def make_bipolar(data_fname, montage_filename):
raw = mne.io.read_raw_edf(data_fname, preload=False, verbose=False)
mne.rename_channels(raw.info, lambda name: re.sub(r'(POL|SEEG)\s+', '', name).strip())
channel_types = dict()
for ch in raw.ch_names:
result = re.match(r'^[A-Z][\']?\d+', ch)
if result:
channel_types[ch] = 'seeg'
raw.set_channel_types(channel_types)
montage = pd.read_csv(montage_filename, delimiter='\t')
montage.drop_duplicates(subset='name', inplace=True)
anode,cathode = clean_montage(raw.ch_names, montage.anode.tolist(), montage.cathode.tolist())
raw.load_data()
bipo = mne.set_bipolar_reference(raw, list(anode), list(cathode), copy=True, verbose=False)
bipo = drop_monopolar_channels(bipo)
bipo.drop_channels(bipo.info['bads'])
picks_seeg = mne.pick_types(bipo.info, meg=False, seeg=True)
non_seeg_chans = [ch_name for ch_idx, ch_name in enumerate(bipo.ch_names) if not(ch_idx in picks_seeg) or len(ch_name.split('-')) == 1]
bipo.drop_channels(non_seeg_chans)
bipo.notch_filter(np.arange(50, bipo.info['sfreq']//2, 50))
return bipo
def for_firas_37(crop_start=None, file_name=None):
edf = 'C:\\Matlab\\D037_28_01_21a.edf'
raw = mne.io.read_raw_edf(edf)
channels_list_original = [
'RA 01', 'RA 02', 'RH 01', 'RH 02', 'LAH 01', 'LAH 02'
]
channels_list_bipolar_original = [
x[:-1] + str(int(x[-1]) + 1) for x in channels_list_original
]
channels_list = [x.replace(' 0', '') for x in channels_list_original]
channels_list_bipolar = [
x.replace(' 0', '') for x in channels_list_bipolar_original
]
raw.pick_channels(
list(
set(channels_list_original) | set(channels_list_bipolar_original)))
channels_mapping = {x: x.replace(' 0', '') for x in raw.ch_names}
raw.rename_channels(channels_mapping)
if crop_start is not None:
raw.crop(tmin=crop_start * 60, tmax=crop_start * 60 + 16 * 60)
raw.resample(250)
raw.load_data()
raw_bi = mne.set_bipolar_reference(raw,
anode=channels_list,
cathode=channels_list_bipolar,
drop_refs=False)
rotem_write_edf(raw_bi,
'416_bipolar.edf' if file_name is None else file_name)
def create_bipolar(raw, electrodes):
""" create EEG data in a bipolar montage
Parameters
----------
raw : MNE Raw
EEG data
electrodes : list of strings
names of electrodes
Returns
-------
bipolar : MNE Raw
EEG data in a bipolar montage
"""
anodes = list()
cathodes = list()
ch_names = list()
for name in electrodes:
temp = setup_bipolar(name, raw.ch_names, raw.info['bads'])
anodes.extend(temp[0])
cathodes.extend(temp[1])
ch_names.extend(temp[2])
bipolar = mne.set_bipolar_reference(raw, anodes, cathodes,
ch_names, verbose=False)
bipolar = bipolar.pick_channels(ch_names)
bipolar = bipolar.reorder_channels(ch_names)
return bipolar
def test_bipolar_combinations():
"""Test bipolar channel generation."""
ch_names = ['CH' + str(ni + 1) for ni in range(10)]
info = create_info(ch_names=ch_names,
sfreq=1000.,
ch_types=['eeg'] * len(ch_names))
raw_data = np.random.randn(len(ch_names), 1000)
raw = RawArray(raw_data, info)
def _check_bipolar(raw_test, ch_a, ch_b):
picks = [raw_test.ch_names.index(ch_a + '-' + ch_b)]
get_data_res = raw_test.get_data(picks=picks)[0, :]
manual_a = raw_data[ch_names.index(ch_a), :]
manual_b = raw_data[ch_names.index(ch_b), :]
assert_array_equal(get_data_res, manual_a - manual_b)
# test classic EOG/ECG bipolar reference (only two channels per pair).
raw_test = set_bipolar_reference(raw, ['CH2'], ['CH1'], copy=True)
_check_bipolar(raw_test, 'CH2', 'CH1')
# test all combinations.
a_channels, b_channels = zip(*itertools.combinations(ch_names, 2))
a_channels, b_channels = list(a_channels), list(b_channels)
raw_test = set_bipolar_reference(raw, a_channels, b_channels, copy=True)
for ch_a, ch_b in zip(a_channels, b_channels):
_check_bipolar(raw_test, ch_a, ch_b)
# check if reference channels have been dropped.
assert (len(raw_test.ch_names) == len(a_channels))
raw_test = set_bipolar_reference(raw,
a_channels,
b_channels,
drop_refs=False,
copy=True)
# check if reference channels have been kept correctly.
assert (len(raw_test.ch_names) == len(a_channels) + len(ch_names))
for idx, ch_label in enumerate(ch_names):
manual_ch = raw_data[idx, :]
assert_array_equal(
raw_test._data[raw_test.ch_names.index(ch_label), :], manual_ch)
# test bipolars with a channel in both list (anode & cathode).
raw_test = set_bipolar_reference(raw, ['CH2', 'CH1'], ['CH1', 'CH2'],
copy=True)
_check_bipolar(raw_test, 'CH2', 'CH1')
_check_bipolar(raw_test, 'CH1', 'CH2')
def add_bipolar_derivation(raw, ch_1, ch_2):
"""
adds a channel to the given raw instance that is ch_1-ch_2
:param raw: raw object to derive channel from
:param ch_1: anode
:param ch_2: cathode
"""
raw = mne.set_bipolar_reference(raw, ch_1, ch_2, drop_refs=False)
return raw
def test_bipolar_combinations():
"""Test bipolar channel generation."""
