, PCA, Adaptive Noise Cancellation (ANC)
• Full evaluation suite: SNR, RMS, Median Artifact, FFT-based metrics
• Batch processing across subjects/sessions with Pipeline.map()
• Generate synthetic EEG for algorithm testing
• Rich progress display in the terminal

Supported artifacts

For a detailed description of use cases (EEG-fMRI research, batch studies, benchmarking, clinical pipelines, and more) see the Use Cases & Supported Artifacts page in the documentation.

Quick start

Quick installation from PyPI (requires Python 3.11/3.12/3.13):

pip install facetpy

Strongly recommended for fast ANC performance:

python -m facet.build

For the Full Setup + Early Access to features (Repository, Examples, Contributing):
Unix (WSL/macOS/Linux) - bootstrap shortcut:

curl -fsSL https://raw.githubusercontent.com/H0mire/facetpy/main/scripts/bootstrap.sh | sh
cd facetpy

Preview

from facet import (
    Pipeline, Loader, EDFExporter,
    TriggerDetector, UpSample, DownSample, AASCorrection,
)

pipeline = Pipeline([
    Loader(path="data.edf", preload=True),
    TriggerDetector(regex=r"\b1\b"),
    UpSample(factor=10),
    AASCorrection(window_size=30),
    DownSample(factor=10),
    EDFExporter(path="corrected.edf", overwrite=True),
], name="Quickstart")

result = pipeline.run()
result.print_summary()   # Done in 4.2s  snr=18.3  rms_ratio=0.14

Installation

Requires Python 3.11, 3.12, or 3.13. For normal usage, Poetry is not required.

Normal usage (recommended): install from PyPI

pip install facetpy

The package name on PyPI is facetpy; import it in Python as facet.

Contributing setup (source + Poetry)

Poetry is required for contribution workflows (tests, linting, docs).

Unix (macOS/Linux) - bootstrap shortcut (installs poetry):

curl -fsSL https://raw.githubusercontent.com/H0mire/facetpy/main/scripts/bootstrap.sh | sh
cd facetpy

Unix (macOS/Linux) - existing clone (installs peotry):

./scripts/install.sh

Other platforms (including Windows) with Poetry installed:

git clone https://github.com/H0mire/facetpy.git
cd facetpy
poetry install --no-interaction

The Unix ./scripts/install.sh script:

• checks for Python 3.11/3.12/3.13
• checks whether Poetry is installed
• asks whether Poetry should be installed if missing
• runs poetry install --no-interaction

The bootstrap script:

• clones FACETpy into ./facetpy
• runs ./scripts/install.sh inside that clone

Manual Poetry installation (contributors):

# 1 — verify Python
python --version

# 2 — install Poetry (pick one)
pipx install poetry
# or: pip install --user poetry

# 3 — install repository dependencies
poetry install --no-interaction

Optional contributor extras:

poetry install -E deeplearning   # TensorFlow-based models
poetry install -E notebooks      # Jupyter notebook support
poetry install -E gui            # PyQt6 GUI components
poetry install -E docs           # Sphinx documentation toolchain
poetry install -E all            # everything above

Run contributor commands with poetry run ... (for example, poetry run pytest).

Build the C extension (strongly recommended for ANC)

The fast Adaptive Noise Cancellation (ANC) path is significantly faster with the compiled FastRANC C extension. Build it once after installing.

Without Poetry:

python -m facet.build

With Poetry:

poetry run build-fastranc

If the extension is not compiled, ANC uses a slower Python fallback and the rest of the toolbox still works.

Running the examples

All examples are in the examples/ folder and use the bundled NiazyFMRI.edf dataset.

To run repository examples, clone the repository and install FACETpy in your active Python environment (no Poetry required):

git clone https://github.com/H0mire/facetpy.git
cd facetpy
pip install facetpy

Then run from the project root:

# Recommended order for new users:
python examples/quickstart.py          # minimal pipeline
python examples/evaluation.py          # metrics & comparison
python examples/advanced_workflows.py  # conditional, parallel, factory
python examples/batch_processing.py    # multiple files at once
python examples/inline_steps.py        # custom steps & pipe operator
python examples/complete_pipeline_example.py  # full clinical pipeline
python examples/eeg_generation_visualization_example.py  # synthetic EEG

Testing

# Run the full test suite
poetry run pytest

# Only fast unit tests (skip slow integration tests)
poetry run pytest -m "not slow"

# A single test file
poetry run pytest tests/test_core_pipeline.py -v

# With coverage report
poetry run pytest --cov=facet --cov-report=html

Open the coverage report:

python -m webbrowser htmlcov/index.html

Documentation

# Install docs dependencies
poetry install -E docs

# Build HTML docs
poetry run sphinx-build -b html docs/source docs/build

Open docs locally:

python -m webbrowser docs/build/index.html

Full online documentation: https://facetpy.readthedocs.io/

For comprehensive build instructions, theme configuration, and contribution guidelines see docs/README.md. For PyPI release steps, see RELEASING.md.

Contributing

Contributing uses the source/Poetry workflow. See installation guide above.
Use ./scripts/install.sh on Unix, otherwise run poetry install --no-interaction. Follow docs/source/development/contributing.rst for the full setup and checks.

Project structure

src/facet/
├── core/           Pipeline, Processor, ProcessingContext, BatchResult
├── io/             Loader, BIDSLoader, EDFExporter, BIDSExporter
├── preprocessing/  Filters, Resample, TriggerDetector, Alignment, Transforms
├── correction/     AASCorrection, PCACorrection, ANCCorrection
├── evaluation/     SNRCalculator, RMSCalculator, MetricsReport, RawPlotter
├── misc/           EEGGenerator (synthetic data)
└── pipelines.py    create_standard_pipeline() factory

examples/
├── quickstart.py                         minimal pipeline
├── evaluation.py                         metrics & comparison
├── advanced_workflows.py                 conditional, parallel, factory
├── batch_processing.py                   multiple files
├── inline_steps.py                       custom steps & pipe operator
├── complete_pipeline_example.py          full clinical pipeline
└── eeg_generation_visualization_example.py  synthetic EEG

VS Code Tasks

Tasks are defined in .vscode/tasks.json and can be run via Ctrl+Shift+P → Tasks: Run Task.

License

GPLv3 — see LICENSE for details.

Author: Janik Michael Mueller