def test_to_data_frame(fname):
"""Test EDF/BDF Raw Pandas exporter."""
ext = op.splitext(fname)[1][1:].lower()
if ext == 'edf':
raw = read_raw_edf(fname, preload=True, verbose='error')
elif ext == 'bdf':
raw = read_raw_bdf(fname, preload=True, verbose='error')
_, times = raw[0, :10]
df = raw.to_data_frame()
assert (df.columns == raw.ch_names).all()
assert_array_equal(np.round(times * 1e3), df.index.values[:10])
df = raw.to_data_frame(index=None, scalings={'eeg': 1e13})
assert 'time' in df.index.names
assert_array_equal(df.values[:, 0], raw._data[0] * 1e13)
def test_load_generator(fname, recwarn):
"""Test IO of annotations from edf and bdf files with raw info."""
ext = op.splitext(fname)[1][1:].lower()
if ext == 'edf':
raw = read_raw_edf(fname)
elif ext == 'bdf':
raw = read_raw_bdf(fname)
assert len(raw.annotations.onset) == 2
found_types = [k for k, v in
channel_indices_by_type(raw.info, picks=None).items()
if v]
assert len(found_types) == 1
events, event_id = events_from_annotations(raw)
ch_names = ['squarewave', 'ramp', 'pulse', 'ECG', 'noise', 'sine 1 Hz',
'sine 8 Hz', 'sine 8.5 Hz', 'sine 15 Hz', 'sine 17 Hz',
'sine 50 Hz']
assert raw.get_data().shape == (11, 120000)
assert raw.ch_names == ch_names
assert event_id == {'RECORD START': 2, 'REC STOP': 1}
assert_array_equal(events, [[0, 0, 2], [120000, 0, 1]])
def test_bdf_crop_save_stim_channel(tmpdir):
"""Test EDF with various sampling rates."""
raw = read_raw_bdf(bdf_stim_channel_path)
raw.save(tmpdir.join('test-raw.fif'), tmin=1.2, tmax=4.0, overwrite=True)
# channels to be exclude from import
exclude = ['EXG5', 'EXG6', 'EXG7', 'EXG8']
# ========================================================================
# --------- loop through files and make bids-files blocks ---------------
for ind, file in enumerate(files):
# --- 1) set up paths and file names -----------------------
filepath, filename = os.path.split(file)
# subject in question
subj = re.findall(r'\d+', filename)[0].rjust(3, '0')
# --- 2) import the data -----------------------------------
raw = read_raw_bdf(file,
preload=False,
exclude=exclude)
# --- 3) dataset info --------------------------------------
# keep the sampling rate
sfreq = raw.info['sfreq']
# all channels in raw
chans = raw.info['ch_names']
# channels in montage
montage_chans = montage.ch_names
# nr of eeg channels
n_eeg = len([chan for chan in chans if chan in montage_chans])
# channel types
types = []
for chan in chans:
if chan in montage_chans:
def copyfile_edf(src, dest, anonymize=None):
"""Copy an EDF, EDF+, or BDF file to a new location, optionally anonymize.
.. warning:: EDF/EDF+/BDF files contain two fields for recording dates:
A generic "startdate" field that supports only 2-digit years,
and a "Startdate" field as part of the "local recording
identification", which supports 4-digit years.
If you want to anonymize your file, MNE-BIDS will set the
"startdate" field to 85 (i.e., 1985), the earliest possible
date for that field. However, the "Startdate" field in the
file's "local recording identification" and the date in the
session's corresponding ``scans.tsv`` will be set correctly
according to the argument provided to the ``anonymize``
parameter. Note that it is possible that not all EDF/EDF+/BDF
reading software parses the accurate recording date, and
that for some reading software, the wrong year (1985) may
be parsed.
Parameters
----------
src : str | pathlib.Path
The source path of the .edf or .bdf file to be copied.
dest : str | pathlib.Path
The destination path of the .edf or .bdf file.
anonymize : dict | None
If None (default), no anonymization is performed.
If dict, data will be anonymized depending on the keys provided with
the dict: `daysback` is a required key, `keep_his` is an optional key.
`daysback` : int
Number of days by which to move back the recording date in time.
In studies with multiple subjects the relative recording date
differences between subjects can be kept by using the same number
of `daysback` for all subject anonymizations. `daysback` should be
great enough to shift the date prior to 1925 to conform with BIDS
anonymization rules. Due to limitations of the EDF/BDF format, the
year of the anonymized date will always be set to 1985 in the
'startdate' field of the file. The correctly-shifted year will be
written to the 'local recording identification' region of the
file header, which may not be parsed by all EDF/EDF+/BDF reader
softwares.