ch_names = ['CH' + str(ni + 1) for ni in range(10)]
info = create_info(
ch_names=ch_names, sfreq=1000., ch_types=['eeg'] * len(ch_names))
raw_data = np.random.randn(len(ch_names), 1000)
raw = RawArray(raw_data, info)
def _check_bipolar(raw_test, ch_a, ch_b):
picks = [raw_test.ch_names.index(ch_a + '-' + ch_b)]
get_data_res = raw_test.get_data(picks=picks)[0, :]
manual_a = raw_data[ch_names.index(ch_a), :]
manual_b = raw_data[ch_names.index(ch_b), :]
assert_array_equal(get_data_res, manual_a - manual_b)
# test classic EOG/ECG bipolar reference (only two channels per pair).
raw_test = set_bipolar_reference(raw, ['CH2'], ['CH1'], copy=True)
_check_bipolar(raw_test, 'CH2', 'CH1')
# test all combinations.
a_channels, b_channels = zip(*itertools.combinations(ch_names, 2))
a_channels, b_channels = list(a_channels), list(b_channels)
raw_test = set_bipolar_reference(raw, a_channels, b_channels, copy=True)
for ch_a, ch_b in zip(a_channels, b_channels):
_check_bipolar(raw_test, ch_a, ch_b)
# check if reference channels have been dropped.
assert (len(raw_test.ch_names) == len(a_channels))
raw_test = set_bipolar_reference(
raw, a_channels, b_channels, drop_refs=False, copy=True)
# check if reference channels have been kept correctly.
assert (len(raw_test.ch_names) == len(a_channels) + len(ch_names))
for idx, ch_label in enumerate(ch_names):
manual_ch = raw_data[idx, :]
assert_array_equal(
raw_test._data[raw_test.ch_names.index(ch_label), :], manual_ch)
# test bipolars with a channel in both list (anode & cathode).
raw_test = set_bipolar_reference(
raw, ['CH2', 'CH1'], ['CH1', 'CH2'], copy=True)
_check_bipolar(raw_test, 'CH2', 'CH1')
_check_bipolar(raw_test, 'CH1', 'CH2')
def write_events_fif_bipolar():
edf = '/Users/rotemfalach/projects/epileptic_activity/results/406/406_overnightData.edf'
channels_list = [
'SEEG LA1-REF', 'SEEG LA2-REF', 'SEEG LAH1-REF', 'SEEG LAH2-REF',
'SEEG RA1-REF', 'SEEG RA2-REF', 'SEEG RAH1-REF', 'SEEG RAH2-REF'
]
channels_list_bipolar = [
x[:-5] + str(int(x[-5]) + 1) + '-REF' for x in channels_list
]
# channels_list = ['LA 02', 'LH 01', 'LH 02', 'RA 02', 'RMH 01', 'RMH 02'] # 036
# channels_list_bipolar = [x[:-1] + str(int(x[-1]) + 1) for x in channels_list]
spikes_files = '_'.join(edf.split('_')[0:2])
original_raw = mne.io.read_raw_edf(edf)
original_raw.pick_channels(
list(set(channels_list) | set(channels_list_bipolar)))
original_raw.load_data()
raw = original_raw
for chan in channels_list:
# original_raw.rename_channels({chan: chan.replace('SEEG ', '').replace('-REF', '')})
original_raw = mne.set_bipolar_reference(
original_raw,
anode=[chan],
cathode=chan[:-5] + str(int(chan[-5]) + 1) + '-REF',
drop_refs=False,
ch_name=chan + '_bi')
stim_amp, stim_grad, stim_env = np.zeros((1, raw.n_times)), np.zeros(
(1, raw.n_times)), np.zeros((1, raw.n_times))
# chan_alias = chan.replace(' 0', '')
chan_alias = chan.replace('SEEG ', '').replace('-REF', '')
spikes_df = pd.read_csv(spikes_files + '_' + chan_alias +
'_biplor_spikes.csv')
for evt in spikes_df.values:
if evt[1] == 'amp':
stim_amp[0][evt[4]] = 1
elif evt[1] == 'grad':
stim_grad[0][evt[4]] = 1
else:
stim_env[0][evt[4]] = 1
info_amp = mne.create_info([chan_alias + '_AMP_bi'], raw.info['sfreq'],
['stim'])
info_grad = mne.create_info([chan_alias + '_GRAD_bi'],
raw.info['sfreq'], ['stim'])
info_env = mne.create_info([chan_alias + '_ENV_bi'], raw.info['sfreq'],
['stim'])
amp_raw = mne.io.RawArray(stim_amp, info_amp)
grad_raw = mne.io.RawArray(stim_grad, info_grad)
env_raw = mne.io.RawArray(stim_env, info_env)
original_raw.add_channels([amp_raw, grad_raw, env_raw],
force_update_info=True)
original_raw.drop_channels(
list(set(channels_list) | set(channels_list_bipolar)))
original_raw.resample(500).save('406_bi.fif')
def bipolar_ref(raw, anode, cathode, new_ch=None, copy=False, ch_info=None):
"""
Sets bipolar montage.
"""
from mne import set_bipolar_reference
cathode = utilsfunc.in_list(cathode)
if len(cathode) == 1:
cathode = cathode * len(utilsfunc.in_list(anode))
if new_ch is None:
new_ch = utilsfunc.in_list(anode)
if copy is False:
set_bipolar_reference(raw,anode=anode,\
cathode=cathode,ch_name=new_ch,\
copy=False, ch_info=ch_info)
else:
return set_bipolar_reference(raw,anode=anode,\
cathode=cathode,ch_name=new_ch,\
copy=True, ch_info=ch_info)
def for_firas_398(crop_start=None, file_name=None):
edf = 'C:\\UCLA\\P398_overnightData.edf'
raw = mne.io.read_raw_edf(edf)
channels_list_original = [
'SEEG R'
'A1-REF', 'SEEG RA2-REF', 'SEEG REC1-REF', 'SEEG REC2-REF',
'SEEG RAH1-REF', 'SEEG RAH2-REF', 'SEEG LA1-REF', 'SEEG LA2-REF',
'SEEG LAH1-REF', 'SEEG LAH2-REF'
]
channels_list_bipolar_original = [
x[:-5] + str(int(x[-5]) + 1) + '-REF' for x in channels_list_original
]
channels_list = [
x.replace('SEEG ', '').replace('-REF', '')
for x in channels_list_original
]
channels_list_bipolar = [
x.replace('SEEG ', '').replace('-REF', '')
for x in channels_list_bipolar_original
]
raw.pick_channels(
list(
set(channels_list_original) | set(channels_list_bipolar_original)))
channels_mapping = {
x: x.replace('SEEG ', '').replace('-REF', '')
for x in raw.ch_names
}
raw.rename_channels(channels_mapping)
if crop_start is not None:
raw.crop(tmin=crop_start * 60, tmax=crop_start * 60 + 16 * 60)
raw.resample(250)
raw.load_data()
raw_bi = mne.set_bipolar_reference(raw,
anode=channels_list,
cathode=channels_list_bipolar,
drop_refs=False)
# raw_bi.plot()
write_edf('398_bipolar.edf' if file_name is None else file_name, raw_bi)
def test_set_bipolar_reference(inst_type):
"""Test bipolar referencing."""
