Compute Rap-Music on evoked data

Compute a Recursively Applied and Projected MUltiple Signal Classification (RAP-MUSIC) [1] on evoked data.

# Author: Yousra Bekhti <yousra.bekhti@gmail.com>
#
# License: BSD-3-Clause
# Copyright the MNE-Python contributors.

import mne
from mne.beamformer import rap_music
from mne.datasets import sample
from mne.viz import plot_dipole_amplitudes, plot_dipole_locations

print(__doc__)

data_path = sample.data_path()
subjects_dir = data_path / "subjects"
meg_path = data_path / "MEG" / "sample"
fwd_fname = meg_path / "sample_audvis-meg-eeg-oct-6-fwd.fif"
evoked_fname = meg_path / "sample_audvis-ave.fif"
cov_fname = meg_path / "sample_audvis-cov.fif"

# Read the evoked response and crop it
condition = "Right Auditory"
evoked = mne.read_evokeds(evoked_fname, condition=condition, baseline=(None, 0))
# select N100
evoked.crop(tmin=0.05, tmax=0.15)

evoked.pick(picks="meg", exclude="bads")

# Read the forward solution
forward = mne.read_forward_solution(fwd_fname)

# Read noise covariance matrix
noise_cov = mne.read_cov(cov_fname)

dipoles, residual = rap_music(
    evoked, forward, noise_cov, n_dipoles=2, return_residual=True, verbose=True
)
trans = forward["mri_head_t"]
plot_dipole_locations(dipoles, trans, "sample", subjects_dir=subjects_dir)
plot_dipole_amplitudes(dipoles)

# Plot the evoked data and the residual.
evoked.plot(ylim=dict(grad=[-300, 300], mag=[-800, 800], eeg=[-6, 8]), time_unit="s")
residual.plot(ylim=dict(grad=[-300, 300], mag=[-800, 800], eeg=[-6, 8]), time_unit="s")

References

[1]

John C. Mosher and Richard M. Leahy. Source localization using recursively applied and projected (RAP) MUSIC. IEEE Transactions on Signal Processing, 47(2):332–340, 1999. doi:10.1109/78.740118.

Estimated memory usage: 0 MB