def _fetch_hbnssi_participants(data_dir, url, verbose):
"""
Helper function to fetch_hbnssi.
This function helps in downloading and loading participants data from .tsv
uploaded on Open Science Framework(OSF).
Parameters
----------
data_dir: str
Path of the data directory. Used to force data storage in a specified
location. If None is given, data are stored in home directory.
url: str, optional
Override download URL. Used for test only(or if you setup a mirror of
the data). Default: None
verbose: int
Defines the level of verbosity of the output.
Returns
-------
participants: numpy.ndarray
Contains data of each subject age, gender, handedness.
"""
if url is None:
url = 'https://osf.io/wtvh3/download'
files = [('participants.csv', url, {'move': 'participants.csv'})]
path_to_participants = _fetch_files(data_dir, files, verbose=verbose)[0]
# Load path to participants
dtype = [('sid', 'U12'), ('age', '<f8'), ('Gender', 'U4'),
('Handedness', 'U4')]
names = ['sid', 'age', 'gender', 'handedness']
participants = csv_to_array(path_to_participants, skip_header=True,
dtype=dtype, names=names)
return participants
def _fetch_ibc_surf_masks(data_dir, url, resume, verbose):
"""Helper function to fetch_ibc.
This function helps in downloading brain and ROI masks for use with IBC
functional alignment and inter-subject decoding.
The files are downloaded from Open Science Framework (OSF).
Parameters
----------
data_dir: str
Path of the data directory. Used to force data storage in a specified
location. If None is given, data are stored in home directory.
url: str, optional
Override download URL. Used for test only (or if you setup a mirror of
the data). Default: None
resume: bool, optional (default True)
Whether to resume download of a partly-downloaded file.
verbose: int
Defines the level of verbosity of the output.
Returns
-------
derivatives_dir: str
Path on disk to the IBC masks data directory.
"""
if url is None:
# Download from the relevant OSF project, using hashes generated
# from the OSF API. Note the trailing slash. For more info, see:
# https://gist.github.com/emdupre/3cb4d564511d495ea6bf89c6a577da74
url = 'https://osf.io/download/{}/'
# The gzip contains unique download keys per Nifti file and CSV
# pre-extracted from OSF. Required for downloading files.
package_directory = os.path.dirname(os.path.abspath(__file__))
dtype = [('filename', 'U52'), ('uid', 'U24')]
names = ['filename', 'uid']
# csv file contains download information
osf_data = csv_to_array(os.path.join(package_directory,
"ibc_surf_masks.csv"),
skip_header=True,
dtype=dtype,
names=names)
derivatives_dir = Path(data_dir, 'masks')
masks = []
for this_osf_id in osf_data:
# Download mask
mask_url = url.format(this_osf_id['uid'])
mask_target = Path(derivatives_dir, this_osf_id['filename'])
mask_file = [(mask_target, mask_url, {'move': mask_target})]
path_to_mask = _fetch_files(data_dir, mask_file, verbose=verbose)[0]
masks.append(path_to_mask)
return derivatives_dir
def test_csv_to_array():
# Create a phony CSV file
with tempfile.NamedTemporaryFile(suffix='.csv', mode='wt') as csv_file:
csv_file.write('1.,2.,3.,4.,5.\n')
csv_file.flush()
assert_true(np.allclose(csv_to_array(csv_file.name),
np.asarray([1., 2., 3., 4., 5.])))
assert_raises(TypeError, csv_to_array, csv_file.name, delimiters='?!')
def test_csv_to_array():
# Create a phony CSV file
with tempfile.NamedTemporaryFile(suffix='.csv', mode='wt') as csv_file:
csv_file.write('1.,2.,3.,4.,5.\n')
csv_file.flush()
assert_true(
np.allclose(csv_to_array(csv_file.name),
np.asarray([1., 2., 3., 4., 5.])))
assert_raises(TypeError, csv_to_array, csv_file.name, delimiters='?!')