raw = read_raw_fif(fif_fname, preload=True)
raw.apply_proj()
if inst_type == 'raw':
inst = raw
del raw
elif inst_type in ['epochs', 'evoked']:
events = find_events(raw, stim_channel='STI 014')
epochs = Epochs(raw, events, tmin=-0.3, tmax=0.7, preload=True)
inst = epochs
if inst_type == 'evoked':
inst = epochs.average()
del epochs
ch_info = {'kind': FIFF.FIFFV_EOG_CH, 'extra': 'some extra value'}
with pytest.raises(KeyError, match='key errantly present'):
set_bipolar_reference(inst, 'EEG 001', 'EEG 002', 'bipolar', ch_info)
ch_info.pop('extra')
reref = set_bipolar_reference(inst, 'EEG 001', 'EEG 002', 'bipolar',
ch_info)
assert (reref.info['custom_ref_applied'])
# Compare result to a manual calculation
a = inst.copy().pick_channels(['EEG 001', 'EEG 002'])
a = a._data[..., 0, :] - a._data[..., 1, :]
b = reref.copy().pick_channels(['bipolar'])._data[..., 0, :]
assert_allclose(a, b)
# Original channels should be replaced by a virtual one
assert ('EEG 001' not in reref.ch_names)
assert ('EEG 002' not in reref.ch_names)
assert ('bipolar' in reref.ch_names)
# Check channel information
bp_info = reref.info['chs'][reref.ch_names.index('bipolar')]
an_info = reref.info['chs'][inst.ch_names.index('EEG 001')]
for key in bp_info:
if key == 'loc':
assert_array_equal(bp_info[key], 0)
elif key == 'coil_type':
assert_equal(bp_info[key], FIFF.FIFFV_COIL_EEG_BIPOLAR)
elif key == 'kind':
assert_equal(bp_info[key], FIFF.FIFFV_EOG_CH)
else:
assert_equal(bp_info[key], an_info[key])
# Minimalist call
reref = set_bipolar_reference(inst, 'EEG 001', 'EEG 002')
assert ('EEG 001-EEG 002' in reref.ch_names)
# Minimalist call with twice the same anode
reref = set_bipolar_reference(inst, ['EEG 001', 'EEG 001', 'EEG 002'],
['EEG 002', 'EEG 003', 'EEG 003'])
assert ('EEG 001-EEG 002' in reref.ch_names)
assert ('EEG 001-EEG 003' in reref.ch_names)
# Set multiple references at once
reref = set_bipolar_reference(
inst,
['EEG 001', 'EEG 003'],
['EEG 002', 'EEG 004'],
['bipolar1', 'bipolar2'],
[{
'kind': FIFF.FIFFV_EOG_CH
}, {
'kind': FIFF.FIFFV_EOG_CH
}],
)
a = inst.copy().pick_channels(['EEG 001', 'EEG 002', 'EEG 003', 'EEG 004'])
a = np.concatenate([
a._data[..., :1, :] - a._data[..., 1:2, :],
a._data[..., 2:3, :] - a._data[..., 3:4, :]
],
axis=-2)
b = reref.copy().pick_channels(['bipolar1', 'bipolar2'])._data
assert_allclose(a, b)
# Test creating a bipolar reference that doesn't involve EEG channels:
# it should not set the custom_ref_applied flag
reref = set_bipolar_reference(inst,
'MEG 0111',
'MEG 0112',
ch_info={'kind': FIFF.FIFFV_MEG_CH},
verbose='error')
assert (not reref.info['custom_ref_applied'])
assert ('MEG 0111-MEG 0112' in reref.ch_names)
# Test a battery of invalid inputs
pytest.raises(ValueError, set_bipolar_reference, inst, 'EEG 001',
['EEG 002', 'EEG 003'], 'bipolar')
pytest.raises(ValueError, set_bipolar_reference, inst,
['EEG 001', 'EEG 002'], 'EEG 003', 'bipolar')
pytest.raises(ValueError, set_bipolar_reference, inst, 'EEG 001',
'EEG 002', ['bipolar1', 'bipolar2'])
pytest.raises(ValueError,
set_bipolar_reference,
inst,
'EEG 001',
'EEG 002',
'bipolar',
ch_info=[{
'foo': 'bar'
}, {
'foo': 'bar'
}])
pytest.raises(ValueError,
set_bipolar_reference,
inst,
'EEG 001',
'EEG 002',
ch_name='EEG 003')
# edf = 'C:\\UCLA\\P406_staging_PSG_and_intracranial_Mref_correct.txt.edf'
# edf = 'C:\\Users\\user\\PycharmProjects\\pythonProject\\results\\402\\402_for_tag.edf'
# edf = "C:ֿֿֿֿ\\analysis\\396\\396_LAH1_100_samples.fif"
subj = '406'
edf = f'C:\\UCLA\\P{subj}_overnightData.edf'
depth_epochs = pd.read_csv(
f'C:\\repos\\spikes_notebooks\\spike_index_{subj}.csv')
scalp_epochs = pd.read_csv(
f'C:\\repos\\spikes_notebooks\\scalp_index_{subj}.csv')
raw = mne.io.read_raw_edf(edf).pick_channels(
['EOG1', 'EOG2', 'C3', 'C4', 'PZ', 'RAH1', 'RAH2', 'LAH1',
'LAH2']).resample(1000)
raw = mne.set_bipolar_reference(raw,
'RAH1',
'RAH2',
ch_name='RAH1-RAH2',
drop_refs=False)
raw = mne.set_bipolar_reference(raw,
'LAH1',
'LAH2',
ch_name='LAH1-LAH2',
drop_refs=False)
depth = np.zeros((1, raw.n_times))
for evt in depth_epochs.values:
depth[0][250 * evt[0]:250 * evt[0] + 250] = 1
scalp = np.zeros((1, raw.n_times))
for evt in scalp_epochs.values:
scalp[0][250 * evt[0]:250 * evt[0] + 250] = 1
def load_data(bids_path):
# read_raw_bids automatically
# - populates bad channels using the BIDS channels.tsv
# - sets channels types according to BIDS channels.tsv `type` column
# - sets raw.annotations using the BIDS events.tsv
subject = bids_path.subject
session = bids_path.session
extra_params = dict()
if config.allow_maxshield:
extra_params['allow_maxshield'] = config.allow_maxshield
raw = read_raw_bids(bids_path=bids_path, extra_params=extra_params)
if subject != 'emptyroom':
# Crop the data.
if config.crop is not None:
raw.crop(*config.crop)
# Rename events.
if config.rename_events:
rename_events(raw=raw, subject=subject, session=session)
raw.load_data()
if hasattr(raw, 'fix_mag_coil_types'):
raw.fix_mag_coil_types()
montage_name = config.eeg_template_montage
if config.get_datatype() == 'eeg' and montage_name:
msg = (f'Setting EEG channel locatiions to template montage: '
f'{montage_name}.')
