def read_edf_file(filename):
f = EdfReader(filename)
chan = 0
data = f.readSignal(chan) / 1000.
sample_freq = f.getSampleFrequency(chan)
f._close()
return data, sample_freq
def _load_file(self, input_file, return_missing=False):
"""
Parameters
----------
input_file : str
path of the file which to load
return_missing : bool
if true, returns triple (x, y, missing)
where missing is true if input_file does
contain one or more of the requested channels.
Returns
-------
tuple (
x : {'channelname': {'sampling_frequency': 100, 'data': np.array}, ... }
y : {'channelname': {'sampling_frequency': 100, 'data': np.array}, ... }
)
"""
reader = EdfReader(input_file)
channel_names = reader.getSignalLabels()
channel_names_dict = {
channel_names[i]: i
for i in range(len(channel_names))
}
x_channels_to_process = set(channel_names).intersection(
set(self.x_channels))
y_channels_to_process = set(channel_names).intersection(
set(self.y_channels))
x_not_present = set(self.x_channels).difference(
set(self.x_channels).intersection(set(channel_names)))
y_not_present = set(self.y_channels).difference(
set(self.y_channels).intersection(set(channel_names)))
not_present = x_not_present.union(y_not_present)
#return this with x and y to detect missing channels.
missing = len(not_present) != 0
#report missing channels
if missing:
not_present = [i for i in not_present]
print(
f'File \'{input_file}\' does not have channels: {not_present}')
# Create data dictionaries
x_channel_data_dict = {}
for i in x_channels_to_process:
x_channel_data_dict[i] = {
"sampling_frequency":
reader.getSampleFrequency(channel_names_dict[i]),
"data":
reader.readSignal(channel_names_dict[i])
}
y_channel_data_dict = {}
for i in y_channels_to_process:
y_channel_data_dict[i] = {
"sampling_frequency":
reader.getSampleFrequency(channel_names_dict[i]),
"data":
reader.readSignal(channel_names_dict[i])
}
if return_missing:
return x_channel_data_dict, y_channel_data_dict, missing
else:
return x_channel_data_dict, y_channel_data_dict
class ChbEdfFile(object):
"""
Edf reader using pyedflib
"""
def __init__(self, filename, patient_id=None):
self._filename = filename
self._patient_id = patient_id
self._file = EdfReader(filename)
def get_filename(self):
return self._filename
def get_n_channels(self):
"""
Number of channels
"""
return len(self._file.getSampleFrequencies())
def get_n_data_points(self):
"""
Number of data points
"""
if len(self._file.getNSamples()) < 1:
raise ValueError("Number of channels is less than 1")
return self._file.getNSamples()[0]
def get_channel_names(self):
"""
Names of channels
"""
return self._file.getSignalLabels()
def get_channel_scalings(self):
"""
Channel scalings as an array
:return:
"""
out = np.zeros(self.get_n_channels())
for i in range(self.get_n_channels()):
out[i] = self._file.getPhysicalMaximum(
i) - self._file.getPhysicalMinimum(i)
return out
def get_file_duration(self):
"""
Returns the file duration in seconds
"""
return self._file.getFileDuration()
def get_sampling_rate(self):
"""
Get the frequency
"""
if len(self._file.getSampleFrequencies()) < 1:
raise ValueError("Number of channels is less than 1")
return self._file.getSampleFrequency(0)
def get_channel_data(self, channel_id):
"""
Get raw data for a single channel
"""
if channel_id >= self.get_n_channels() or channel_id < 0:
raise ValueError("Illegal channel id selected %d" % channel_id)
return self._file.readSignal(channel_id)
def get_data(self):
"""
Get raw data for all channels
"""
output_data = np.zeros(
(self.get_n_data_points(), self.get_n_channels()))
for i in range(self.get_n_channels()):
output_data[:, i] = self._file.readSignal(i)
return output_data
def get_start_datetime(self):
"""
Get the starting date and time
"""
return self._file.getStartdatetime()
def get_end_datetime(self):
return self._file.getStartdatetime() + datetime.timedelta(
seconds=self._file.getFileDuration())