EegFun

Appearance

Channel Metrics ​

This demo demonstrates how to calculate and visualize channel quality metrics for identifying bad channels and detecting artifacts.

What are Channel Metrics? ​

Channel quality metrics quantify various characteristics of the EEG signal at each electrode. These metrics help identify problematic channels that may need repair or rejection before analysis.

Variance:

  • Measures overall signal power/amplitude variability at each channel

  • High variance may indicate artifacts (muscle, movement)

  • Very low variance suggests poor contact or disconnected electrode

Kurtosis:

  • Measures the distribution shape (tailedness) of the signal

  • High kurtosis indicates spiky signals with extreme values (artifacts)

Correlation:

  • Average correlation between a channel and its spatial neighbors

  • Low correlation suggests the channel is behaving differently from nearby electrodes

  • Detects bad contacts, bridging, or electrode-specific noise

Joint Probability:

  • Statistical outlier detection across channels simultaneously

  • Identifies channels that deviate from the typical multi-channel pattern

  • Combines information from variance and correlation metrics

Use Cases ​

Quality control:

  • Identify bad channels before epoching or averaging

  • Screen data quality during or after acquisition

  • Track bad channels across participants - if a specific electrode is consistently bad across multiple participants, this may indicate a broken electrode or faulty electrode set that needs replacement

Workflow Summary ​

This demo shows channel quality assessment workflows:

Basic Channel Metrics ​

  • Load and preprocess raw data (average reference, high-pass filter)

  • Calculate channel joint probability metrics

  • Identify channels with extreme values

EOG Correlation Analysis ​

  • Compute vertical and horizontal EOG channels

  • Detect EOG onsets automatically

  • Calculate correlation between EEG channels and EOG

  • Partition bad channels into EOG-related vs. non-EOG artifacts

Metric Interpretation ​

  • Add z-score columns for standardized thresholds

  • Identify channels that are artifact-related vs. bad contacts

  • Guide decisions on repair (interpolation) vs. rejection

Code Examples ​

Show Code
julia
# Demo: Channel Metrics
# Shows how to compute and visualize channel-wise metrics (variance, kurtosis, etc.).

# Note: EegFun.example_path() resolves bundled example data paths.
# When using your own data, simply pass the file path directly, e.g.:
# dat = EegFun.read_raw_data("/path/to/your/data.bdf")

using EegFun
using GLMakie

# read raw data
dat = EegFun.read_raw_data(EegFun.example_path("data/bdf/example1.bdf"));

# read and prepare layout file
layout = EegFun.read_layout(EegFun.example_path("layouts/biosemi/biosemi72.csv"));
EegFun.polar_to_cartesian_xy!(layout)

dat = EegFun.create_eegfun_data(dat, layout)

# minimal preprocessing
EegFun.rereference!(dat, :avg)
EegFun.highpass_filter!(dat, 0.1)

# extreme value bool column
EegFun.is_extreme_value!(dat, 100)

# channel joint probability
channel_joint_probability = EegFun.channel_joint_probability(dat)
channel_joint_probability = EegFun.channel_joint_probability(dat, sample_selection = EegFun.samples_not(:is_extreme_value_100))

# Calculate EOG signals
EegFun.channel_difference!(
    dat,
    channel_selection1 = EegFun.channels([:Fp1, :Fp2]),
    channel_selection2 = EegFun.channels([:IO1, :IO2]),
    channel_out = :vEOG,
); # vertical EOG = mean(Fp1, Fp2) - mean(IO1, I02)
EegFun.channel_difference!(
    dat,
    channel_selection1 = EegFun.channels([:F9]),
    channel_selection2 = EegFun.channels([:F10]),
    channel_out = :hEOG,
); # horizontal EOG = F9 - F10

EegFun.detect_eog_onsets!(dat, 50, :vEOG, :is_vEOG)
EegFun.detect_eog_onsets!(dat, 30, :hEOG, :is_hEOG)


# Calculate correlations between all channels and EOG channels
cm = EegFun.correlation_matrix_dual_selection(
    dat,
    sample_selection = EegFun.samples(),  # All samples
    channel_selection1 = EegFun.channels(),  # All EEG channels
    channel_selection2 = EegFun.channels([:vEOG, :hEOG]),  # EOG channels
)
EegFun.add_zscore_columns!(cm)

bad_channels = [:Fp1, :AF3]
non_eog_related, eog_related = EegFun.partition_channels_by_eog_correlation(bad_channels, cm)

See Also ​