def task_csd():
"""Step 05: Compute cross-spectral density (CSD) matrices"""
for subject in subjects:
epo_fname = fname.epo(subject=subject)
csd_fnames = [fname.csd(subject=subject, condition=cond)
for cond in conditions + ['baseline']]
yield dict(
name=subject,
task_dep=['epochs'],
file_dep=[epo_fname, '05_csd.py'],
targets=csd_fnames,
actions=['python 05_csd.py %s' % subject],
)
def task_figures():
"""Make all figures. Each figure is a sub-task."""
# Make figure 1: plot of the CSD matrices.
yield dict(
name='csd',
task_dep=['connectivity_stats'],
file_dep=[fname.epo(subject=subjects[0]),
fname.csd(subject=subjects[0], condition='face')],
targets=['../paper/figures/csd.pdf'],
actions=['python figure_csd.py'],
)
# Make figure 2: plot of the source space and forward model.
yield dict(
name='forward',
file_dep=[fname.fwd(subject=subjects[0]),
fname.fwd_r(subject=subjects[0]),
fname.trans(subject=subjects[0])],
targets=['../paper/figures/forward1.png',
'../paper/figures/forward2.png'],
actions=['python figure_forward.py'],
)
# Make figure 3: grand average power maps.
file_dep = [fname.ga_power_hemi(condition=cond, hemi='lh') for cond in conditions]
file_dep += [fname.ga_power_hemi(condition=cond, hemi='rh') for cond in conditions]
targets = ['../paper/figures/power_face_lh.png',
'../paper/figures/power_face_rh.png',
'../paper/figures/power_scrambled_lh.png',
'../paper/figures/power_scrambled_rh.png']
targets += ['../paper/figures/power_contrast_%s-%s-lh.png' % (freq[0], freq[1]) for freq in freq_bands]
yield dict(
name='power',
file_dep=file_dep,
targets=targets,
actions=['python figure_power.py'],
)
# Make figure 4: plot of the functional connectivity.
yield dict(
name='connectivity',
file_dep=[fname.ga_con(condition='pruned'),
fname.ga_con(condition='parcelled')],
targets=['../paper/figures/degree_lh.png',
'../paper/figures/degree_rh.png',
'../paper/figures/squircle.pdf'],
actions=['python figure_connectivity.py'],
)
def task_connectivity():
"""Step 10: Compute DICS connectivity."""
for subject in subjects:
fwd_r_fname = fname.fwd_r(subject=subject)
csd_fnames = []
con_fnames = []
for cond in conditions:
csd_fnames.append(fname.csd(subject=subject, condition=cond))
con_fnames.append(fname.con(subject=subject, condition=cond))
yield dict(
name=subject,
task_dep=['power'],
file_dep=[fwd_r_fname, fname.pairs, '10_connectivity.py'] + csd_fnames,
targets=con_fnames,
actions=['python 10_connectivity.py %s' % subject],
)
def task_power():
"""Step 09: Compute DICS power maps."""
for subject in subjects:
fwd_r_fname = fname.fwd_r(subject=subject)
csd_fnames = []
stc_fnames = []
for cond in conditions + ['baseline']:
csd_fnames.append(fname.csd(subject=subject, condition=cond))
stc_fnames.append(fname.power_hemi(subject=subject, condition=cond, hemi='lh'))
stc_fnames.append(fname.power_hemi(subject=subject, condition=cond, hemi='rh'))
yield dict(
name=subject,
task_dep=['select_vertices'],
file_dep=[fwd_r_fname, '09_power.py'] + csd_fnames,
targets=stc_fnames,
actions=['python 09_power.py %s' % subject],
)
for condition in conditions:
print('Condition:', condition)
# Remove the mean during the time interval for which we compute the CSD
epochs_baselined = epochs[condition].apply_baseline((csd_tmin, csd_tmax))
# Compute CSD for the desired time interval
csd = csd_morlet(epochs_baselined,
frequencies=frequencies,
tmin=csd_tmin,
tmax=csd_tmax,
decim=20,
n_jobs=n_jobs,
verbose=True)
# Save the CSD matrices
csd.save(fname.csd(condition=condition, subject=subject))
report.add_figs_to_section(csd.plot(show=False),
['CSD for %s' % condition],
section='Sensor-level')
# Also compute the CSD for the baseline period (use all epochs for this,
# regardless of condition). This way, we can compare the change in power caused
# by the presentation of the stimulus.
epochs = epochs.apply_baseline((-0.2, 0)) # Make sure data is zero-mean
csd_baseline = csd_morlet(epochs,
frequencies=frequencies,
tmin=-0.2,
tmax=0,
decim=20,
n_jobs=n_jobs,
verbose=True)
from matplotlib import pyplot as plt
import mne
from mne.time_frequency import read_csd, pick_channels_csd
from config import fname, subjects, freq_bands
info = mne.io.read_info(fname.epo(subject=subjects[0]))
grads = [info['ch_names'][ch] for ch in mne.pick_types(info, meg='grad')]
csd = read_csd(fname.csd(subject=subjects[0], condition='face'))
csd = pick_channels_csd(csd, grads)
csd = csd.mean([f[0] for f in freq_bands], [f[1] for f in freq_bands])
# Plot theta, alpha, low beta
csd[:3].plot(info, n_cols=3, show=False)
plt.savefig('../paper/figures/csd1.pdf', bbox_inches='tight')
# Plot high beta 1, high beta 2 and low gamma
csd[3:].plot(info, n_cols=3, show=False)
plt.savefig('../paper/figures/csd2.pdf', bbox_inches='tight')
# Read the forward model
fwd = mne.read_forward_solution(fname.fwd(subject=subject))
# Read the info structure
info = mne.io.read_info(fname.epo(subject=subject))
# Compute source power for all frequency bands and all conditions
fmin = [f[0] for f in freq_bands]
fmax = [f[1] for f in freq_bands]
csds = dict()
for condition in conditions + ['baseline']:
print('Reading CSD matrix for condition:', condition)
# Read the CSD matrix
csds[condition] = read_csd(fname.csd(condition=condition, subject=subject))
# Average the CSDs of both conditions
csd = csds['face'].copy()
csd._data += csds['scrambled']._data
csd._data /= 2
# Compute the DICS beamformer using the CSDs from both conditions, for all
# frequency bands.
filters = make_dics(info=info, forward=fwd, csd=csd.mean(fmin, fmax), reg=reg,
pick_ori='max-power')
stcs = dict()
for condition in conditions + ['baseline']:
print('Computing power for condition:', condition)
stcs[condition], _ = apply_dics_csd(csds[condition].mean(fmin, fmax),