logger.info(
gen_log_message(message=msg,
step=1,
subject=subject,
session=session))
montage = mne.channels.make_standard_montage(montage_name)
raw.set_montage(montage, match_case=False, on_missing='warn')
if config.ch_types == ['eeg'] and config.eeg_bipolar_channels:
msg = 'Creating bipolar channels …'
logger.info(
gen_log_message(message=msg,
step=1,
subject=subject,
session=session))
raw.load_data()
for ch_name, (anode, cathode) in config.eeg_bipolar_channels.items():
msg = f' {anode} – {cathode} -> {ch_name}'
logger.info(
gen_log_message(message=msg,
step=1,
subject=subject,
session=session))
mne.set_bipolar_reference(raw,
anode=anode,
cathode=cathode,
ch_name=ch_name,
drop_refs=False,
copy=False)
# If we created a new bipolar channel that the user wishes to
# use as an EOG channel, it is probably a good idea to set its channel
# type to 'eog'. Bipolar channels, by default, don't have a location,
# so one might get unexpected results otherwise, as the channel would
# influence e.g. in GFP calculations, but not appear on topographic
# maps.
if any([
eog_ch_name in config.eeg_bipolar_channels
for eog_ch_name in config.eog_channels
]):
msg = 'Setting channel type of new bipolar EOG channel(s) …'
logger.info(
gen_log_message(message=msg,
step=1,
subject=subject,
session=session))
for eog_ch_name in config.eog_channels:
if eog_ch_name in config.eeg_bipolar_channels:
msg = f' {eog_ch_name} -> eog'
logger.info(
gen_log_message(message=msg,
step=1,
subject=subject,
session=session))
raw.set_channel_types({eog_ch_name: 'eog'})
if config.drop_channels:
msg = f'Dropping channels: {", ".join(config.drop_channels)}'
logger.info(
gen_log_message(message=msg,
step=1,
subject=subject,
session=session))
raw.drop_channels(config.drop_channels)
return raw
def test_set_bipolar_reference():
"""Test bipolar referencing."""
raw = read_raw_fif(fif_fname, preload=True)
raw.apply_proj()
reref = set_bipolar_reference(raw, 'EEG 001', 'EEG 002', 'bipolar',
{'kind': FIFF.FIFFV_EOG_CH,
'extra': 'some extra value'})
assert (reref.info['custom_ref_applied'])
# Compare result to a manual calculation
a = raw.copy().pick_channels(['EEG 001', 'EEG 002'])
a = a._data[0, :] - a._data[1, :]
b = reref.copy().pick_channels(['bipolar'])._data[0, :]
assert_allclose(a, b)
# Original channels should be replaced by a virtual one
assert ('EEG 001' not in reref.ch_names)
assert ('EEG 002' not in reref.ch_names)
assert ('bipolar' in reref.ch_names)
# Check channel information
bp_info = reref.info['chs'][reref.ch_names.index('bipolar')]
an_info = reref.info['chs'][raw.ch_names.index('EEG 001')]
for key in bp_info:
if key == 'loc':
assert_array_equal(bp_info[key], 0)
elif key == 'coil_type':
assert_equal(bp_info[key], FIFF.FIFFV_COIL_EEG_BIPOLAR)
elif key == 'kind':
assert_equal(bp_info[key], FIFF.FIFFV_EOG_CH)
else:
assert_equal(bp_info[key], an_info[key])
assert_equal(bp_info['extra'], 'some extra value')
# Minimalist call
reref = set_bipolar_reference(raw, 'EEG 001', 'EEG 002')
assert ('EEG 001-EEG 002' in reref.ch_names)
# Minimalist call with twice the same anode
reref = set_bipolar_reference(raw,
['EEG 001', 'EEG 001', 'EEG 002'],
['EEG 002', 'EEG 003', 'EEG 003'])
assert ('EEG 001-EEG 002' in reref.ch_names)
assert ('EEG 001-EEG 003' in reref.ch_names)
# Set multiple references at once
reref = set_bipolar_reference(
raw,
['EEG 001', 'EEG 003'],
['EEG 002', 'EEG 004'],
['bipolar1', 'bipolar2'],
[{'kind': FIFF.FIFFV_EOG_CH, 'extra': 'some extra value'},
{'kind': FIFF.FIFFV_EOG_CH, 'extra': 'some extra value'}],
)
a = raw.copy().pick_channels(['EEG 001', 'EEG 002', 'EEG 003', 'EEG 004'])
a = np.array([a._data[0, :] - a._data[1, :],
a._data[2, :] - a._data[3, :]])
b = reref.copy().pick_channels(['bipolar1', 'bipolar2'])._data
assert_allclose(a, b)
# Test creating a bipolar reference that doesn't involve EEG channels:
# it should not set the custom_ref_applied flag
reref = set_bipolar_reference(raw, 'MEG 0111', 'MEG 0112',
ch_info={'kind': FIFF.FIFFV_MEG_CH},
verbose='error')
assert (not reref.info['custom_ref_applied'])
assert ('MEG 0111-MEG 0112'[:15] in reref.ch_names)
# Test a battery of invalid inputs
pytest.raises(ValueError, set_bipolar_reference, raw,
'EEG 001', ['EEG 002', 'EEG 003'], 'bipolar')
pytest.raises(ValueError, set_bipolar_reference, raw,
['EEG 001', 'EEG 002'], 'EEG 003', 'bipolar')
pytest.raises(ValueError, set_bipolar_reference, raw,
'EEG 001', 'EEG 002', ['bipolar1', 'bipolar2'])
pytest.raises(ValueError, set_bipolar_reference, raw,
'EEG 001', 'EEG 002', 'bipolar',
ch_info=[{'foo': 'bar'}, {'foo': 'bar'}])
pytest.raises(ValueError, set_bipolar_reference, raw,
'EEG 001', 'EEG 002', ch_name='EEG 003')
def test_set_bipolar_reference():
"""Test bipolar referencing."""
