def _read_eeg(self, path_eeg_file, channel_name):
"""Loads signal from 'path_eeg_file', does filtering and resampling."""
proper_name_dict = {
KEY_EEG_FRONTAL: 'EEG F3-CLE', # Frontal
KEY_EEG_CENTRAL: 'EEG C3-CLE', # Central
KEY_EEG_OCCIPITAL: 'EEG O1-CLE', # Occipital
KEY_EMG: 'EMG Chin', # EMG
KEY_EOG_LEFT: 'EOG Left Horiz', # Left eye
KEY_EOG_RIGHT: 'EOG Right Horiz', # Right eye
}
checks.check_valid_value(channel_name, 'channel_name',
list(proper_name_dict.keys()))
with pyedflib.EdfReader(path_eeg_file) as file:
channel_names = file.getSignalLabels()
channel_to_extract = channel_names.index(
proper_name_dict[channel_name])
signal = file.readSignal(channel_to_extract)
fs_old = file.samplefrequency(channel_to_extract)
# Check
print('Channel extracted: %s' % file.getLabel(channel_to_extract))
fs_old_round = int(np.round(fs_old))
# Transform the original fs frequency with decimals to rounded version
signal = utils.resample_signal_linear(signal,
fs_old=fs_old,
fs_new=fs_old_round)
signal = signal.astype(np.float32)
return signal
def _read_eeg(self, path_eeg_file, channel_name):
"""Loads signal from 'path_eeg_file', does filtering and resampling."""
proper_name_dict = {
KEY_EEG_FP1: 'EEG Fp1-CLE', # Frontal
KEY_EEG_FP2: 'EEG Fp2-CLE', # Central
KEY_ECG: 'ECG I', # ECG
}
checks.check_valid_value(channel_name, 'channel_name',
list(proper_name_dict.keys()))
with pyedflib.EdfReader(path_eeg_file) as file:
channel_names = file.getSignalLabels()
channel_to_extract = channel_names.index(
proper_name_dict[channel_name])
signal = file.readSignal(channel_to_extract)
fs_old = file.samplefrequency(channel_to_extract)
# Check
print('Channel extracted: %s' % file.getLabel(channel_to_extract))
fs_old_round = int(np.round(fs_old))
# Transform the original fs frequency with decimals to rounded version
signal = utils.resample_signal_linear(signal,
fs_old=fs_old,
fs_new=fs_old_round)
signal = signal.astype(np.float32)
return signal
def get_subject_hypnogram(self, subject_id, verbose=False):
"""Returns the hypogram of this subject."""
checks.check_valid_value(subject_id, 'ID', self.all_ids)
ind_dict = self.data[subject_id]
hypno = ind_dict[KEY_HYPNOGRAM]
if verbose:
print('Getting Hypnogram of ID %s' % subject_id)
return hypno
def get_subject_signal(self, subject_id, verbose=False):
checks.check_valid_value(subject_id, 'ID', self.all_ids)
ind_dict = self.data[subject_id]
signal = np.stack(
[ind_dict[KEY_EEG_FP1], ind_dict[KEY_EEG_FP2], ind_dict[KEY_ECG]],
axis=1)
if verbose:
print('Getting signal of ID %s' % subject_id)
return signal
def get_subject_data(self,
subject_id,
output_fs=100,
border_duration=0,
ignore_unknown=True,
verbose=False):
"""
Returns segments of the signals for the given id and their
corresponding sleep stage.
"""
checks.check_valid_value(subject_id, 'ID', self.all_ids)
if output_fs > self.fs:
raise ValueError('output fs cannot be greater than original fs')
signal = self.get_subject_signal(subject_id)
hypnogram = self.get_subject_hypnogram(subject_id)
# Resample if needed
if output_fs < self.fs:
if verbose:
print('Resampling from %d Hz to %d Hz' % (self.fs, output_fs))
resampled_list = []
for chn in range(signal.shape[1]):
this_signal = signal[:, chn]
this_signal = utils.resample_signal(this_signal, self.fs,
output_fs)
resampled_list.append(this_signal)
signal = np.stack(resampled_list, axis=1)
# Extract segments
segments_list = []
stages_list = []
page_size = int(self.page_duration * output_fs)
border_size = int(border_duration * output_fs)
for i, stage in enumerate(hypnogram):
if i == 0:
if verbose:
"Skipping first segment."
elif stage == "?" and ignore_unknown:
if verbose:
"Dropped unknown stage '?'."
else:
sample_start = i * page_size - border_size
sample_end = (i + 1) * page_size + border_size
this_segment = signal[sample_start:sample_end, :]
segments_list.append(this_segment)
stages_list.append(stage)
x = np.stack(segments_list, axis=0)
y = np.stack(stages_list, axis=0)
if verbose:
print('S%02d with %d segments' % (subject_id, x.shape[0]))
return x, y
def get_subject_signal(self, subject_id, verbose=False):
checks.check_valid_value(subject_id, 'ID', self.all_ids)
ind_dict = self.data[subject_id]
signal = np.stack([
ind_dict[KEY_EEG_FRONTAL], ind_dict[KEY_EEG_CENTRAL],
ind_dict[KEY_EEG_OCCIPITAL], ind_dict[KEY_EOG_LEFT],
ind_dict[KEY_EOG_RIGHT], ind_dict[KEY_EMG]
],
axis=1)
if verbose:
print('Getting signal of ID %s' % subject_id)
return signal