def high_variance_confounds(file_names):
""" Return confounds time series extracted from voxels with high
variance.
Parameters
===========
filenames: list of 3d nifti or analyze files
The ras (non preprocessed) dataset
Notes
======
This method is related to what has been published in the
literature as 'CompCorr' (Behzadi NeuroImage 2007).
"""
mask = np.ones(nibabel.load(filenames[0]).get_data().shape,
dtype=np.bool)
series, _ = mask_tools.series_from_mask(
filenames,
mask)
for serie in series:
serie[:] = signal.detrend(serie)
# Retrieve the 1% high variance voxels
var = np.mean(series**2, axis=-1)
var_thr = stats.scoreatpercentile(var, 99)
series = series[var > var_thr]
u, s, v = linalg.svd(series, full_matrices=False)
v = v[:10]
return v
def high_variance_confounds(file_names):
""" Return confounds time series extracted from voxels with high
variance.
Parameters
===========
filenames: list of 3d nifti or analyze files
The ras (non preprocessed) dataset
Notes
======
This method is related to what has been published in the
literature as 'CompCorr' (Behzadi NeuroImage 2007).
"""
mask = np.ones(nibabel.load(filenames[0]).get_data().shape, dtype=np.bool)
series, _ = mask_tools.series_from_mask(filenames, mask)
for serie in series:
serie[:] = signal.detrend(serie)
# Retrieve the 1% high variance voxels
var = np.mean(series**2, axis=-1)
var_thr = stats.scoreatpercentile(var, 99)
series = series[var > var_thr]
u, s, v = linalg.svd(series, full_matrices=False)
v = v[:10]
return v
def session_pca(raw_filenames, mask, smooth=False,
two_levels=False, n_first_components=None):
""" Do the preprocessing and calculate the PCA components for a
single session.
Parameters
-----------
mask: 3d ndarray
3D mask array: true where a voxel should be used.
smooth: False or float, optional
If smooth is not False, it gives the size, in voxel of the
spatial smoothing to apply to the signal.
two_levels: boolean, optional
If two_levels is True, the filenames are a list of filenames
corresponding to various sessions, and a multi-level model is
applied based on a first CCA.
n_first_components: integer, optional
The number of components to retain at the first level. This
must be specified if two_levels is True.
"""
# Data preprocessing and loading.
if not two_levels:
series, header = mask_utils.series_from_mask(raw_filenames,
mask, smooth=smooth)
# XXX: this should go in series_from_mask
series -= series.mean(axis=-1)[:, np.newaxis]
std = series.std(axis=-1)
std[std==0] = 1
series /= std[:, np.newaxis]
del std
# PCA
components, loadings, _ = linalg.svd(series, full_matrices=False)
else:
if n_first_components is None:
raise ValueError('If two_levels is True, n_first_components '
'must be specified')
components = list()
for session_files in raw_filenames:
these_components, _, header = \
session_pca(session_files, mask, smooth=smooth)
components.append(these_components[:,
:n_first_components])
del these_components
# CCA
components = np.hstack(components)
components, loadings, _ = linalg.svd(components,
full_matrices=False)
return components, loadings, header
gm_img = as_volume_img(gm_file)
epi_ref_img = as_volume_img(epi_ref_file)
# Resample tissue mask to grid of epi
gm_img = gm_img.resampled_to_img(epi_ref_img)
# Extract tissue mask
gm = gm_img.get_data()
# Normalize the mask to [0,1]
gm -= gm.min()
gm /= gm.max()
# Threshold tissue mask
gm_mask = (gm > .5)
# Find largest connected component
gm_mask = mask_utils.largest_cc(gm_mask)
# Extract graymatter voxel timecourses
time_series_gm, header_gm = mask_utils.series_from_mask(epi_files, gm_mask)
time_series_gm = preprocessing.standardize(time_series_gm).T
n_tpts = time_series_gm.shape[0]
# Load motion regressors
motion_regressor = np.loadtxt(os.path.join(BASE_DIR, "fmri", "rp_fga070108233-0004-00002-000002-01.txt"))
motion_regressor = preprocessing.standardize(motion_regressor)
# Bandpass filter to remove DC offset and low frequency drifts
beta, _, _, _ = linalg.lstsq(motion_regressor,time_series_gm)
time_series_gm -= np.dot(motion_regressor,beta)
f_cut = np.array([0.01, 0.1])
tr = 2.4
samp_freq = 1 / tr
w_cut = f_cut * 2 / samp_freq
b, a = signal.butter(5, w_cut, btype='bandpass', analog=0, output='ba')
for tc in time_series_gm.T:
tc[:] = signal.filtfilt(b, a, tc) # Modify timer_series_gm in place