`keep_his` : bool
By default (False), all subject information next to the recording
date will be overwritten as well. If True, keep subject information
apart from the recording date. Participant names and birthdates
will always be anonymized if present, regardless of this setting.
See Also
--------
mne.io.anonymize_info
copyfile_brainvision
copyfile_bti
copyfile_ctf
copyfile_eeglab
copyfile_kit
"""
# Ensure source & destination extensions are the same
fname_src, ext_src = _parse_ext(src)
fname_dest, ext_dest = _parse_ext(dest)
if ext_src != ext_dest:
raise ValueError(f'Need to move data with same extension, '
f' but got "{ext_src}" and "{ext_dest}"')
# Copy data prior to any anonymization
sh.copyfile(src, dest)
# Anonymize EDF/BDF data, if requested
if anonymize is not None:
if ext_src == '.bdf':
raw = read_raw_bdf(dest, preload=False, verbose=0)
elif ext_src == '.edf':
raw = read_raw_edf(dest, preload=False, verbose=0)
else:
raise ValueError('Unsupported file type ({0})'.format(ext_src))
# Get subject info, recording info, and recording date
with open(dest, 'rb') as f:
f.seek(8) # id_info field starts 8 bytes in
id_info = f.read(80).decode('ascii').rstrip()
rec_info = f.read(80).decode('ascii').rstrip()
# Parse metadata from file
if len(id_info) == 0 or len(id_info.split(' ')) != 4:
id_info = "X X X X"
if len(rec_info) == 0 or len(rec_info.split(' ')) != 5:
rec_info = "Startdate X X X X"
pid, sex, birthdate, name = id_info.split(' ')
start_date, admin_code, tech, equip = rec_info.split(' ')[1:5]
# Try to anonymize the recording date
daysback, keep_his = _check_anonymize(anonymize, raw, '.edf')
anonymize_info(raw.info, daysback=daysback, keep_his=keep_his)
start_date = '01-JAN-1985'
meas_date = '01.01.85'
# Anonymize ID info and write to file
if keep_his:
# Always remove participant birthdate and name to be safe
id_info = [pid, sex, "X", "X"]
rec_info = ["Startdate", start_date, admin_code, tech, equip]
else:
id_info = ["0", "X", "X", "X"]
rec_info = ["Startdate", start_date, "X",
"mne-bids_anonymize", "X"]
with open(dest, 'r+b') as f:
f.seek(8) # id_info field starts 8 bytes in
f.write(bytes(" ".join(id_info).ljust(80), 'ascii'))
f.write(bytes(" ".join(rec_info).ljust(80), 'ascii'))
f.write(bytes(meas_date, 'ascii'))
from config import fname, task_name, montage, parser, LoggingFormat
###############################################################################
# Start processing step
# Handle command line arguments
args = parser.parse_args()
subject = args.subject
print(LoggingFormat.PURPLE + LoggingFormat.BOLD +
'Converting to BIDS: Subject %s' % subject + LoggingFormat.END)
###############################################################################
# 1) Import the data
input_file = fname.source(source_type='eeg', subject=subject)
raw = read_raw_bdf(input_file, preload=False)
# sampling rate
sfreq = raw.info['sfreq']
# channels in dataset
channels = raw.info['ch_names']
###############################################################################
# 2) Modify dataset info
# identify channel types based on matching names in montage
types = []
for channel in channels:
if channel in montage.ch_names:
types.append('eeg')
elif channel.startswith('EOG') | channel.startswith('EXG'):
types.append('eog')
def importbdf(bdfname,
nchans=34,
refchans=['EXG1', 'EXG2'],
hptsname=None,
mask=255,
extrachans=[],
exclude=[],
verbose=None):
"""Wrapper around mne-python to import BDF files
Parameters
----------
bdfname - Name of the biosemi .bdf filename with full path
nchans - Number of EEG channels (including references)
(Optional) By default, 34 (32 + 2 references)
refchans - list of strings with reference channel names
(Optional) By default ['EXG1','EXG2'].
Use None for average reference.
hptsname - Name of the electrode position file in .hpts format with path
(Optional) By default a 32 channel Biosemi layout is used. If
the nchans is >= 64 and < 96, a 64 channel Biosemi layout is
used. If nchans >= 96, a 96 channel biosemi layout is used.
Formats other than .hpts will also likely work, but behavior
may vary.
mask - Integer mask to use for trigger channel (Default is 255).
extrachans - Additional channels other than EEG and EXG that may be in the
bdf file. These will be marked as MISC in mne-python.
Specify as list of names.
exclude - List of channel names to exclude from importing
verbose - bool, str, int, or None (Optional)
The verbosity of messages to print. If a str, it can be either DEBUG,
INFO, WARNING, ERROR, or CRITICAL.
