Advanced Neurofeedback Toolbox
Real-time M/EEG neurofeedback for research and clinical use

---

ANT is an open-source Python package for real-time closed-loop M/EEG neurofeedback. Built on MNE-Python and the Lab Streaming Layer (LSL), it covers the full pipeline — from amplifier to 3D brain display — in a single, researcher-friendly API.

Highlights

Feature	Details
14+ NF modalities	Alpha power, ERD/ERS, laterality, Hjorth, spectral centroid, CFC, graph metrics, source power …
Sensor & source space	Full MNE inverse-operator pipeline for source-level NF
Real-time artifact correction	ORICA (online ICA), adaptive LMS, GEDAI (GED-based spatial filters)
Three parallel windows	Raw stream viewer · NF signal monitor · 3D brain activation
OSC output	Send feedback values to Max/MSP, SuperCollider, Pure Data
CLI	ANT demo, ANT baseline, ANT run — no Python required
Mock mode	Works without hardware using bundled sample EEG data

Installation

pip install ANT                 # core package (OSC output included)
pip install "ANT[full]"         # all extras (viz, dev, docs)

Other installation methods

uv (fast Rust-based installer):

uv pip install ANT

conda / mamba:

conda env create -f environment.yml
conda activate ant

Development install from source:

git clone https://github.com/payamsash/ANT.git
cd ANT
pip install -e ".[dev]"

Verify the installation:

ANT info     # print versions of ANT and all key dependencies
ANT demo     # run a 60-second mock neurofeedback session

Quick start

from ant import NFRealtime

# 1 — Create a session object
nf = NFRealtime(
    subject_id="sub01",
    visit=1,
    session="main",
    subjects_dir="/data/subjects",
    montage="easycap-M1",
)

# 2 — Connect to a live LSL stream (or a mock replay)
nf.connect_to_lsl(mock_lsl=True)

# 3 — Record a resting-state baseline (computes inverse operator)
nf.record_baseline(baseline_duration=120)

# 4 — Run the closed-loop NF session
nf.record_main(
    duration=300,
    modality=["sensor_power", "erd_ers", "laterality"],
    show_nf_signal=True,
    show_brain_activation=True,
)

Architecture

_{Click to open the interactive diagram}

The acquisition loop runs in a background daemon thread; all three visualisation windows share a Qt event loop on the main thread, updated at ~30 fps via a pump timer.

Available NF modalities

Key	Description
sensor_power	Mean band power across channels
band_ratio	Power ratio between two bands (e.g. θ/β)
erd_ers	Event-related de/synchronisation (baseline-normalised)
laterality	Log power asymmetry right vs. left hemisphere
hjorth	Hjorth activity, mobility, complexity
spectral_centroid	Frequency-weighted spectral centroid
entropy	Spectral / approximate / sample entropy
argmax_freq	Dominant frequency peak
individual_peak_power	Power at the individual spectral peak
cfc_sensor	Cross-frequency coupling (sensor space)
sensor_connectivity	Functional connectivity (PLI, correlation)
sensor_graph	Graph Laplacian from sensor connectivity
source_power	Source-space band power
source_connectivity	Source-space functional connectivity
source_graph	Graph Laplacian from source connectivity

CLI

# Quick demo — no amplifier needed
ANT demo --duration 60 --modality sensor_power band_ratio

# Record a resting-state baseline
ANT baseline --subject sub01 --subjects-dir /data --duration 120 --mock

# Run a full NF session with OSC output
ANT run --subject sub01 --subjects-dir /data --duration 600 \
        --modality sensor_power erd_ers \
        --osc-host 127.0.0.1 --osc-port 9000

Documentation

Full documentation, API reference, and gallery examples:
payamsash.github.io/ANT

Cite

If you use ANT, please cite:

@inproceedings{shabestari2025advances,
  title     = {Advances on Real Time M/EEG Neural Feature Extraction},
  author    = {Shabestari, Payam S and Ribes, Delphine and D{\'e}fayes, Lara
               and Cai, Danpeng and Groves, Emily and Behjat, Harry H
               and Van de Ville, Dimitri and Kleinjung, Tobias
               and Naas, Adrian and Henchoz, Nicolas and others},
  booktitle = {2025 IEEE 38th International Symposium on Computer-Based
               Medical Systems (CBMS)},
  pages     = {337--338},
  year      = {2025},
  organization = {IEEE}
}

Acknowledgements

Development was supported by the Swiss National Science Foundation (grant number - 208164 — Advancing Neurofeedback in Tinnitus).

License

MIT License — © 2025 Payam S. Shabestari