Skip to main content

MNE Scan Quickstart

A 10–15 minute first-run walkthrough that brings a fresh install of MNE-CPP from "binary installed" to "FIFF file recorded from a simulated stream". No hardware required.

1. Install MNE-CPP​

Pick one of the following:

  • Binary release — download the latest installer for your OS from mne-cpp releases and run it. The installer lays out bin/, lib/, and plugins/ under a single prefix; see Deployment for the layout.
  • From source — follow the CMake build guide. The default build produces mne_scan plus all platform-compatible sensor plugins.

Verify the install by listing the plugins directory; you should see one shared library per entry in the Hardware Support Matrix.

2. Launch mne_scan​

From the install prefix:

# Linux / macOS
./bin/mne_scan

# Windows
bin\mne_scan.exe

The main window opens with three docks:

  • Plugin Toolbar (left) — every loaded plugin, grouped Sensor / Algorithm / IO.
  • Plugin Scene (centre) — the drag-and-drop canvas where you wire up the pipeline.
  • Log (bottom) — runtime messages from each plugin.

Empty main window after first launch

3. Add a Source Plugin (FiffSimulator)​

The FiffSimulator plugin replays a pre-recorded FIFF file and is the easiest way to test the pipeline without hardware.

  1. Drag FiffSimulator from the Plugin Toolbar onto the Plugin Scene.

  2. Click the plugin to open its configuration panel, then select the Connection tab.

  3. Start mne_rt_server in a separate terminal, pointing at the sample dataset:

    ./bin/mne_rt_server --file ./resources/data/MNE-sample-data/MEG/sample/ernoise_raw.fif

  4. Back in MNE Scan, set the host to 127.0.0.1 and press Connect. The status indicator should switch to Connected.

  5. Open the Preferences tab and set Block size equal to the sampling frequency (so the display refreshes once per second).

See Stream Pre-Recorded Data for a fuller walkthrough of FiffSimulator.

4. Add a Display / Recording Plugin​

You can either look at the data live, record it, or both. For the quickstart, do both:

  1. Drag writetofile from the IO group onto the scene.
  2. Hover over FiffSimulator's output connector (right side) and drag a connection to the writetofile input connector (left side). A solid line confirms the link is established.
  3. Click writetofile and pick an output path in its Preferences tab.

Optionally drag any algorithm plugin (e.g. noisereduction) between the two to verify processing also works.

Wired pipeline: FiffSimulator → writetofile

5. Start Streaming​

Press the green Run button in the top-left of the toolbar. You should observe:

  • The Log dock prints initialisation messages from every plugin in the pipeline.
  • Clicking on FiffSimulator shows the real-time channel display refreshing once per second.
  • The output FIFF file grows on disk as samples are written.

If any plugin fails to start, the failing connector turns red and the Log dock prints an [ERROR] line tagged with the plugin name.

6. Stop and Verify the FIFF Output​

Press the red Stop button when you have collected enough data. The output file is finalised (FIFF closing tags written) only after Stop completes — kill-9'ing mne_scan will leave a truncated file.

Verify the recorded file:

./bin/mne_show_fiff <path-to-output.fif>

You should see the standard FIFF block structure (FIFFB_MEAS_INFO, FIFFB_RAW_DATA, …) and a sample count that matches your recording duration × the source sampling frequency.

Next Steps​