CLI

MNE-RT provides an mne-rt shell command with five sub-commands.

mne-rt --help
mne-rt --version
mne-rt info
mne-rt demo         [options]
mne-rt demo-erp     [options]
mne-rt baseline     --subject ID --subjects-dir DIR [options]
mne-rt run          --subject ID --subjects-dir DIR --duration N [options]

Global options:

Flag	Default	Description
--verbose, -v	WARNING	Logging verbosity level: DEBUG, INFO, WARNING, or ERROR

mne-rt info

Print the installed MNE-RT version and all key dependency versions.

$ mne-rt info
MNE-RT — Real-Time M/EEG Analysis
──────────────────────────────────
  mne-rt version : 1.0.0
  Python         : 3.11.9
  mne            : 1.8.0
  ...

mne-rt demo

Launch a full demo real-time session from simulated EEG (no amplifier needed). The real-time scalp topomap (δ/θ/α/β/γ) is shown by default. The 3-D brain display is enabled automatically when a FreeSurfer fsaverage5 directory is found (via FREESURFER_HOME or --subjects-fs-dir).

$ mne-rt demo
$ mne-rt demo --duration 120 --modality sensor_power erd_ers
$ mne-rt demo --no-topomap
$ mne-rt demo --no-brain

Options:

Flag	Default	Description
--duration	120	Session duration in seconds
--modality	sensor_power band_ratio entropy hjorth	Feature modality(ies) to demonstrate
--winsize	1.0	Analysis window length (s)
--no-nf	—	Disable the scrolling real-time NF signal plot (NFPlot)
--no-raw	—	Disable the scrolling raw M/EEG viewer (RawPlot)
--no-topomap	—	Disable the real-time scalp topomap (enabled by default)
--no-brain	—	Disable 3-D brain display (auto-enabled when FreeSurfer is found)
--subjects-fs-dir	—	FreeSurfer subjects directory (auto-detected from FREESURFER_HOME or SUBJECTS_DIR if not given)
--surf	inflated	Brain surface geometry (inflated, pial, or white)
--smoothing	0.25	EMA smoothing factor applied to each feature value (1.0 = no smoothing, 0.1 = heavy smoothing). Lower values produce smoother output at the cost of slower response to real changes in brain activity.
--no-save	—	Skip saving session data and report

mne-rt demo-erp

Launch an ERP demo that streams MNE sample-dataset EEG through a mock LSL player, collects auditory epochs trial-by-trial via RTEpochs, and drives the four epoch visualisation windows. Downloads the MNE sample dataset on first run (~1.5 GB).

$ mne-rt demo-erp                          # all four plot windows
$ mne-rt demo-erp --n-trials 50            # stop after 50 trials
$ mne-rt demo-erp --no-tfr                 # skip the TFR heatmap
$ mne-rt demo-erp --no-compare --no-tfr    # TopoPlot + ButterflyPlot only

Options:

Flag	Default	Description
--n-trials	70	Number of EEG trials to collect before stopping
--no-topo	—	Disable the scalp-layout TopoPlot (ERP display)
--no-butterfly	—	Disable the ButterflyPlot (all-channel overlay)
--no-compare	—	Disable CompareEvoked (per-channel comparison)
--no-tfr	—	Disable the TFRPlot (Morlet wavelet heatmaps)

mne-rt baseline

Record a resting-state baseline and compute the noise covariance and inverse operator needed for source-space modalities.

$ mne-rt baseline --subject sub01 --subjects-dir /data/subjects
$ mne-rt baseline --subject sub01 --subjects-dir /data --duration 180 --mock
$ mne-rt baseline --subject sub01 --subjects-dir /data --mock \
                  --fname /path/to/recording.fif

Required:

• --subject ID — subject identifier string

• --subjects-dir DIR — root directory with one folder per subject

Options:

Flag	Default	Description
--duration	120	Baseline duration in seconds
--session	01	BIDS session label (e.g. 01, pre, week1)
--mock	—	Use simulated data instead of live LSL
--fname	—	Any MNE-readable file (.fif, .vhdr, .edf, .bdf, .set, …) for --mock playback
--montage	easycap-M1	EEG montage name or .bvct path
--data-type	eeg	eeg or meg
--subjects-fs-dir	—	FreeSurfer subjects directory (required for source-space modalities)

mne-rt run

Run a real-time M/EEG session with feature extraction, feedback protocols, and live visualisation.

