Processing Resting State EEG Data#

Resting state EEG is one of the most common paradigms in neuroscience research. This guide shows you how to process resting state data effectively with AutoClean.

🎯 What is Resting State EEG?#

Resting state EEG records brain activity while participants are at rest, typically: - Sitting quietly with eyes open or closed - Not performing any specific task - Usually 5-10 minutes of recording - Used to study baseline brain activity and connectivity

Common research applications: - Brain connectivity analysis - Power spectral analysis - Individual differences in brain activity - Clinical assessments - Biomarker research

📊 The RestingEyesOpen Task#

AutoClean includes a specialized task for resting state data that: - Removes common artifacts (eye blinks, muscle tension, line noise) - Applies optimal filtering for resting state analysis - Creates clean epochs suitable for connectivity analysis - Generates comprehensive quality control reports

What it does automatically: 1. Filtering: Removes slow drifts and high-frequency noise 2. Artifact Detection: Identifies bad channels and time periods 3. ICA Cleaning: Removes eye movements, muscle artifacts, and heartbeat 4. Epoching: Creates analysis-ready data segments 5. Quality Control: Provides detailed processing reports

🚀 Quick Start: Process Resting State Data#

For a single file:

autoclean process RestingEyesOpen subject001_rest.raw

For multiple files in a folder:

autoclean process RestingEyesOpen data_folder/

With custom output location:

autoclean process RestingEyesOpen subject001_rest.raw --output results/

⚙️ Understanding the Processing Steps#

The RestingEyesOpen task performs these steps automatically:

1. Data Import and Validation

  • Loads your EEG file

  • Validates data integrity

  • Sets up electrode positions

2. Basic Preprocessing

  • Resamples to 250 Hz (reduces file size, maintains quality)

  • Applies bandpass filter (1-100 Hz) to remove drifts and noise

  • Removes line noise (50/60 Hz) automatically

3. Channel and Artifact Detection

  • Identifies bad channels with poor signal quality

  • Detects noisy time periods

  • Flags excessive artifacts for review

4. Advanced Cleaning

  • Runs Independent Component Analysis (ICA)

  • Automatically classifies components as brain activity vs. artifacts

  • Removes eye movements, muscle tension, and heartbeat artifacts

5. Epoching and Quality Control

  • Creates 2-second epochs for analysis

  • Removes epochs with residual artifacts

  • Generates comprehensive quality reports

📈 Interpreting Your Results#

After processing, you’ll find these key outputs:

Quality Control Report (metadata/run_report.pdf)

Look for these important metrics:

Data Quality Indicators: - Channels kept: Should be >90% for good quality data - Epochs kept: Should be >70% for reliable results - ICA components removed: Typically 3-8 artifact components

Red flags to watch for: - ❌ <70% channels kept = poor electrode contact - ❌ <50% epochs kept = very noisy data - ❌ >15 ICA components removed = possible over-cleaning

Processed Data Files

bids/derivatives/cleaned_data/ - continuous_clean.fif: Artifact-free continuous data - epochs_clean.fif: Clean 2-second epochs ready for analysis - ica_solution.fif: ICA decomposition for review

stage/ (intermediate files) - Raw data at different processing stages - Useful for troubleshooting or custom analysis

🔧 Customizing for Your Research#

The default RestingEyesOpen task works well for most studies, but you can customize it:

Common modifications needed:

Different epoch length: - Default: 2-second epochs - For connectivity: Often want 4-8 second epochs - For spectral analysis: 2-4 second epochs work well

Different frequency bands: - Default: 1-100 Hz bandpass - For alpha analysis: Might want 0.5-40 Hz - For gamma: Might want 1-150 Hz

Stricter/looser artifact rejection: - Default: Balanced for most studies - Clinical data: Often needs looser criteria - High-precision research: Might need stricter criteria

To customize: See Creating a Custom Task for detailed instructions.

📊 Best Practices for Resting State#

Data Collection Tips: - Record at least 5 minutes (8-10 minutes preferred) - Consistent instructions across participants - Note eyes open vs. eyes closed conditions - Minimize environmental distractions

Processing Recommendations: - Always review quality control reports - Check that >70% of epochs are retained - Verify ICA removed appropriate artifacts - Document any custom processing parameters

Analysis Considerations: - First/last minute often noisier - consider excluding - Eyes open vs. closed have different spectral profiles - Individual differences in alpha frequency are normal - Connectivity measures sensitive to residual artifacts

🆘 Troubleshooting Common Issues#

“Too many bad channels” warning:

# Check your electrode montage and impedances
# Bad channels usually indicate:
# - Poor electrode contact
# - Broken electrodes
# - Wrong montage specification

Solutions: - Verify electrode positions were set correctly - Check original data quality - Consider manual bad channel marking before processing

“Insufficient clean epochs” error:

# This means >50% of data was marked as artifactual
# Common causes:
# - Very noisy environment during recording
# - Participant movement/talking
# - Equipment malfunction

Solutions: - Review original recording quality - Consider looser artifact detection settings - Check if data is actually resting state (not task)

ICA removes too many/few components:

Too many (>10): - Data might be very noisy - ICA may be over-fitting - Consider pre-cleaning steps

Too few (<2): - Participant had very little eye movement - Very clean data (this is good!) - Verify eye artifacts are actually removed

Processing takes very long:

Normal processing time: - 10 minutes data: ~3-5 minutes processing - 30 minutes data: ~8-12 minutes processing

If much slower: - Computer may be low on memory - Other programs using CPU - Very large file size

📋 Batch Processing Multiple Participants#

Organize your data:

data/
├── sub-001_rest.raw
├── sub-002_rest.raw
├── sub-003_rest.raw
└── ...

Process all files:

autoclean process RestingEyesOpen data/

Monitor progress: - AutoClean will process each file sequentially - Check the command window for progress updates - Each file gets its own results folder

Quality control for batches: - Review summary statistics across participants - Flag participants with unusual processing metrics - Check for systematic issues across the dataset

🎯 Advanced Analysis Integration#

For Python users:

import mne
from autoclean import Pipeline

# Process data
pipeline = Pipeline()
pipeline.process_file("subject001_rest.raw", "RestingEyesOpen")

# Load results for analysis
epochs = mne.read_epochs("output/subject001_rest_*/bids/derivatives/epochs_clean.fif")

# Your analysis code here
psd = epochs.compute_psd()
connectivity = mne_connectivity.spectral_connectivity_epochs(epochs)

For MATLAB/EEGLAB users:

% Load AutoClean results
EEG = pop_loadset('epochs_clean.set', 'output/subject001_rest_*/bids/derivatives/');

% Continue with your analysis
[spectra, freqs] = spectopo(EEG.data, 0, EEG.srate);

For R users:

library(eegUtils)

# Load processed data
eeg_data <- import_set("output/subject001_rest_*/bids/derivatives/epochs_clean.set")

# Continue analysis
psd <- compute_psd(eeg_data)

🎉 Next Steps#

Now that you can process resting state data:

  1. Quality Control: Learn to systematically review processing quality

  2. Batch Processing: Scale up to process entire datasets

  3. Custom Tasks: Create specialized workflows for your research

  4. Integration: Connect AutoClean to your analysis pipeline

Recommended follow-up tutorials: - quality_control_best_practices - batch_processing_datasets - python_integration - Creating a Custom Task