Interpolate bad channels for MEG/EEG channels

This example shows how to interpolate bad MEG/EEG channels

• Using spherical splines from [1] for EEG data.

• Using field interpolation for MEG and EEG data.

In this example, the bad channels will still be marked as bad. Only the data in those channels is replaced.

# Authors: Denis A. Engemann <denis.engemann@gmail.com>
#          Mainak Jas <mainak.jas@telecom-paristech.fr>
#
# License: BSD-3-Clause

import mne
from mne.datasets import sample

print(__doc__)

data_path = sample.data_path()
meg_path = data_path / 'MEG' / 'sample'
fname = meg_path / 'sample_audvis-ave.fif'
evoked = mne.read_evokeds(fname, condition='Left Auditory',
                          baseline=(None, 0))

# plot with bads
evoked.plot(exclude=[], picks=('grad', 'eeg'))

Reading /home/circleci/mne_data/MNE-sample-data/MEG/sample/sample_audvis-ave.fif ...
    Read a total of 4 projection items:
        PCA-v1 (1 x 102) active
        PCA-v2 (1 x 102) active
        PCA-v3 (1 x 102) active
        Average EEG reference (1 x 60) active
    Found the data of interest:
        t =    -199.80 ...     499.49 ms (Left Auditory)
        0 CTF compensation matrices available
        nave = 55 - aspect type = 100
Projections have already been applied. Setting proj attribute to True.
Applying baseline correction (mode: mean)

Compute interpolation (also works with Raw and Epochs objects)

evoked_interp = evoked.copy().interpolate_bads(reset_bads=False)
evoked_interp.plot(exclude=[], picks=('grad', 'eeg'))

Interpolating bad channels
    Automatic origin fit: head of radius 91.0 mm
Computing interpolation matrix from 59 sensor positions
Interpolating 1 sensors
    Computing dot products for 305 MEG channels...
[Parallel(n_jobs=1)]: Using backend SequentialBackend with 1 concurrent workers.
[Parallel(n_jobs=1)]: Done   1 out of   1 | elapsed:    4.6s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   1 out of   1 | elapsed:    4.6s finished
    Computing cross products for 305 → 1 MEG channel...
    Preparing the mapping matrix...
    Truncating at 85/305 components to omit less than 0.0001 (9.4e-05)

You can also use minimum-norm for EEG as well as MEG

evoked_interp_mne = evoked.copy().interpolate_bads(
    reset_bads=False, method=dict(eeg='MNE'), verbose=True)
evoked_interp_mne.plot(exclude=[], picks=('grad', 'eeg'))

Interpolating bad channels
    Automatic origin fit: head of radius 91.0 mm
    Computing dot products for 305 MEG channels...
[Parallel(n_jobs=1)]: Using backend SequentialBackend with 1 concurrent workers.
[Parallel(n_jobs=1)]: Done   1 out of   1 | elapsed:    4.7s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   1 out of   1 | elapsed:    4.7s finished
    Computing cross products for 305 → 1 MEG channel...
    Preparing the mapping matrix...
    Truncating at 85/305 components to omit less than 0.0001 (9.4e-05)
    Computing dot products for 59 EEG channels...
[Parallel(n_jobs=1)]: Using backend SequentialBackend with 1 concurrent workers.
[Parallel(n_jobs=1)]: Done   1 out of   1 | elapsed:    0.1s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   1 out of   1 | elapsed:    0.1s finished
    Computing cross products for 59 → 60 EEG channels...
    Preparing the mapping matrix...
    Truncating at 58/59 components and regularizing with α=1.0e-01
    The map has an average electrode reference (60 channels)

References

[1]

François M. Perrin, Jacques Pernier, Olivier M. Bertrand, and Jean Franćois Echallier. Spherical splines for scalp potential and current density mapping. Electroencephalography and Clinical Neurophysiology, 72(2):184–187, 1989. doi:10.1016/0013-4694(89)90180-6.

Estimated memory usage: 12 MB