Hi all,
I’d like to introduce MNE-RT (link), an open-source Python package for real-time M/EEG signal processing, built on top of MNE-Python and MNE-LSL. It covers the entire closed-loop pipeline in a single, researcher-friendly API, aimed at neurofeedback, BCI, and real-time clinical/basic-science monitoring.
What it does:
-
21 real-time neural feature modalities in both sensor and source space
-
Real-time single-trial decoding (CSP + any scikit-learn classifier)
-
Adaptive feedback protocols: z-score, threshold, percentile, staircase, operant/RL-based, sham, multi-band, and cross-session transfer
-
Online artifact-correction methods: ASR, adaptive LMS, GEDAI, ORICA, real-time Maxwell/SSS filtering for MEG
-
Live visualization windows: Raw signal, NF feedback curves, epoch overlays, scalp topographies, 3D brain activity, TFR heatmaps, …
-
External feedback output via OSC (Max/MSP, SuperCollider) and LSL outlets (PsychoPy, OpenViBE, BCI2000)
-
BIDS-compatible session saving
-
Full CLI
Use cases it’s built for: neurofeedback research, real-time BCI/motor-imagery decoding, and any closed-loop paradigm needing live feature extraction + adaptive feedback + artifact correction in one place.
Any feedback, issues, feature requests, or contributions, especially from anyone doing real-time work who can stress-test it against their own hardware/paradigms are welcome!
Thanks!
Payam