def test_power_band():
"""Test computing mean PSD on specified band."""
fmin = 0.1
fmax = 40
band = [[8, 12], [13, 29]]
power_node = pe.Node(interface=Power(), name='welch_psd')
power_node.inputs.data_file = raw_fname
power_node.inputs.fmin = fmin
power_node.inputs.fmax = fmax
power_node.inputs.method = 'welch'
power_node.run()
power_band_node = pe.Node(interface=PowerBand(), name='power_band')
power_band_node.inputs.psds_file = power_node.result.outputs.psds_file
power_band_node.inputs.freq_bands = band
power_band_node.run()
mean_psd = np.load(power_band_node.result.outputs.mean_power_band_file)
raw = mne.io.read_raw_fif(raw_fname)
picks = mne.pick_types(raw.info, meg=True, ref_meg=False, exclude='bads')
assert len(picks) == mean_psd.shape[0]
assert len(band) == mean_psd.shape[1]
def create_pipeline_power(main_path, freq_bands, pipeline_name='power_pipeline', # noqa
fmin=0, fmax=300, method='welch',
is_epoched=False):
"""Power pipeline.
Wraps functions of MNE to compute PSD of epoch or raw data.
Parameters
----------
main_path : str
the main path of the pipeline
freq_bands : list
frequency bands
pipeline_name : str (default 'power')
name of the pipeline
fmin : float (default 0)
min frequency of interest
fmax : float (default 300)
max frequency of interest
method : str (default 'welch')
if 'welch' the power spectral density (PSD) is computed by Welch's
method; otherwise, if 'multitaper' the PSD is computed by
multitapers
is_epoched : bool (default False)
True if the input data are in epoch format (-epo.fif); False
if the input data are raw data (-raw.fif)
fif_file (inputnode): str
path to raw or epoched meg data in fif format
Returns
-------
pipeline : instance of Workflow
"""
pipeline = pe.Workflow(name=pipeline_name)
pipeline.base_dir = main_path
print(('*** main_path -> %s' % main_path + ' ***'))
# define the inputs of the pipeline
inputnode = pe.Node(IdentityInterface(fields=['fif_file']),
name='inputnode')
power_node = pe.Node(interface=Power(), name='power')
power_node.inputs.fmin = fmin
power_node.inputs.fmax = fmax
power_node.inputs.method = method
power_node.inputs.is_epoched = is_epoched
pipeline.connect(inputnode, 'fif_file', power_node, 'data_file')
power_band_node = pe.Node(interface=PowerBand(), name='power_band')
power_band_node.inputs.freq_bands = freq_bands
pipeline.connect(power_node, 'psds_file', power_band_node, 'psds_file')
return pipeline
def test_power_src_space():
"""Test computing mean PSD on specified band."""
fmin = 0.1
fmax = 40
# read source estimate and save in .npy
stc = mne.read_source_estimate(stc_fname)
data_fname = op.abspath('test_data.npy')
np.save(data_fname, stc.data)
power_node = pe.Node(interface=Power(), name='psd_src_space')
power_node.inputs.data_file = data_fname
power_node.inputs.fmin = fmin
power_node.inputs.fmax = fmax
power_node.inputs.sfreq = 300.
power_node.inputs.is_sensor_space = False
power_node.run()
def psd(fmin, fmax):
"""Compute power spectral density.
Lower and higher frequency bounds for computation
can be changed
Takes as input epochs in .fif format
EXAMPLE:
$ neuropycon pwr input ~/fif_epochs/*/*-epo.fif
"""
from ephypype.interfaces.mne.power import Power
# click.echo(list(fif_files))
power = pe.Node(interface=Power(), name='pwr')
power.inputs.fmin = fmin
power.inputs.fmax = fmax
power.inputs.method = 'welch'
return power
def test_power_multitaper():
"""Test computing and saving PSD."""
fmin = 0.1
fmax = 40
power_node = pe.Node(interface=Power(), name='multitaper_psd')
power_node.inputs.data_file = raw_fname
power_node.inputs.fmin = fmin
power_node.inputs.fmax = fmax
power_node.inputs.method = 'multitaper'
power_node.run()
psd_file = power_node.result.outputs.psds_file
power = np.load(psd_file)
raw = mne.io.read_raw_fif(raw_fname)
picks = mne.pick_types(raw.info, meg=True, ref_meg=False, exclude='bads')
assert len(picks) == power['psds'].shape[0]
assert len(power['freqs']) == power['psds'].shape[1]
def create_pipeline_power_src_space(main_path,
sfreq,
freq_bands,
pipeline_name='source_power',
fmin=0,
fmax=300,
nfft=256,
overlap=0,
is_epoched=False):
"""Power pipeline: wraps functions of MNE to compute source PSD.
Parameters
----------
main_path : str
the main path of the pipeline
sfreq : float
sampling frequency
freq_bands : list
frequency bands
pipeline_name : str (default 'source_power')
name of the pipeline
fmin : float (default 0)
min frequency of interest
fmax : float (default 300)
max frequency of interest
n_fft : int (default 256)
the length of FFT used
n_overlap : int
The number of points of overlap between segments
is_epoched : bool (default False)
True if the input data are in epoch format
Inputs (inputnode)
------------------
src_file : str
path to source reconstruction matrix (.npy format)
Returns
-------
pipeline : instance of Workflow
"""
pipeline = pe.Workflow(name=pipeline_name)
pipeline.base_dir = main_path
print('*** main_path -> %s' % main_path + ' ***')
# define the inputs of the pipeline
inputnode = pe.Node(IdentityInterface(fields=['data_file']),
name='inputnode')
power_node = pe.Node(interface=Power(), name='power')
power_node.inputs.sfreq = sfreq
power_node.inputs.fmin = fmin
power_node.inputs.fmax = fmax
power_node.inputs.nfft = nfft
power_node.inputs.overlap = overlap
power_node.inputs.is_epoched = is_epoched
power_node.inputs.is_sensor_space = False
pipeline.connect(inputnode, 'data_file', power_node, 'data_file')
power_band_node = pe.Node(interface=PowerBand(), name='power_band')
power_band_node.inputs.freq_bands = freq_bands
pipeline.connect(power_node, 'psds_file', power_band_node, 'psds_file')
return pipeline