montage = mne.channels.read_montage('biosemi128')
raw.set_montage(montage)
# mne.viz.plot_sensors(raw.info)
# Bad channels
bads = bad_channs[subj]
raw.info['bads'] = bads
picks_eeg = mne.pick_types(raw.info, eeg=True, meg=False, eog=False, exclude='bads')
# raw.plot(n_channels=128)
# Log
log = read_log_file(log_filename)
# Events
events = mne.find_events(raw)
events = check_events(events)
durations = durations_from_log(log)
new_events = create_events(events, durations, fs)
raw.add_events(new_events)
events_updated = mne.find_events(raw, shortest_event=1)
# mne.viz.plot_events(events_updated)
# Reject
reject = {'eeg': 120e-6}
if do_ica:
n_components = 20
method = 'fastica'
decim = 3
ica = mne.preprocessing.ICA(n_components=n_components, method=method)
ica.fit(raw, picks=picks_eeg, decim=decim, reject=reject)
53,
54,
55
]
# Filters
notchs = np.arange(50, 151, 50)
raw.filter(l_freq=1, h_freq=140, n_jobs=n_jobs)
raw.notch_filter(freqs=notchs, n_jobs=n_jobs)
# Log
log_filename = log_path + 'ETG{}a_{}.mat'.format(ratios, subj)
log = read_log_file(log_filename)
events = mne.find_events(raw)
events = check_events(events)
durations = durations_from_log_ieeg(log)
if ratios == 3:
durations = [d for d in durations if ~np.isnan(d)]
log = log[:-11]
new_events = create_events(events, durations, raw.info['sfreq'])
raw.add_events(new_events)
events_updated = mne.find_events(raw, shortest_event=1)
events_updated = check_events(events_updated)
events_updated, log = add_event_condition(events_updated, log)
events_tr_id = add_event_tr_id(events_updated)
# Epoch
t_min = -0.7
t_max = 0.7
def run_prepro(subj):
ses_x_subj = sessions[subj]
for ses in ses_x_subj:
archivo = 'etg_su{}_se{}'.format(subj, ses)
full_path = op.join(data_path, archivo + '.bdf')
if not op.isfile(full_path):
print('The session {} of subject {} is missing'.format(ses, subj))
continue
print('Processing Subject {} - Session {}'.format(subj, ses))
raw = mne.io.read_raw_edf(full_path, preload=True)
# raw.plot(n_channels=32, duration=2)
montage = mne.channels.read_montage('biosemi128')
head_channs = montage.ch_names[:128]
head_channs.append('STI 014')
raw.pick_channels(ch_names=head_channs)
raw.set_montage(montage)
bads = bad_channs[subj]
raw.info['bads'] = bads
picks_eeg = mne.pick_types(raw.info,
eeg=True,
meg=False,
eog=False,
exclude='bads')
raw, ref = mne.io.set_eeg_reference(raw, ref_channels=None)
raw.apply_proj()
raw.filter(l_freq=0.1,
h_freq=40,
filter_length='auto',
l_trans_bandwidth='auto',
h_trans_bandwidth='auto')
raw.plot(n_channels=32, duration=5)
# ICA
reject = {'eeg': 120e-6}
n_components = 20
method = 'infomax'
decim = 4
ica = mne.preprocessing.ICA(n_components=n_components, method=method)
ica.fit(raw, picks=picks_eeg, decim=decim, reject=reject)
print(ica)
ica.plot_components()
comp_to_delete = [
int(x)
for x in input('Components to delete (ej: 1 2 3): ').split()
]
if comp_to_delete:
ica.exclude = comp_to_delete
raw = ica.apply(raw)
raw.interpolate_bads()
# Log
log_filename = op.join(log_path,
archivo[:4] + 'scalp_' + archivo[4:] + '.mat')
log = read_log_file(log_filename)
# Events
events = mne.find_events(raw)
events = check_events(events)
durations = durations_from_log(log)
new_events = create_events(events, durations, raw.info['sfreq'])
