BCI Feature Extraction

Extracts features from epoched EEG data for Brain-Computer Interface classification. Supports band power, CSP, and Riemannian geometry methods.

Usage

eegBCIfeatures(
  x,
  method = c("bandpower", "csp", "riemannian"),
  labels = NULL,
  bands = NULL,
  assay_name = NULL
)

Arguments

x

A PhysioExperiment object with epoched (3D) EEG data (time x channels x trials).

method

Feature extraction method: "bandpower" (log band power), "csp" (Common Spatial Pattern log-variance), or "riemannian" (tangent space projection of covariance matrices).

labels

Character or factor vector of class labels (required for "csp" method). One label per trial.

bands

Named list of frequency bands for "bandpower" method. Default: list(mu = c(8, 13), beta = c(13, 30)).

assay_name

Input assay name (default: first assay).

Value

A numeric matrix with n_trials rows and feature columns. Number of columns depends on method:

• "bandpower": n_channels * n_bands

• "csp": 2 * n_filters (default: 6)

• "riemannian": n_channels * (n_channels + 1) / 2

References

Blankertz, B., et al. (2008). Optimizing spatial filters for robust EEG single-trial analysis. IEEE Signal Processing Magazine, 25(1), 41-56.

See also

eegCSP(), eegMotorImagery(), eegBCIclassify()

Examples

if (FALSE) { # \dontrun{
pe <- make_eeg_bci(n_trials = 20, n_channels = 8, sr = 256)
features <- eegBCIfeatures(pe, method = "bandpower")
} # }