Maxwell filter data with movement compensation

Demonstrate movement compensation on simulated data. The simulated data contains bilateral activation of auditory cortices, repeated over 14 different head rotations (head center held fixed). See the following for details:

# Authors: Eric Larson <larson.eric.d@gmail.com>
#
# License: BSD (3-clause)

from os import path as op

import mne
from mne.preprocessing import maxwell_filter

print(__doc__)

data_path = op.join(mne.datasets.misc.data_path(verbose=True), 'movement')

head_pos = mne.chpi.read_head_pos(op.join(data_path, 'simulated_quats.pos'))
raw = mne.io.read_raw_fif(op.join(data_path, 'simulated_movement_raw.fif'))
raw_stat = mne.io.read_raw_fif(op.join(data_path,
                                       'simulated_stationary_raw.fif'))

Out:

Opening raw data file /home/circleci/mne_data/MNE-misc-data/movement/simulated_movement_raw.fif...
    Range : 25800 ... 34208 =     42.956 ...    56.955 secs
Ready.
Current compensation grade : 0
Opening raw data file /home/circleci/mne_data/MNE-misc-data/movement/simulated_stationary_raw.fif...
    Range : 25800 ... 34208 =     42.956 ...    56.955 secs
Ready.
Current compensation grade : 0

Visualize the “subject” head movements. By providing the measurement information, the distance to the nearest sensor in each direction (e.g., left/right for the X direction, forward/backward for Y) can be shown in blue, and the destination (if given) shown in red.

mne.viz.plot_head_positions(
    head_pos, mode='traces', destination=raw.info['dev_head_t'], info=raw.info)
../../_images/sphx_glr_plot_movement_compensation_001.png

This can also be visualized using a quiver.

mne.viz.plot_head_positions(
    head_pos, mode='field', destination=raw.info['dev_head_t'], info=raw.info)
../../_images/sphx_glr_plot_movement_compensation_002.png

Out:

Getting helmet for system 306m

Process our simulated raw data (taking into account head movements).

# extract our resulting events
events = mne.find_events(raw, stim_channel='STI 014')
events[:, 2] = 1
raw.plot(events=events)

topo_kwargs = dict(times=[0, 0.1, 0.2], ch_type='mag', vmin=-500, vmax=500,
                   time_unit='s')
../../_images/sphx_glr_plot_movement_compensation_003.png

Out:

14 events found
Event IDs: [ 1  2  3  4  5  6  7  8  9 10 11 12 13 14]

First, take the average of stationary data (bilateral auditory patterns).

evoked_stat = mne.Epochs(raw_stat, events, 1, -0.2, 0.8).average()
evoked_stat.plot_topomap(title='Stationary', **topo_kwargs)
../../_images/sphx_glr_plot_movement_compensation_004.png

Out:

14 matching events found
Applying baseline correction (mode: mean)
Not setting metadata
0 projection items activated

Second, take a naive average, which averages across epochs that have been simulated to have different head positions and orientations, thereby spatially smearing the activity.

evoked = mne.Epochs(raw, events, 1, -0.2, 0.8).average()
evoked.plot_topomap(title='Moving: naive average', **topo_kwargs)
../../_images/sphx_glr_plot_movement_compensation_005.png

Out:

14 matching events found
Applying baseline correction (mode: mean)
Not setting metadata
0 projection items activated

Third, use raw movement compensation (restores pattern).

raw_sss = maxwell_filter(raw, head_pos=head_pos)
evoked_raw_mc = mne.Epochs(raw_sss, events, 1, -0.2, 0.8).average()
evoked_raw_mc.plot_topomap(title='Moving: movement compensated', **topo_kwargs)
../../_images/sphx_glr_plot_movement_compensation_006.png

Out:

Maxwell filtering raw data
    Appending head position result channels and loading raw data from disk
    No bad MEG channels
    Processing 203 gradiometers and 102 magnetometers
    Automatic origin fit: head of radius 91.0 mm
    Using origin -4.1, 16.0, 51.7 mm in the head frame
        Using 90/95 harmonic components for    0.000  (75/80 in, 15/15 out)
    Processing 1 data chunks of (at least) 10.0 sec
        Using 87/95 harmonic components for    0.000  (72/80 in, 15/15 out)
        Using 88/95 harmonic components for    1.001  (73/80 in, 15/15 out)
        Using 90/95 harmonic components for    2.000  (75/80 in, 15/15 out)
        Using 88/95 harmonic components for    3.000  (73/80 in, 15/15 out)
        Using 88/95 harmonic components for    3.999  (73/80 in, 15/15 out)
        Using 88/95 harmonic components for    5.000  (73/80 in, 15/15 out)
        Using 89/95 harmonic components for    6.001  (74/80 in, 15/15 out)
        Using 93/95 harmonic components for    6.999  (78/80 in, 15/15 out)
        Using 88/95 harmonic components for    8.000  (73/80 in, 15/15 out)
        Using 91/95 harmonic components for    9.001  (76/80 in, 15/15 out)
        Using 93/95 harmonic components for   10.000  (78/80 in, 15/15 out)
        Using 93/95 harmonic components for   11.000  (78/80 in, 15/15 out)
        Using 89/95 harmonic components for   11.999  (74/80 in, 15/15 out)
        Using 88/95 harmonic components for   13.000  (73/80 in, 15/15 out)
        Used  14 head positions for    0.000 -   13.999 sec (#1/1)
[done]
14 matching events found
Applying baseline correction (mode: mean)
Not setting metadata
0 projection items activated

Total running time of the script: ( 0 minutes 7.248 seconds)

Gallery generated by Sphinx-Gallery