mne_denoise.spectrum_interpolation.interpolate_spectrum#

mne_denoise.spectrum_interpolation.interpolate_spectrum(data: ndarray, sfreq: float, freqs: ndarray, *, bandwidth: float = 1.0, neighbour_width: float = 2.0) → ndarray[source]#

Interpolate line-noise amplitudes in a 2-D signal spectrum.

Parameters:

datandarray, shape (n_channels, n_times): Real-valued channel-first data.
sfreqfloat: Sampling frequency in Hz.
freqsarray-like: Target frequencies in Hz.
bandwidthfloat, default=1.0: Half-width of each replaced target band in Hz.
neighbour_widthfloat, default=2.0: Width of the neighboring reference bands in Hz.

Returns:

ndarray, shape (n_channels, n_times): Cleaned data with the same shape and units as data.

See also

SpectrumInterpolation: Estimator that resolves target frequencies and preserves MNE containers.

Notes

Amplitudes in target bins are replaced using neighboring amplitudes while the original FFT phase is retained [1].

References

Examples

>>> import numpy as np
>>> from mne_denoise.spectrum_interpolation import interpolate_spectrum
>>> rng = np.random.default_rng(0)
>>> data = rng.standard_normal((8, 2000))
>>> clean = interpolate_spectrum(data, sfreq=250.0, freqs=np.array([60.0, 120.0]))