def test_csv_to_array():
# Create a phony CSV file
filename = tempfile.mktemp(suffix='.csv')
try:
with open(filename, mode='wt') as fp:
fp.write('1.,2.,3.,4.,5.\n')
assert_true(np.allclose(csv_to_array(filename),
np.asarray([1., 2., 3., 4., 5.])))
assert_raises(TypeError, csv_to_array, filename, delimiters='?!')
finally:
os.remove(filename)
def test_csv_to_array():
# Create a phony CSV file
filename = tempfile.mktemp(suffix='.csv')
try:
with open(filename, mode='wt') as fp:
fp.write('1.,2.,3.,4.,5.\n')
assert np.allclose(csv_to_array(filename),
np.asarray([1., 2., 3., 4., 5.]))
pytest.raises(TypeError, csv_to_array, filename, delimiters='?!')
finally:
os.remove(filename)
def transform(self, imgs, confounds=None):
"""Signal extraction from regions learned on the images.
Parameters
----------
imgs : Nifti like images, list
confounds : csv file or array-like, optional
Contains signals like motion, high variance confounds from
white matter, csf. This is passed to signal.clean
"""
self._check_fitted()
confounds1_ = None
confounds2_ = None
SUBJECTS_TIMESERIES = dict()
# Getting Masker to transform fMRI images in Nifti to timeseries signals
# based on atlas learning
if self.masker is None:
raise ValueError("Could not find masker attribute. Masker is missing")
if not isinstance(imgs, collections.Iterable) or \
isinstance(imgs, _basestring):
imgs = [imgs, ]
if self.compute_confounds not in ['compcor_5', 'compcor_10', None]:
warnings.warn("Given invalid input compute_confounds={0}. Given "
"input is diverting to compute_confounds=None"
.format(self.compute_confounds))
self.compute_confounds = None
if self.compute_not_mask_confounds not in ['compcor_5', 'compcor_10', None]:
warnings.warn("Invalid input type of 'compute_not_mask_confounds'={0}"
"is provided. Switching to None"
.format(self.compute_not_mask_confounds))
self.compute_not_mask_confounds = None
if confounds is None and self.compute_confounds is None and \
self.compute_not_mask_confounds is None:
confounds = [None] * len(imgs)
if self.compute_confounds is None and \
self.compute_not_mask_confounds is None:
confounds_ = None
if self.compute_confounds is not None:
if self.compute_confounds_mask_img is None:
raise ValueError("Parameter `compute_confounds` is given as {0} but "
"`compute_confounds_mask_img` is given as None. "
.format(self.compute_confounds))
if not isinstance(self.compute_confounds_mask_img, collections.Iterable) or \
isinstance(self.compute_confounds_mask_img, _basestring):
mask_imgs1 = [self.compute_confounds_mask_img]
if self.compute_confounds == 'compcor_5':
n_confounds = 5
elif self.compute_confounds == 'compcor_10':
n_confounds = 10
confounds1_ = []
for mask_img1 in mask_imgs1:
conf1_ = self.masker.memory.cache(compute_confounds)(
imgs, mask_img1, n_confounds=n_confounds)
confounds1_.append(conf1_)
confounds1_ = np.hstack(confounds1_)
if self.compute_not_mask_confounds is not None:
if self.compute_confounds_non_mask_img is None:
raise ValueError("Parameter `compute_not_mask_confounds` is given as {0}"
" but `compute_confounds_non_mask_img` is given as None."