raw = read_raw_fif(fif_fname, preload=True)
raw.apply_proj()
reref = set_bipolar_reference(raw, 'EEG 001', 'EEG 002', 'bipolar',
{'kind': FIFF.FIFFV_EOG_CH,
'extra': 'some extra value'})
assert (reref.info['custom_ref_applied'])
# Compare result to a manual calculation
a = raw.copy().pick_channels(['EEG 001', 'EEG 002'])
a = a._data[0, :] - a._data[1, :]
b = reref.copy().pick_channels(['bipolar'])._data[0, :]
assert_allclose(a, b)
# Original channels should be replaced by a virtual one
assert ('EEG 001' not in reref.ch_names)
assert ('EEG 002' not in reref.ch_names)
assert ('bipolar' in reref.ch_names)
# Check channel information
bp_info = reref.info['chs'][reref.ch_names.index('bipolar')]
an_info = reref.info['chs'][raw.ch_names.index('EEG 001')]
for key in bp_info:
if key == 'loc':
assert_array_equal(bp_info[key], 0)
elif key == 'coil_type':
assert_equal(bp_info[key], FIFF.FIFFV_COIL_EEG_BIPOLAR)
elif key == 'kind':
assert_equal(bp_info[key], FIFF.FIFFV_EOG_CH)
else:
assert_equal(bp_info[key], an_info[key])
assert_equal(bp_info['extra'], 'some extra value')
# Minimalist call
reref = set_bipolar_reference(raw, 'EEG 001', 'EEG 002')
assert ('EEG 001-EEG 002' in reref.ch_names)
# Minimalist call with twice the same anode
reref = set_bipolar_reference(raw,
['EEG 001', 'EEG 001', 'EEG 002'],
['EEG 002', 'EEG 003', 'EEG 003'])
assert ('EEG 001-EEG 002' in reref.ch_names)
assert ('EEG 001-EEG 003' in reref.ch_names)
# Set multiple references at once
reref = set_bipolar_reference(
raw,
['EEG 001', 'EEG 003'],
['EEG 002', 'EEG 004'],
['bipolar1', 'bipolar2'],
[{'kind': FIFF.FIFFV_EOG_CH, 'extra': 'some extra value'},
{'kind': FIFF.FIFFV_EOG_CH, 'extra': 'some extra value'}],
)
a = raw.copy().pick_channels(['EEG 001', 'EEG 002', 'EEG 003', 'EEG 004'])
a = np.array([a._data[0, :] - a._data[1, :],
a._data[2, :] - a._data[3, :]])
b = reref.copy().pick_channels(['bipolar1', 'bipolar2'])._data
assert_allclose(a, b)
# Test creating a bipolar reference that doesn't involve EEG channels:
# it should not set the custom_ref_applied flag
reref = set_bipolar_reference(raw, 'MEG 0111', 'MEG 0112',
ch_info={'kind': FIFF.FIFFV_MEG_CH},
verbose='error')
assert (not reref.info['custom_ref_applied'])
assert ('MEG 0111-MEG 0112'[:15] in reref.ch_names)
# Test a battery of invalid inputs
pytest.raises(ValueError, set_bipolar_reference, raw,
'EEG 001', ['EEG 002', 'EEG 003'], 'bipolar')
pytest.raises(ValueError, set_bipolar_reference, raw,
['EEG 001', 'EEG 002'], 'EEG 003', 'bipolar')
pytest.raises(ValueError, set_bipolar_reference, raw,
'EEG 001', 'EEG 002', ['bipolar1', 'bipolar2'])
pytest.raises(ValueError, set_bipolar_reference, raw,
'EEG 001', 'EEG 002', 'bipolar',
ch_info=[{'foo': 'bar'}, {'foo': 'bar'}])
pytest.raises(ValueError, set_bipolar_reference, raw,
'EEG 001', 'EEG 002', ch_name='EEG 003')
'C32', 'EXG1', 'EXG2', 'EXG3', 'EXG4', 'EXG5', 'EXG6', 'EXG7', 'EXG8', 'GSR1', 'GSR2', 'Erg1', 'Erg2', 'Resp', 'Plet', 'Temp'])
Data_rawEEG.rename_channels(mapping={'A1':'Fp1', 'A2':'AF7', 'A3':'AF3', 'A4':'F1', 'A5':'F3', 'A6':'F5', 'A7':'F7',
'A8':'FT7', 'A9':'FC5', 'A10':'FC3', 'A11':'FC1', 'A12':'C1', 'A13':'C3', 'A14':'C5',
'A15':'T7', 'A16':'TP7', 'A17':'CP5', 'A18':'CP3', 'A19':'CP1', 'A20':'P1', 'A21':'P3',
'A22':'P5', 'A23':'P7', 'A24':'P9', 'A25':'PO7', 'A26':'PO3', 'A27':'O1', 'A28':'Iz',
'A29':'Oz', 'A30':'POz', 'A31':'Pz', 'A32':'CPz', 'B1':'Fpz', 'B2':'Fp2', 'B3':'AF8',
'B4':'AF4', 'B5':'AFz', 'B6':'Fz', 'B7':'F2', 'B8':'F4', 'B9':'F6', 'B10':'F8',
'B11':'FT8', 'B12':'FC6', 'B13':'FC4', 'B14':'FC2', 'B15':'FCz', 'B16':'Cz', 'B17':'C2',
'B18':'C4', 'B19':'C6', 'B20':'T8', 'B21':'TP8', 'B22':'CP6', 'B23':'CP4', 'B24':'CP2',
'B25':'P2', 'B26':'P4', 'B27':'P6', 'B28':'P8', 'B29':'P10', 'B30':'PO8', 'B31':'PO4',
'B32':'O2','C1':'mastoidleft', 'C2':'mastoidright', 'C3':'HEOGleft','C4':'HEOGright',
'C5':'VEOGup', 'C6':'VEOGdown','Status':'triggers'})
#VEOG< HEOG setup
Data_rawEEG = mne.set_bipolar_reference(Data_rawEEG,anode='VEOGup',cathode='VEOGdown',ch_name='VEOG')
Data_rawEEG = mne.set_bipolar_reference(Data_rawEEG,anode='HEOGleft',cathode='HEOGright',ch_name='HEOG')
Data_rawEEG.set_channel_types({'VEOG':'eog','HEOG':'eog'})
## Setting EEG reference
Data_rawEEG.set_eeg_reference(['mastoidleft','mastoidright'])
Data_rawEEG.drop_channels(['mastoidleft','mastoidright'])
## Filter on data
low_f = 0.1 #Low edge of the high-pass filter
high_f = 30 #High edge of the low-pass filter
picks_eeg_eog = mne.pick_types(Data_rawEEG.info, meg=False, eeg=True, eog=True) #select the eeg and the eog channels for filtering
raw = mne.io.read_raw_bdf(data_fname, preload=True)
raw = raw.pick(picks=[
'EXG1', 'EXG2', 'EXG3', 'EXG4', 'EXG5', 'EXG6', 'EXG7', 'EXG8', 'Erg1',
'Erg2', 'Status'
])
mne_events = mne.find_events(raw)
if name_subj == "S3": #removes incpmleted trail for S3
mne_events = mne_events[:-5]
mne_events[:, 2] = mne_events[:, 2] - mne_events[:, 1]
mne_events = np.array(
[list(x) for x in mne_events if x[2] in initial_expctd_events])
mne.set_bipolar_reference(raw,anode=['EXG1','EXG3'], cathode=['EXG2','EXG4'],ch_name=['EMG_L','EMG_R'],\
copy=False)
raw.pick_channels(['EMG_L', 'EMG_R', 'Erg1', 'Erg2'])
raw.set_channel_types({
'EMG_L': 'emg',
'EMG_R': 'emg',
'Erg1': 'misc',
'Erg2': 'misc'
})
emg_channels_idx = {0: 'EMG_L', 1: 'EMG_R'}
