mne.viz.plot_evoked_white

mne.viz.plot_evoked_white(evoked, noise_cov, show=True, rank=None, time_unit=None, verbose=None)[source]

Plot whitened evoked response.

Plots the whitened evoked response and the whitened GFP as described in [1]. This function is especially useful for investigating noise covariance properties to determine if data are properly whitened (e.g., achieving expected values in line with model assumptions, see Notes below).

Parameters:
evoked : instance of mne.Evoked

The evoked response.

noise_cov : list | instance of Covariance | str

The noise covariance. Can be a string to load a covariance from disk.

show : bool

Show figure if True.

rank : dict of int | None

Dict of ints where keys are ‘eeg’, ‘meg’, mag’ or ‘grad’. If None, the rank is detected automatically. Defaults to None. ‘mag’ or ‘grad’ cannot be specified jointly with ‘meg’. For SSS’d data, only ‘meg’ is valid. For non-SSS’d data, ‘mag’ and/or ‘grad’ must be specified separately. If only one is specified, the other one gets estimated. Note. The rank estimation will be printed by the logger for each noise covariance estimator that is passed.

time_unit : str

The units for the time axis, can be “ms” (default in 0.16) or “s” (will become the default in 0.17).

New in version 0.16.

verbose : bool, str, int, or None

If not None, override default verbose level (see mne.verbose() and Logging documentation for more).

Returns:
fig : instance of matplotlib.figure.Figure

The figure object containing the plot.

See also

mne.Evoked.plot

Notes

If baseline signals match the assumption of Gaussian white noise, values should be centered at 0, and be within 2 standard deviations (±1.96) for 95% of the time points. For the global field power (GFP), we expect it to fluctuate around a value of 1.

If one single covariance object is passed, the GFP panel (bottom) will depict different sensor types. If multiple covariance objects are passed as a list, the left column will display the whitened evoked responses for each channel based on the whitener from the noise covariance that has the highest log-likelihood. The left column will depict the whitened GFPs based on each estimator separately for each sensor type. Instead of numbers of channels the GFP display shows the estimated rank. Note. The rank estimation will be printed by the logger (if verbose=True) for each noise covariance estimator that is passed.

References

[1](1, 2) Engemann D. and Gramfort A. (2015) Automated model selection in covariance estimation and spatial whitening of MEG and EEG signals, vol. 108, 328-342, NeuroImage.