.format(self.compute_not_mask_confounds))
if not isinstance(self.compute_confounds_non_mask_img, collections.Iterable) or \
isinstance(self.compute_confounds_non_mask_img, _basestring):
mask_imgs2 = [self.compute_confounds_non_mask_img]
if self.compute_not_mask_confounds == 'compcor_5':
n_confounds = 5
elif self.compute_not_mask_confounds == 'compcor_10':
n_confounds = 10
confounds2_ = []
for mask_img2 in mask_imgs2:
conf2_ = self.masker.memory.cache(compute_confounds)(
imgs, mask_img2, n_confounds=n_confounds, compute_not_mask=True)
confounds2_.append(conf2_)
if confounds1_ is not None and confounds2_ is not None:
confounds_ = np.hstack((confounds1_, confounds2_))
if confounds1_ is not None and confounds2_ is None:
confounds_ = confounds1_
if confounds1_ is None and confounds2_ is not None:
confounds_ = confounds2_
if confounds_ is not None:
if confounds is not None:
confs_ = []
if isinstance(confounds, collections.Iterable):
for conf_gen, conf_given in zip(confounds_, confounds):
if isinstance(conf_given, _basestring):
conf_given = csv_to_array(conf_given)
stack_confounds = np.hstack((conf_gen, conf_given))
confs_.append(stack_confounds)
confounds = confs_
else:
confounds = confounds_
if confounds is not None and isinstance(confounds, collections.Iterable):
if len(confounds) != len(imgs):
raise ValueError("Number of confounds given doesnot match with "
"the given number of subjects. Add missing "
"confound in a list.")
for model in self.models_:
subjects_timeseries = []
if self.verbose > 0:
print("[Timeseries Extraction] {0} atlas image is selected"
.format(model))
atlas_img = self.rois_[model]
masker = check_embedded_atlas_masker(self.masker, atlas_type='auto',
img=atlas_img, t_r=2.53,
low_pass=0.1, high_pass=0.01)
subjects_timeseries = Parallel(n_jobs=self.masker.n_jobs)(
delayed(_atlas_masker_extraction)(img, masker, confound)
for img, confound in zip(imgs, confounds))
if subjects_timeseries is not None:
SUBJECTS_TIMESERIES[model] = subjects_timeseries
else:
warnings.warn("Timeseries signals extraction are found empty "
"for model:{0}".format(model))
self.subjects_timeseries_ = SUBJECTS_TIMESERIES
if self.connectome_convert:
if self.connectome_measure is not None:
if isinstance(self.connectome_measure, collections.Iterable):
catalog = self.connectome_measure
else:
if isinstance(self.connectome_measure, _basestring):
catalog = [self.connectome_measure, ]
else:
warnings.warn("Given connectome_convert=True but connectome "
"are given as None. Taking connectome measure "
"kind='correlation'", stacklevel=2)
catalog = ['correlation']
self.connectome_measures_ = catalog
connectivities = self._connectome_converter(
catalog=self.connectome_measures_,
confounds=self.connectome_confounds)
return self
def _fetch_hbnssi_functional(participants, data_dir, url, resume, verbose):
"""Helper function to fetch_development_fmri.
This function helps in downloading functional MRI data in Nifti
and its confound corresponding to each subject.
The files are downloaded from Open Science Framework (OSF).
Parameters
----------
participants : numpy.ndarray
Should contain column participant_id which represents subjects id. The
number of files are fetched based on ids in this column.
data_dir: str
Path of the data directory. Used to force data storage in a specified
location. If None is given, data are stored in home directory.
url: str, optional
Override download URL. Used for test only (or if you setup a mirror of
the data). Default: None
resume: bool, optional (default True)
Whether to resume download of a partly-downloaded file.
verbose: int
Defines the level of verbosity of the output.
Returns
-------
func: list of str (Nifti files)
Paths to functional MRI data (4D) for each subject.
"""
if url is None:
# Download from the relevant OSF project, using hashes generated
# from the OSF API. Note the trailing slash. For more info, see:
# https://gist.github.com/emdupre/3cb4d564511d495ea6bf89c6a577da74
url = 'https://osf.io/download/{}/'
raiders = 'sub-{0}_task-RAIDERS_space-MNI152NLin2009cAsym_desc-postproc_bold.nii.gz'
flanker = 'sub-{0}_task-FLANKERTASK_space-MNI152NLin2009cAsym_desc-postproc_bold.nii.gz'
conditions = 'sub-{0}_labels.csv'
runs = 'sub-{0}_runs.csv'