force_channels_idx = {2: 'Erg1', 3: 'Erg2'}
channel_names = list(emg_channels_idx.values()) + list(
force_channels_idx.values())
#Filtering
raw = dbt.use_mne.apply_filter(raw, ch_names=['EMG_L', 'EMG_R'], low_cutoff=10)
nameBDF = f
Trigger = [11, 21, 31, 41]
#65314
fname = os.path.join(pathBDF, nameBDF)
# raw = mne.io.read_raw_edf(fname,preload=True, verbose=True)
raw = mne.io.read_raw_brainvision(fname,
montage=None,
eog=('HEOG_m', 'HEOG_p', 'VEOG_m',
'VEOG_p'),
misc='auto',
scale=1.0,
preload=True,
verbose=True)
raw
mne.set_bipolar_reference(raw,anode=['EMGg_m','EMGd_m'],\
cathode=['EMGg_p','EMGd_p'],ch_name=['EMG_L','EMG_R'],\
copy=False)
# events = mne.find_events(raw, stim_channel = 'STI 014', min_duration=0.002, mask=2**24-64)
events = mne.find_events(raw)
raw.pick_channels(['EMG_L', 'EMG_R'])
raw.set_channel_types({
'EMG_L': 'emg',
'EMG_R': 'emg',
# 'EXG5' :'eog',
# 'EXG6' :'eog',
# 'Erg1' :'misc',
# 'Erg2': 'misc',
# 'STI 014': 'stim'
})
'C1': 'M1',
'C2': 'M2',
'C3': 'HEOGleft',
'C4': 'HEOGright',
'C5': 'VEOGup',
'C6': 'VEOGdown',
'Status': 'triggers'
})
# Setting channel montage
montage = mne.channels.read_montage('biosemi64')
raw.set_montage(montage)
# Creating EOG channels
raw = mne.set_bipolar_reference(raw,
anode='VEOGup',
cathode='VEOGdown',
ch_name='VEOG')
raw = mne.set_bipolar_reference(raw,
anode='HEOGleft',
cathode='HEOGright',
ch_name='HEOG')
raw.set_channel_types({'VEOG': 'eog', 'HEOG': 'eog'})
# Setting reference (mastoids, but given different name for better electrode location)
raw.set_eeg_reference(['M1', 'M2'])
raw.drop_channels(['M1', 'M2'])
# raw.set_eeg_reference(ref_channels='average')
# # Investigating raw data (Identify bad channels and add to list above)
# raw.plot()
# raw.plot_psd(fmax=70)
def rereference(edf: mne.io.RawArray,
desired_ref: str,
current_ref: str = None,
pick_chans: list = None) -> Tuple[mne.io.RawArray, str]:
"""
reference an edf. M1 and M2 should be included, but NOT as eeg stype in the mne.io.RawArray.
contra_mastoid reref will be undone if needed but this requires M1 and M2 PRIOR to any referencing.
:param edf: edf to re-ref
:param current_ref: current reference type, can be "contra_mastoid",'linked_ear' or a channel name
:param desired_ref: current reference type, can be "contra_mastoid",'linked_ear' or a channel name.
If None, assumed it is a common channels
:return:
"""
if pick_chans is None:
chans = edf.ch_names
else:
chans = pick_chans
if current_ref == desired_ref:
return edf, current_ref
if desired_ref in ['linked_ear'
] and 'M1' not in chans or 'M2' not in chans:
warnings.warn(
'Trying to reference to linked ear, but missing M1 and M2 channels. EEG file will not be re-referenced',
EEGWarning)
return edf, current_ref
if current_ref == 'contra_mastoid':
to_reref = [ch for ch in chans if ch not in ['M1', 'M2']]
left = [
ch for ch in to_reref
if len([n for n in ['1', '3', '5', '7', '9'] if n in ch]) > 0
]
right = [
ch for ch in to_reref
if len([n for n in ['2', '4', '6', '8', '10'] if n in ch]) > 0
]
if len(left) > 0 and 'M2' not in chans:
warnings.warn(
'Trying to reference to left channels to M2 ear, but missing M2 channel. left channels cannot be unreferenced'
)
left_ref = []
left = []
else:
left_ref = ['M2'] * len(left)
if len(right) > 0 and 'M1' not in chans:
warnings.warn(
'Trying to reference to right channels to M1 ear, but missing M1 channel. right channels cannot be unreferenced'
)
right_ref = []
right = []
else:
right_ref = ['M1'] * len(right)
edf = edf.apply_function(lambda x: -x, picks=['M1', 'M2'])
edf = mne.set_bipolar_reference(edf,
left + right,
left_ref + right_ref,
drop_refs=False,
verbose=False)
edf = edf.drop_channels(left + right)
edf.rename_channels({ch: ch.split('-')[0] for ch in edf.ch_names})
edf = edf.apply_function(lambda x: -x, picks=['M1', 'M2'])
ref_type = desired_ref
if desired_ref == 'contra_mastoid':
if current_ref == 'linked_ear':
edf = edf.apply_function(lambda x: -x, picks=['M1', 'M2'])
edf, _ = mne.set_eeg_reference(edf,
ref_channels=['M1', 'M2'],
verbose=False)
edf = edf.apply_function(lambda x: -x, picks=['M1', 'M2'])
to_reref = [ch for ch in chans if ch not in ['M1', 'M2']]
left = [
ch for ch in to_reref
if len([n for n in ['1', '3', '5', '7', '9', 'z'] if n in ch]) > 0
]
right = [
ch for ch in to_reref
if len([n for n in ['2', '4', '6', '8', '10'] if n in ch]) > 0
]
if len(left) > 0 and 'M2' not in chans:
warnings.warn(
'Trying to reference to left channels to M2 ear, but missing M2 channel. left channels will not be re-referenced'
)
left_ref = []
left = []
ref_type = 'contra_right_only'
else:
left_ref = ['M2'] * len(left)
if len(right) > 0 and 'M1' not in chans:
warnings.warn(
'Trying to reference to right channels to M1 ear, but missing M1 channel. right channels will not be re-referenced'
)
right_ref = []
right = []
ref_type = 'contra_left_only'
else:
right_ref = ['M1'] * len(right)
edf = mne.set_bipolar_reference(edf,
left + right,
left_ref + right_ref,
drop_refs=False,
verbose=False)
edf = edf.drop_channels(left + right)
edf.rename_channels({ch: ch.split('-')[0] for ch in edf.ch_names})
elif desired_ref == 'linked_ear':
edf, _ = mne.set_eeg_reference(edf,
ref_channels=['M1', 'M2'],
verbose=False)
else:
edf, _ = mne.set_eeg_reference(edf,
ref_channels=desired_ref,
verbose=False)
