def _load_raw(subject,
mne_data_root=None,
verbose=False,
onsets=None,
interpolate_bad_channels=False,
has_mastoid_channels=None, # None=True, False, or callable(subject) returning True/False
reference_mastoids=True):
if mne_data_root is None:
# use default data root
import deepthought
data_root = os.path.join(deepthought.DATA_PATH, 'OpenMIIR')
mne_data_root = os.path.join(data_root, 'eeg', 'mne')
mne_data_filepath = os.path.join(mne_data_root, '{}-raw.fif'.format(subject))
log.info('Loading raw data for subject "{}" from {}'.format(subject, mne_data_filepath))
raw = mne.io.Raw(mne_data_filepath, preload=True, verbose=verbose)
## referencing to mastoids
if has_mastoid_channels is None \
or has_mastoid_channels is True \
or has_mastoid_channels(subject) is True:
if reference_mastoids:
log.info('Referencing to mastoid channels: {}'.format(MASTOID_CHANNELS))
mne.io.set_eeg_reference(raw, MASTOID_CHANNELS, copy=False) # inplace
else:
log.info('This recording has unused mastoid channels: {} '
'To use them, re-run with reference_mastoids=True.'.format(MASTOID_CHANNELS))
raw.drop_channels(MASTOID_CHANNELS)
## optional event merging
if onsets == 'audio':
merge_trial_and_audio_onsets(raw,
use_audio_onsets=True,
inplace=True,
stim_channel='STI 014',
verbose=verbose)
elif onsets == 'trials':
merge_trial_and_audio_onsets(raw,
use_audio_onsets=True,
inplace=True,
stim_channel='STI 014',
verbose=verbose)
# else: keep both
bads = raw.info['bads']
if bads is not None and len(bads) > 0:
if interpolate_bad_channels:
log.info('Interpolating bad channels: {}'.format(bads))
raw.interpolate_bads()
else:
log.info('This file contains some EEG channels marked as bad: {}\n'
'To interpolate bad channels run load_raw() with interpolate_bad_channels=True.'
''.format(bads))
return raw
def merge_trial_and_audio_onsets(self, use_audio_onsets=True):
raw = self.raw
# save original events
self.orig_trial_events = self.trial_events
# merge
merge_trial_and_audio_onsets(raw, use_audio_onsets=use_audio_onsets, inplace=True)
# recompute trial_events and times
trial_events = mne.find_events(raw, stim_channel='STI 014', shortest_event=0)
trial_event_times = raw.index_as_time(trial_events[:,0])
self.trial_events = trial_events
self.trial_event_times = trial_event_times
def merge_trial_and_audio_onsets(self, use_audio_onsets=True):
raw = self.raw
# save original events
self.orig_trial_events = self.trial_events
# merge
merge_trial_and_audio_onsets(raw, use_audio_onsets=use_audio_onsets, inplace=True)
# recompute trial_events and times
trial_events = mne.find_events(raw, stim_channel='STI 014', shortest_event=0)
trial_event_times = raw.times[trial_events[:,0]]
self.trial_events = trial_events
self.trial_event_times = trial_event_times
def check_trial_audio_onset_merge(self, use_audio_onsets=True, verbose=None):
# assert self.filtered is False
assert self.downsampled is False
raw = self.raw
## check whether trial and audio events are merged correctly
merged_events = merge_trial_and_audio_onsets(raw, use_audio_onsets=use_audio_onsets, inplace=False)
if verbose:
for event in merged_events:
print event
plt.figure(figsize=(17,10))
axes = plt.gca()
mne.viz.plot_events(merged_events, raw.info['sfreq'], raw.first_samp, axes=axes)
def check_trial_audio_onset_merge(self, use_audio_onsets=True, verbose=None):
# assert self.filtered is False
assert self.downsampled is False
raw = self.raw
## check whether trial and audio events are merged correctly
merged_events = merge_trial_and_audio_onsets(raw, use_audio_onsets=use_audio_onsets, inplace=False)
if verbose:
for event in merged_events:
print event
plt.figure(figsize=(17,10))
axes = plt.gca()
mne.viz.plot_events(merged_events, raw.info['sfreq'], raw.first_samp, axes=axes)
def _load_raw(
subject,
mne_data_root=None,
verbose=False,
onsets=None,
interpolate_bad_channels=False,
has_mastoid_channels=None, # None=True, False, or callable(subject) returning True/False
apply_reference=True, # by default, reference the data
reference_mastoids=True):
if mne_data_root is None:
# use default data root
import deepthought
data_root = os.path.join(deepthought.DATA_PATH, 'OpenMIIR')
mne_data_root = os.path.join(data_root, 'eeg', 'mne')
mne_data_filepath = os.path.join(mne_data_root,
'{}-raw.fif'.format(subject))
log.info('Loading raw data for subject "{}" from {}'.format(
subject, mne_data_filepath))
raw = mne.io.Raw(mne_data_filepath, preload=True, verbose=verbose)
if apply_reference:
if has_mastoid_channels is None \
or has_mastoid_channels is True \
or has_mastoid_channels(subject) is True:
## referencing to mastoids
if reference_mastoids:
log.info('Referencing to mastoid channels: {}'.format(
MASTOID_CHANNELS))
mne.io.set_eeg_reference(raw, MASTOID_CHANNELS,
copy=False) # inplace
else:
log.info(
'This recording has unused mastoid channels: {} '
'To use them, re-run with reference_mastoids=True.'.format(
MASTOID_CHANNELS))
raw.drop_channels(MASTOID_CHANNELS)
else:
## referencing to average
log.info('Referencing to average.')
mne.io.set_eeg_reference(raw, copy=False)
## optional event merging
if onsets == 'audio':
merge_trial_and_audio_onsets(raw,
use_audio_onsets=True,
inplace=True,
stim_channel='STI 014',
verbose=verbose)
elif onsets == 'trials':
merge_trial_and_audio_onsets(raw,
use_audio_onsets=True,
inplace=True,
stim_channel='STI 014',
verbose=verbose)
# else: keep both
bads = raw.info['bads']
if bads is not None and len(bads) > 0:
if interpolate_bad_channels:
log.info('Interpolating bad channels: {}'.format(bads))
raw.interpolate_bads()
else:
log.info(
'This file contains some EEG channels marked as bad: {}\n'
'To interpolate bad channels run load_raw() with interpolate_bad_channels=True.'
''.format(bads))
return raw