# The gzip contains unique download keys per Nifti file and confound
# pre-extracted from OSF. Required for downloading files.
package_directory = os.path.dirname(os.path.abspath(__file__))
dtype = [('sid', 'U12'), ('raiders', 'U24'), ('flanker', 'U24'),
('condition', 'U24'), ('run', 'U24')]
names = ['sid', 'raiders', 'flanker', 'condition', 'run']
# csv file contains download information related to OpenScience(osf)
osf_data = csv_to_array(os.path.join(package_directory, "hbnssi.csv"),
skip_header=True, dtype=dtype, names=names)
derivatives_dir = Path(data_dir, 'derivatives')
align, decode, labels, sessions = [], [], [], []
for sid in participants['sid']:
this_osf_id = osf_data[osf_data['sid'] == sid]
# Download raiders
raiders_url = url.format(this_osf_id['raiders'][0])
raiders_target = Path(derivatives_dir, raiders.format(sid))
raiders_file = [(raiders_target,
raiders_url,
{'move': raiders_target})]
path_to_raiders = _fetch_files(data_dir, raiders_file,
verbose=verbose)[0]
align.append(path_to_raiders)
# Download flanker
flanker_url = url.format(this_osf_id['flanker'][0])
flanker_target = Path(derivatives_dir, flanker.format(sid))
flanker_file = [(flanker_target,
flanker_url,
{'move': flanker_target})]
path_to_flanker = _fetch_files(data_dir, flanker_file,
verbose=verbose)[0]
decode.append(path_to_flanker)
# Download condition labels
label_url = url.format(this_osf_id['condition'][0])
label_target = Path(derivatives_dir, conditions.format(sid))
label_file = [(label_target,
label_url,
{'move': label_target})]
path_to_labels = _fetch_files(data_dir, label_file,
verbose=verbose)[0]
labels.append(path_to_labels)
# Download session run numbers
session_url = url.format(this_osf_id['run'][0])
session_target = Path(derivatives_dir, runs.format(sid))
session_file = [(session_target,
session_url,
{'move': session_target})]
path_to_sessions = _fetch_files(data_dir, session_file,
verbose=verbose)[0]
sessions.append(path_to_sessions)
return derivatives_dir
def _fetch_rsvp_trial_surf(participants, data_dir, url, resume, verbose):
"""Helper function to fetch_ibc.
This function helps in downloading functional MRI data in Nifti format
and its corresponding CSVs each subject for functional alignment and
decoding.
The files are downloaded from Open Science Framework (OSF).
Parameters
----------
participants : numpy.ndarray
Should contain column participant_id which represents subjects id. The
number of files are fetched based on ids in this column.
data_dir: str
Path of the data directory. Used to force data storage in a specified
location. If None is given, data are stored in home directory.
url: str, optional
Override download URL. Used for test only (or if you setup a mirror of
the data). Default: None
resume: bool, optional (default True)
Whether to resume download of a partly-downloaded file.
verbose: int
Defines the level of verbosity of the output.
Returns
-------
func: list of str (Nifti files)
Paths to functional MRI data (4D) for each subject.
"""
if url is None:
# Download from the relevant OSF project, using hashes generated
# from the OSF API. Note the trailing slash. For more info, see:
# https://gist.github.com/emdupre/3cb4d564511d495ea6bf89c6a577da74
url = 'https://osf.io/download/{}/'
surf = '{0}_{1}.gii'
highres_surf = '{0}_{1}_fullres.gii'
conditions = '{0}_{1}_labels.csv'
runs = '{0}_{1}_runs.csv'