return edf, ref_type
# %%
# Using a bipolar reference
# ^^^^^^^^^^^^^^^^^^^^^^^^^
#
# To create a bipolar reference, you can use :meth:`~mne.set_bipolar_reference`
# along with the respective channel names for ``anode`` and ``cathode`` which
# creates a new virtual channel that takes the difference between two
# specified channels (anode and cathode) and drops the original channels by
# default. The new virtual channel will be annotated with the channel info of
# the anode with location set to ``(0, 0, 0)`` and coil type set to
# ``EEG_BIPOLAR`` by default. Here we use a contralateral/transverse bipolar
# reference between channels ``EEG 054`` and ``EEG 055`` as described in
# :footcite:`YaoEtAl2019` which creates a new virtual channel
# named ``EEG 054-EEG 055``.
raw_bip_ref = mne.set_bipolar_reference(raw, anode=['EEG 054'],
cathode=['EEG 055'])
raw_bip_ref.plot()
# %%
# EEG reference and source modeling
# ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
#
# If you plan to perform source modeling (either with EEG or combined EEG/MEG
# data), it is **strongly recommended** to use the
# average-reference-as-projection approach. It is important to use an average
# reference because using a specific
# reference sensor (or even an average of a few sensors) spreads the forward
# model error from the reference sensor(s) into all sensors, effectively
# amplifying the importance of the reference sensor(s) when computing source
# estimates. In contrast, using the average of all EEG channels as reference
# spreads the forward modeling error evenly across channels, so no one channel
anodes = [
'T1', 'Fp1', 'F7', 'T3', 'T5', 'F3', 'T3', 'Fz', 'Cz', 'F4', 'T4', 'Fp2',
'F8', 'T4', 'T6', 'T2'
]
cathodes = [
'A1', 'F7', 'T3', 'T5', 'O1', 'T3', 'P3', 'Cz', 'Pz', 'T4', 'P4', 'F8',
'T4', 'T6', 'O2', 'A2'
]
#Read and set original montage
montage = mne.channels.make_standard_montage('standard_postfixed')
epochs.set_montage(montage, raise_if_subset=False)
# Set bipolar montage
epochs_bipolar = mne.set_bipolar_reference(epochs,
anodes,
cathodes,
drop_refs=False)
# Removing old channels (keeping only biploar)
epochs_bipolar.picks = None
epochs_bipolar.drop_channels(epochs.info['ch_names'])
#Print info for bipolar (double banana) reference raw data
print(epochs_bipolar)
print(epochs_bipolar.info['ch_names'])
# =============================================================================
# # reordering bipolar channels (given original setup of channels)
# ch_order = ['Fp1-F7', 'F7-T3', 'T3-T5', 'T5-O1',
# 'Fp1-F3', 'F3-C3', 'C3-P3', 'P3-O1',
# 'Fz-Cz', 'Cz-Pz',
def preprocess(raw):
## Input Variables ##############################
trialLen = 1 # Number of seconds per trial
#cutoff = 0.5 # Cut first/last .5 seconds from each
downsample_fs = 1500 # What frequency to downsample to
##################################################
# Filter
raw.filter(0.1, 200)
# data resample
raw.resample(downsample_fs)
# Cut data into trials (epochs) mne.EpochsArray
picks = mne.pick_types(raw.info, eeg=True)
data, times = raw[:]
events = mne.find_events(raw)
# Make trial events
trialSamps = downsample_fs * trialLen
dataShape = data.shape
overflow = dataShape[1] % trialSamps
# MNE forces you to use events to set timing for trials
# Create events with event_id=9 for trial start timings (after our 1-8 ttls)
trialEvents = []
n = int(raw.first_time) + 1
while n <= dataShape[1] - overflow + 1:
trialEvents.append([n, 0, 9]) # [timestamp, N/A, event_id]
n += trialSamps
raw.add_events(np.array(trialEvents), stim_channel='TTL') # adding our trial events
events = mne.find_events(raw)
# Split into trials based on trial events
processedObj = mne.Epochs(raw, events, event_id=9, tmin = 0.0, tmax = trialLen, baseline = None, preload=True)
scalings = {'eeg':2000}
# Plot to look for bad trials/channesl
processedObj.plot(scalings = scalings, n_epochs = 2, show=True, block=True)
print(processedObj)
# Bipolar ref
bi1 = []
bi2 = []
for n in range(0,16, 2):
if 'CH'+str(n) not in processedObj.info['bads']:
bi1.append(processedObj.ch_names[n])
if 'CH'+str(n+1) not in processedObj.info['bads']:
bi2.append(processedObj.ch_names[n+1])
print(bi1, bi2)
if len(processedObj.info['bads']) > 0:
print('Some bad channels')
print(processedObj.ch_names)
# Bipolar based upon channel names
processedObj = mne.set_bipolar_reference(processedObj, bi1, bi2)
return processedObj
def load_data(bids_path):
# read_raw_bids automatically
# - populates bad channels using the BIDS channels.tsv
# - sets channels types according to BIDS channels.tsv `type` column
# - sets raw.annotations using the BIDS events.tsv
subject = bids_path.subject
session = bids_path.session
extra_params = dict()
if config.allow_maxshield:
extra_params['allow_maxshield'] = config.allow_maxshield
raw = read_raw_bids(bids_path=bids_path, extra_params=extra_params)
if subject != 'emptyroom':
# Crop the data.
if config.crop is not None:
raw.crop(*config.crop)
# Rename events.
if config.rename_events:
rename_events(raw=raw, subject=subject, session=session)
raw.load_data()
if hasattr(raw, 'fix_mag_coil_types'):
raw.fix_mag_coil_types()
montage_name = config.eeg_template_montage
if config.get_datatype() == 'eeg' and montage_name:
msg = (f'Setting EEG channel locatiions to template montage: '
f'{montage_name}.')