$ mne-rt run --subject sub01 --subjects-dir /data --duration 600 \
             --modality sensor_power erd_ers

# With epoch plots (requires a stimulus channel in the recording)
$ mne-rt run --subject sub01 --subjects-dir /data --duration 600 \
             --modality sensor_power \
             --topo --tfr --stim-ch "STI 014" \
             --event-id left=1 right=2

# With topomap display
$ mne-rt run --subject sub01 --subjects-dir /data --duration 600 \
             --modality sensor_power --topomap

# With OSC output to Max/MSP
$ mne-rt run --subject sub01 --subjects-dir /data --duration 600 \
             --modality sensor_power --osc-host 127.0.0.1 --osc-port 9000

# With LSL output (faster, same-machine integration)
$ mne-rt run --subject sub01 --subjects-dir /data --duration 600 \
             --modality sensor_power --lsl-output

# From a mock file with artifact correction
$ mne-rt run --subject sub01 --subjects-dir /data --duration 300 \
             --mock --fname /data/sub01/session.fif \
             --artifact-correction asr --topo

Options (inherits all baseline flags above, plus):

Flag	Default	Description
--duration	required	Session duration in seconds
--modality	sensor_power	One or more feature modalities (space-separated). See table below.
--winsize	1.0	Analysis window length (s)
--artifact-correction	—	lms, orica, gedai, asr, or maxwell (MEG only)
--no-nf	—	Disable the scrolling real-time NF signal plot (NFPlot)
--no-raw	—	Disable the scrolling raw M/EEG viewer (RawPlot)
--topomap	—	Enable real-time scalp topomap (δ/θ/α/β/γ bands)
--brain	—	Enable 3-D brain display (requires inverse operator)
--topo	—	Enable TopoPlot ERP display (requires --stim-ch)
--butterfly	—	Enable ButterflyPlot (requires --stim-ch)
--compare-evoked	—	Enable CompareEvoked (requires --stim-ch)
--tfr	—	Enable TFRPlot Morlet wavelet heatmap (requires --stim-ch)
--stim-ch	—	Stimulus/trigger channel name, e.g. STI 014 — required when any epoch plot flag is used
--tmin	-0.1	Epoch start relative to stimulus (s)
--tmax	0.5	Epoch end relative to stimulus (s)
--event-id	stimulus=1	Condition definitions as NAME=CODE pairs, e.g. --event-id left=1 right=2
--surf	inflated	Brain surface geometry
--osc-host	—	Enable OSC output; target hostname (e.g. 127.0.0.1)
--osc-port	9000	OSC destination port
--osc-prefix	/mne_rt	OSC address prefix
--lsl-output	—	Push feature values into an LSL stream outlet (faster than OSC for same-machine integration; readable by PsychoPy, Psychtoolbox, …)
--lsl-stream-name	MNE_RT	LSL stream name (only with --lsl-output)
--smoothing	0.25	EMA smoothing factor applied to each feature value (1.0 = no smoothing, 0.1 = heavy smoothing)

Available feature modalities

Modality key	Description
sensor_power	Mean band power across channels
band_ratio	Power ratio between two frequency bands (e.g. θ/β)
erd_ers	Event-related de/synchronisation (baseline normalised)
laterality	Log power asymmetry between right and left hemispheres
laterality_erd_ers	ERD/ERS computed separately per hemisphere
hjorth	Hjorth activity, mobility, complexity (time-domain)
spectral_centroid	Frequency-weighted spectral centroid
entropy	Spectral, approximate, or sample entropy
argmax_freq	Dominant frequency peak
individual_peak_power	Power at the individualised spectral peak
cfc_sensor	Cross-frequency coupling (sensor space)
instantaneous_phase	Instantaneous phase and amplitude envelope via Hilbert analytic signal
scp	Slow cortical potential — low-pass mean amplitude shift (DC-coupled)
peak_alpha_freq	Real-time peak alpha frequency tracker with EMA smoothing
sensor_connectivity	Functional connectivity (coh, plv, pli, wpli, corr, …); method set in config
connectivity_ratio	Ratio of connectivity between two channel pairs (e.g. C3–C4 / F3–F4)
sensor_graph	Graph-Laplacian learning from sensor connectivity
source_power	Source-space band power (requires inverse operator)
source_connectivity	Source-space functional connectivity
source_graph	Graph-Laplacian learning from source connectivity

For the mathematical background of every modality, see Real-time Feature Modalities.

Real-time visualisation

All eight plot windows are available as CLI flags and can be combined freely:

Flag	Plot class
--nf (default on)	NFPlot — scrolling multi-channel NF signal monitor
--raw (default on)	RawPlot — scrolling raw M/EEG channel viewer (bad-channel / bad-segment marking)
--epoch-plot	EpochPlot — scrolling raw viewer with trigger/epoch overlays (requires --stim-ch)
--topomap	TopomapPlot — live per-band scalp topography
--brain	BrainPlot — interactive 3-D cortical surface
--topo	TopoPlot — scalp-layout ERP with ±SEM shading
--butterfly	ButterflyPlot — all channels overlaid, region-coloured
--compare-evoked	CompareEvoked — per-channel comparison with SEM ribbons
--tfr	TFRPlot — Morlet wavelet TFR heatmaps

The first four (--nf, --raw, --topomap, --brain) work with continuous feature extraction (mne-rt run). The epoch-based windows (--epoch-plot, --erp, --butterfly, --compare-evoked, --tfr) require --stim-ch. Use mne-rt demo-erp to try them without a live recording.

See Visualisation for screenshots and feature descriptions of each plot.