Artifact Handling#

Remove or correct artifacts such as eye blinks, muscle activity, and bad segments from your Wearable Sensing EEG recordings.

Manual Artifact Rejection#

Mark bad segments manually:

  1. Plot → Plot data

  2. Tools → Annotations to enable annotation mode

  3. Mark artifacts by clicking and dragging on the plot

  4. Label annotations (e.g., “blink”, “movement”)

Annotations vs. Rejection

Annotated segments are marked but not deleted. This allows you to:

  • Exclude them from epochs/analysis later

  • Review annotations before committing

  • Keep a record of data quality

Independent Component Analysis (ICA)#

Use ICA for automated artifact correction:

Fitting ICA#

  1. Tools → ICA

  2. Click Fit ICA

  3. Choose number of components (default: all channels)

  4. Wait for decomposition (may take time for long recordings)

Identifying Artifact Components#

After fitting, MNELAB displays:

  • Time courses of each component

  • Topographic maps

  • Power spectra

Common artifacts to look for:

  • Eye blinks: Frontal topography, low frequency

  • Lateral eye movements: Temporal topography, horizontal pattern

  • Muscle activity: Temporal topography, high frequency

  • Cardiac: Regular rhythmic pattern

Removing Components#

  1. Select artifact components (click to highlight)

  2. Click Apply to remove selected components

  3. A new dataset is created with artifacts removed

ICA Best Practices

  • Filter data (1-40 Hz) before ICA for better decomposition

  • Use enough data (at least 1-2 minutes) for stable components

  • Remove extreme artifacts manually before ICA

  • Verify component removal doesn’t affect signal of interest

Marking Bad Channels#

If specific channels are consistently noisy:

  1. Plot → Plot data

  2. Right-click on channel name

  3. Select Mark as bad

  4. Bad channels are excluded from average reference and analyses

Bad Channel Interpolation

After marking bad channels, use Tools → Interpolate bad channels to replace them with interpolated values from neighboring channels.

Next Steps#

After cleaning your Wearable Sensing data:

  1. Create epochs - Extract event-related segments for analysis

  2. Save cleaned data: File → Save as (FIF format recommended for MNE compatibility)

  3. Export to MNE-Python for advanced statistics and visualization

Resources#