logger.info(
gen_log_message(message=msg,
step=1,
subject=subject,
session=session))
montage = mne.channels.make_standard_montage(montage_name)
raw.set_montage(montage, match_case=False, on_missing='warn')
if config.ch_types == ['eeg'] and config.eeg_bipolar_channels:
msg = 'Creating bipolar channels …'
logger.info(
gen_log_message(message=msg,
step=1,
subject=subject,
session=session))
raw.load_data()
for ch_name, (anode, cathode) in config.eeg_bipolar_channels.items():
msg = f' {anode} – {cathode} -> {ch_name}'
logger.info(
gen_log_message(message=msg,
step=1,
subject=subject,
session=session))
mne.set_bipolar_reference(raw,
anode=anode,
cathode=cathode,
ch_name=ch_name,
drop_refs=False,
copy=False)
if config.drop_channels:
msg = f'Dropping channels: {", ".join(config.drop_channels)}'
logger.info(
gen_log_message(message=msg,
step=1,
subject=subject,
session=session))
raw.drop_channels(config.drop_channels)
return raw
def sync_matlab_and_eeg_events(edf_filename: str,
timing_filename: str,
event_filename: str = None,
init_slice: int = None) -> object:
"""Syncs the matlab events with times in the EEG data, returns imported eeg raw eeg and events"""
sample_rate = 128
channels_to_exclude = [
'COUNTER', 'INTERPOLATED', 'RAW_CQ', 'GYROX', 'GYROY', 'MARKER',
'MARKER_HARDWARE', 'SYNC', 'TIME_STAMP_s', 'TIME_STAMP_ms'
]
impedance_channel_map = {
'CQ_AF3': 'im_F8',
'CQ_F7': 'im_F5',
'CQ_F3': 'im_Cz',
'CQ_FC5': 'im_C3',
'CQ_T7': 'im_Fz',
'CQ_P7': 'im_Sync1',
'CQ_O1': 'im_F7',
'CQ_O2': 'im_Sync2',
'CQ_P8': 'im_P7',
'CQ_T8': 'im_P3',
'CQ_FC6': 'im_P8',
'CQ_F4': 'im_Oz',
'CQ_F8': 'im_Pz',
'CQ_AF4': 'im_P4',
'CQ_CMS': 'im_CMS',
'CQ_DRL': 'im_DRL'
}
channel_map = {
'O1': 'F7',
'P7': 'Sync1',
'T7': 'Fz',
'F7': 'F5',
'AF3': 'F8',
'FC5': 'C3',
'F3': 'Cz',
'O2': 'Sync2',
'P8': 'P7',
'T8': 'P3',
'F8': 'Pz',
'AF4': 'P4',
'FC6': 'P8',
'F4': 'Oz'
}
data_channels = [value for key, value in channel_map.items()]
immedence_channels = [
value for key, value in impedance_channel_map.items()
]
channel_map.update(impedance_channel_map)
raw_edf = mne.io.read_raw_edf(edf_filename,
stim_channel=None,
exclude=channels_to_exclude)
try:
raw_edf.load_data()
except IndexError:
warnings.warn("Failed to load data, skipping subject")
return
raw_edf.rename_channels(channel_map)
raw_edf = mne.set_bipolar_reference(raw_edf, "Sync1", "Sync2", "SyncRR")
sync_channel0 = raw_edf.copy().pick_channels(["SyncRR"])
raw_im = raw_edf.copy().pick_channels(immedence_channels)
raw_edf.pick_channels(data_channels)
##Process sync channels
data, times = sync_channel0[:] # Extract data
data = data.transpose()
data = zscore(data) # normalize
times -= init_slice
data = data[times > 0]
times = times[times > 0]
# plt.plot(np.asarray(times), data)
# plt.show()
diff_data_raw = np.insert(np.diff(data, axis=0), 0, 0)
diff_data = diff_data_raw.copy()
threshold_moving = window_stdev(
diff_data, 5000) * 3.5 # Calculate a moving threshold by
diff_data[np.where((diff_data > -threshold_moving)
& (diff_data < threshold_moving))] = 0 # Threshold
diff_data[np.where(
diff_data < 0)] = -diff_data[np.where(diff_data < 0)] # Rectify
# plt.plot(np.asarray(times), diff_data)
eeg_times = detect_peaks(diff_data, mph=0.2)
eeg_times_edge = np.where(diff_data_raw[eeg_times] > 0, 1,
0) # rising edge is 1
rising = eeg_times[eeg_times_edge == 1]
falling = eeg_times[eeg_times_edge == 0]
# plt.plot(rising / sample_rate, np.ones(rising.shape), '*')
# plt.plot(falling / sample_rate, np.ones(falling.shape), '*')
eeg_times_secs = eeg_times / sample_rate
event_data = pd.read_excel(timing_filename)
event_data = event_data[['label', 'getSecsLogTime', 'realTime']]
event_data.columns = ["label", "matlab_time", 'datetime']
event_data['matlab_time'] -= event_data['matlab_time'][0] # Start at zero
event_data['eeg_time'] = np.nan
recogRespTimes = event_data.label.apply(
lambda lab: 'recogRespTime' not in lab)
event_data_working = add_expected_transitions(event_data[recogRespTimes])
max_expected_lag = 5 # Initial Differences between matlab and eeg should not be more than 5
event_data_working['label_generic'] = event_data_working['label'].apply(
lambda lab: lab[0:lab.rfind("_")])
# from collections import defaultdict
# time_testing_dict = defaultdict(list)
event_data.set_index('label', inplace=True, drop=False)
matlab_labels = event_data_working['label_generic'].unique()
for label in matlab_labels:
timing_data_slice = event_data_working[
label == event_data_working['label_generic']]
print("For", label, "attempting to find",
len(timing_data_slice.matlab_time), "events")
# plt.figure()
# plt.plot(timing_data_slice.matlab_time, np.ones(len(timing_data_slice.matlab_time)), '*')
# a = eeg_times_secs[(eeg_times_secs < (timing_data_slice.matlab_time.iloc[-1] + max_expected_lag)) & (
# eeg_times_secs > (timing_data_slice.matlab_time.iloc[0] - max_expected_lag))]
# plt.plot(a, np.ones(len(a)), '*')
# plt.show()
expected_diff = np.diff(timing_data_slice.matlab_time)
if label in ['syncDataLight_BeforeInter_flipTime']:
sync_idx, sync_lag = find_template_in_sequence(
eeg_times_secs, eeg_times_edge, expected_diff,
timing_data_slice.trans_bool, 0, eeg_times_secs[-1] / 4)
if sync_idx is None:
warnings.warn('Did not fine sync signal. Terminating....')
return
eeg_times_secs = eeg_times_secs[sync_idx[0]:] - sync_lag
eeg_times_edge = eeg_times_edge[sync_idx[0]:]
lag_eeg_to_matlab = sync_lag
print('Found start at', sync_lag)
sync_idx = range(0, len(sync_idx))
else:
sync_idx, sync_lag = find_template_in_sequence(
eeg_times_secs, eeg_times_edge, expected_diff,
timing_data_slice.trans_bool,
timing_data_slice.matlab_time.iloc[0] - max_expected_lag,
timing_data_slice.matlab_time.iloc[-1] + max_expected_lag)
# Linearly increasing lag
max_expected_lag = 5 + 0.01 * timing_data_slice.matlab_time.iloc[-1]
if sync_idx is None:
print('Did not find all points for', label)
continue
# for key, val in time_testing.items():
# time_testing_dict[key].extend(val)
# print("Found", len(sync_idx), "events at", sync_lag)
for lab, val in zip(timing_data_slice.label, eeg_times_secs[sync_idx]):
event_data.set_value(lab, 'eeg_time', val)
event_data = add_event_data_missing_sync(event_data)
event_data.sort_values('matlab_time', inplace=True)
events = interpolate_missing(event_data)
events['type'] = events['label'].apply(lambda lab: lab[0:lab.rfind("_")])
events['eeg_samples'] = events['eeg_time'] * sample_rate
events.set_index('type', inplace=True, drop=False)
events.lag_eeg_to_matlab = lag_eeg_to_matlab
events.to_csv(event_filename)
# times -= lag_eeg_to_matlab
# data = data[times > 0]
# times = times[times > 0]
# plot_sync_times(data, times, event_data)
eeg = DotDict({})
eeg.raw = raw_edf
eeg.impedance = raw_im
slice_and_add_eeg_montage(eeg, lag_eeg_to_matlab) #Modifies inplace
return eeg, events