# The gzip contains unique download keys per Nifti file and CSV
# pre-extracted from OSF. Required for downloading files.
package_directory = os.path.dirname(os.path.abspath(__file__))
dtype = [('sid', 'U12'), ('hemi', 'U12'), ('surf', 'U24'),
('highres_surf', 'U24'), ('condition', 'U24'), ('run', 'U24')]
names = ['sid', 'hemi', 'surf', 'highres_surf', 'condition', 'run']
# csv file contains download information
osf_data = csv_to_array(os.path.join(package_directory,
"rsvp_trial_surf.csv"),
skip_header=True,
dtype=dtype,
names=names)
derivatives_dir = Path(data_dir, 'rsvp_trial', 'surf_derivatives')
surf_files, highres_surf_files, labels, sessions = [], [], [], []
for sid in participants['sid']:
hemi1, hemi2 = osf_data[osf_data['sid'] == sid]
for hemi in (hemi1, hemi2):
# Download surface for decoding
surf_url = url.format(hemi['surf'])
surf_target = Path(derivatives_dir, surf.format(sid, hemi['hemi']))
surf_file = [(surf_target, surf_url, {'move': surf_target})]
path_to_surf = _fetch_files(data_dir, surf_file,
verbose=verbose)[0]
surf_files.append(path_to_surf)
# Download surface
highres_surf_url = url.format(hemi['highres_surf'])
highres_surf_target = Path(derivatives_dir,
highres_surf.format(sid, hemi['hemi']))
highres_surf_file = [(highres_surf_target, highres_surf_url, {
'move': highres_surf_target
})]
path_to_highres_surf = _fetch_files(data_dir,
highres_surf_file,
verbose=verbose)[0]
highres_surf_files.append(path_to_highres_surf)
# Download condition labels
label_url = url.format(hemi['condition'])
label_target = Path(derivatives_dir,
conditions.format(sid, hemi['hemi']))
label_file = [(label_target, label_url, {'move': label_target})]
path_to_labels = _fetch_files(data_dir,
label_file,
verbose=verbose)[0]
labels.append(path_to_labels)
# Download session run numbers
session_url = url.format(hemi['run'])
session_target = Path(derivatives_dir,
runs.format(sid, hemi['hemi']))
session_file = [(session_target, session_url, {
'move': session_target
})]
path_to_sessions = _fetch_files(data_dir,
session_file,
verbose=verbose)[0]
sessions.append(path_to_sessions)
# create out_dir
Path(data_dir, "rsvp_trial", "decoding").mkdir(parents=True, exist_ok=True)
# create mask_cache
Path(data_dir, "rsvp_trial", "mask_cache").mkdir(parents=True,
exist_ok=True)
return derivatives_dir
def _fetch_ibc_alignment(participants, data_dir, url, resume, verbose):
"""Helper function to fetch_ibc.
This function helps in downloading IBC functional MRI data in Nifti format
and its corresponding CSVs each subject for functional alignment and
decoding.
The files are downloaded from Open Science Framework (OSF).
Parameters
----------
participants : numpy.ndarray
Should contain column participant_id which represents subjects id. The
number of files are fetched based on ids in this column.
data_dir: str
Path of the data directory. Used to force data storage in a specified
location. If None is given, data are stored in home directory.
url: str, optional
Override download URL. Used for test only (or if you setup a mirror of
the data). Default: None
resume: bool, optional (default True)
Whether to resume download of a partly-downloaded file.
verbose: int
Defines the level of verbosity of the output.
Returns
-------
derivatives_dir: str
Path on disk to the IBC alignment data directory.
"""
if url is None:
# Download from the relevant OSF project, using hashes generated
# from the OSF API. Note the trailing slash. For more info, see:
# https://gist.github.com/emdupre/3cb4d564511d495ea6bf89c6a577da74
url = 'https://osf.io/download/{}/'
alignment = '{0}_53_contrasts.nii.gz'
# The gzip contains unique download keys per Nifti file and CSV
# pre-extracted from OSF. Required for downloading files.
package_directory = os.path.dirname(os.path.abspath(__file__))
dtype = [('sid', 'U12'), ('alignment', 'U24')]
names = ['sid', 'alignment']
# csv file contains download information
osf_data = csv_to_array(os.path.join(package_directory,
"ibc_alignment.csv"),
skip_header=True,
dtype=dtype,
names=names)
derivatives_dir = Path(data_dir, 'alignment')
align = []
for sid in participants['sid']:
this_osf_id = osf_data[osf_data['sid'] == sid]
# Download alignment
alignment_url = url.format(this_osf_id['alignment'][0])
alignment_target = Path(derivatives_dir, alignment.format(sid))
alignment_file = [(alignment_target, alignment_url, {
'move': alignment_target
})]
path_to_alignment = _fetch_files(data_dir,
alignment_file,
verbose=verbose)[0]
align.append(path_to_alignment)
return derivatives_dir
