cogpy.spectral.features
Spectral features derived from pre-computed PSD estimates.
Status
STATUS: ACTIVE Reason: PSD-first spectral feature extraction for iEEG channel characterization. Superseded by: n/a Safe to remove: no
- Convention:
psd : (…, freq) as returned by psd_welch or psd_multitaper freqs : (freq,) Hz, strictly increasing scalar output : (…,) freq axis reduced vector output : (…, freq) same freq axis as input
Functions
|
log10(P_sideband / P_background). |
|
log10(P_sideband / P_carrier). |
|
Fit 1/f aperiodic exponent via specparam (fooof). |
|
Integrate PSD over band using trapezoidal rule. |
|
log10(P_signal / P_noise). |
|
Periodic component above 1/f background via specparam (fooof). |
|
F-test line scan across all frequency bins. |
|
P_line / P_flanks. |
|
Narrowband prominence ratio per frequency bin. |
|
Reduce a frequency axis to per-band scalars. |
|
band_power(band) / band_power(norm_range). |
|
Frequency below which fraction p of total power lies (SEF). |
|
Shannon entropy of normalized PSD. |
|
Spectral flatness (Wiener entropy). |
|
Detect discrete peak frequencies in a PSD. |