Returns
-------
raw - MNE raw data object of rereferences and preloaded data
eves - Event list (3 column array as required by mne.Epochs)
Requires
--------
mne-python module > release 0.7
"""
# Default HPTS file
if (hptsname is None):
anlffr_root = os.path.dirname(sys.modules['anlffr'].__file__)
if nchans >= 64 and nchans < 96:
logger.info('Number of channels is greater than 64.'
' Hence loading a 64 channel montage.')
hptspath = os.path.join(anlffr_root, 'helper/sysfiles/')
hptsname = 'biosemi64.hpts'
montage = read_dig_hpts(hptspath + hptsname)
misc = ['EXG3', 'EXG4', 'EXG5', 'EXG6', 'EXG7', 'EXG8']
else:
if nchans >= 96:
logger.info('Number of channels is greater than 96.'
' Hence loading a 96 channel montage.')
hptspath = os.path.join(anlffr_root, 'helper/sysfiles/')
hptsname = 'biosemi96.hpts'
montage = read_dig_hpts(hptspath + hptsname)
misc = ['EXG3', 'EXG4', 'EXG5', 'EXG6', 'EXG7', 'EXG8']
else:
if nchans == 2:
logger.info('Number of channels is 2.'
'Guessing ABR montage or saccades.')
montage = None
misc = []
else:
logger.info('Loading a default 32 channel montage.')
hptspath = os.path.join(anlffr_root, 'helper/sysfiles/')
hptsname = 'biosemi32.hpts'
montage = read_dig_hpts(hptspath + hptsname)
misc = ['EXG3', 'EXG4', 'EXG5', 'EXG6', 'EXG7', 'EXG8']
else:
montage = read_dig_hpts(hptsname) # User-supplied
misc = ['EXG3', 'EXG4', 'EXG5', 'EXG6', 'EXG7', 'EXG8']
misc += extrachans
raw = read_raw_bdf(bdfname,
preload=True,
misc=misc,
exclude=exclude,
stim_channel='auto')
raw.set_montage(montage, on_missing='warn')
# Rereference
if refchans is not None:
sys.stdout.write('Re-referencing data to: ' + ' '.join(refchans))
(raw, ref_data) = set_eeg_reference(raw, refchans, copy=False)
raw.info['bads'] += refchans
else:
# Add average reference operator for possible use later
ave_ref_operator = make_eeg_average_ref_proj(raw.info, activate=False)
raw = raw.add_proj(ave_ref_operator)
eves = find_events(raw, shortest_event=1, mask=mask)
return (raw, eves)
montage = make_standard_montage(kind='standard_1020')
# channels to be exclude from import
exclude = ['EXG5', 'EXG6', 'EXG7', 'EXG8']
# ========================================================================
# ------------ loop through files and extract blocks --------------------
for file in files:
# --- 1) set up paths and file names -----------------------
filepath, filename = op.split(file)
# subject in question
subj = re.findall(r'\d+', filename)[0].rjust(3, '0')
print(subj)
# --- 2) import the data -----------------------------------
raw = read_raw_bdf(file, preload=True, exclude=exclude)
# check data and remove channel with low (i.e., near zero variance)
flats = np.where(np.std(raw.get_data(), axis=1) * 1e6 < 10.)[0]
# names
flats = [raw.ch_names[ch] for ch in flats]
# print summary
print('Following channels were dropped as variance ~ 0:', flats)
# remove them from data set
raw.drop_channels(flats)
# --- 3) modify data set information ----------------------
# keep the sampling rate
sfreq = raw.info['sfreq']
# and date of measurement
date_of_record = raw.annotations.orig_time
clip = None
# === LOOP THROUGH FILES AND RUN PRE-PROCESSING ==========================
for file in files:
# --- 1) Set up paths and file names -----------------------
filepath, filename = os.path.split(file)
filename, ext = os.path.splitext(filename)
subj = re.findall(r'\d+', filename)[0]
# --- 2) READ IN THE DATA ----------------------------------
# Set EEG arrangement
montage = read_montage(kind='biosemi64')
# Import raw data
raw = read_raw_bdf(file, montage=montage, preload=True, exclude=exclude)
# Note the sampling rate of recording
sfreq = raw.info['sfreq']
# all channels in raw
chans = raw.info['ch_names']
# channels in montage
montage_chans = montage.ch_names
# nr of eeg channels
n_eeg = len([chan for chan in chans if chan in montage_chans])
# channel types
types = []
for chan in chans:
if chan in montage_chans:
types.append('eeg')
elif re.match('EOG|EXG', chan):