raw.add_events(new_events)
events_updated = mne.find_events(raw, shortest_event=1)
events_updated = check_events(events_updated)
# events_updated = add_event_condition(events_updated, log)
# mne.viz.plot_events(events_updated)
# Epoch
t_min = -0.4
t_max = 4
baseline = (None, 0)
event_id = dict(s2=2)
# event_id = marks
epoch = mne.Epochs(raw,
events_updated,
event_id=event_id,
tmin=t_min,
tmax=t_max,
baseline=baseline,
reject=reject,
preload=True)
# epoch.plot(n_epochs=10)
# Downsample
epoch.resample(256, n_jobs=n_jobs)
epoch.save(op.join(study_path, 'fif/',
archivo + 'avg_subcond-epo.fif'))
print('\n' * 4)
def run_prepro(subj):
ses_x_subj = sessions[subj]
montage = mne.channels.read_montage('biosemi128')
head_channs = montage.ch_names[:128]
head_channs.append('STI 014')
all_ses = list()
all_log = list()
for ses in ses_x_subj:
# EEG
file = 'etg_su{}_se{}' .format(subj, ses)
full_path = op.join(data_path, file + '.bdf')
if not op.isfile(full_path):
print('The session {} of subject {} is missing' .format(ses, subj))
continue
print('Processing Subject {} - Session {}' .format(subj, ses))
raw = mne.io.read_raw_edf(full_path, preload=True)
# LOG
log_filename = op.join(log_path, file[:4] + 'scalp_' + file[4:] + '.mat')
log = read_log_file(log_filename)
log['session'] = ses
log['subject'] = subj
all_log.append(log)
# Events
events = mne.find_events(raw)
events = check_events(events)
durations = durations_from_log(log)
new_events = create_events(events, durations, raw.info['sfreq'])
raw.add_events(new_events)
raw.pick_channels(ch_names=head_channs)
all_ses.append(raw)
if len(all_ses) > 1:
raw_cat = mne.concatenate_raws(all_ses, preload=True)
logs = pd.concat(all_log, ignore_index=True)
else:
raw_cat = raw
logs = log
del all_ses
del raw
events = mne.find_events(raw_cat)
events = check_events(events)
# mne.viz.plot_events(events)
try:
check_events_and_log(events, logs)
except ValueError as err:
print(err, ' - subj:', subj)
return
raw_cat.set_montage(montage)
bads = bad_channs[subj]
raw_cat.info['bads'] = bads
picks_eeg = mne.pick_types(raw_cat.info, eeg=True, meg=False, eog=False, exclude='bads')
raw, ref = mne.io.set_eeg_reference(raw_cat, ref_channels=None)
raw.apply_proj()
raw.filter(l_freq=0.1, h_freq=40, filter_length='auto', l_trans_bandwidth='auto', h_trans_bandwidth='auto', n_jobs=n_jobs)
#raw.plot(events, n_channels=128, duration=20, scalings={'eeg': 40e-6})
# ICA
reject = {'eeg': 250e-6}
n_components = 20
method = 'extended-infomax'
decim = 4
ica = mne.preprocessing.ICA(n_components=n_components, method=method)
ica.fit(raw_cat, picks=picks_eeg, decim=decim, reject=reject)
print(ica)
ica.plot_components()
comp_to_delete = [int(x) for x in input('Components to delete (ej: 1 2 3): ').split()]
if comp_to_delete:
ica.plot_properties(raw_cat, picks=comp_to_delete)
comp_to_delete = [int(x) for x in input('Confirm components to delete (ej: 1 2 3): ').split()]
if comp_to_delete:
ica.exclude = comp_to_delete
raw_cat = ica.apply(raw_cat)
raw_cat.interpolate_bads()
raw_cat.save(op.join(study_path, 'fif', 'v2', '{}_prep-raw.fif' .format(subj)), overwrite=True)
logs.to_csv(op.join(study_path, 'logs', '{}_prep-raw.csv' .format(subj)))
del raw_cat
print('\n' * 4)