def parse_z(path, conditions):
"""
Get images, sub, group and labels from a directory of images
Parameters
----------
path: str
path to the dir with the images. No files besides the images can be
in the directory. Images must be .nii or .nii.gz images
conditions: list of str
labels to search for in the target folders
Returns
-------
filenames: list
list of paths the z-maps of the directory
"""
# Get all (subject, session) pairs
task_list = ['audio1', 'audio2']
filenames = []
for task in task_list:
session_list = sorted(get_subject_session(task))
for sub, ses in session_list:
session_path = os.path.join(path, sub, ses)
if not os.path.exists(session_path):
print("Folder not found for {}, {}. Skipping...".format(sub, ses))
continue
print("Code reached for {}, {}".format(sub, ses))
# ses_glob = glob.glob(os.path.join(session_path, "res_stats_audio_*_*/stat_maps"))
ses_glob = glob.glob(os.path.join(session_path, "res_stats_audio_*_*/z_score_maps"))
for run_glob in ses_glob:
try:
file_list = [os.path.join(run_glob, file) for
file in os.listdir(run_glob) if
any(x + '-others.nii.gz' in file for x in conditions)]
except IndexError:
print("Found empty folder for {}, {}. Skipping...".format(sub, ses))
continue
filenames.extend(file_list)
return filenames
# Read the jobfile
list_subjects, params = _generate_preproc_pipeline(output_name)
# Preproc and Dump data
subject_data = do_subjects_preproc(output_name, report=True)
return subject_data
if __name__ == '__main__':
# correction of distortion_parameters
# custom solution, to be improved in the future
main_dir = '/neurospin/ibc/'
cache_dir = '/neurospin/tmp/ibc'
prepare_derivatives(main_dir)
do_topup = True
protocol = 'biological_motion'
subject_session = sorted(get_subject_session([protocol]))
# subject_session = [('sub-04', 'ses-28'), ('sub-05', 'ses-28'),]
if do_topup:
acq = None
if protocol in ['rs']:
acq = 'mb6'
elif protocol in ['mtt1', 'mtt2']:
acq = 'mb3'
apply_topup(main_dir, cache_dir, subject_session, acq=acq)
subject_data = []
jobfile = 'ini_files/IBC_preproc_%s.ini' % protocol
subject_data_ = Parallel(n_jobs=3)(
delayed(run_subject_preproc)(jobfile, subject, session)
for subject, session in subject_session)
subject_data = subject_data + subject_data_[0]
smooth=smooth,
mask_img=mask_img)
else:
first_level(subject,
compcorr=True,
smooth=smooth,
mask_img=mask_img)
fixed_effects_analysis(subject, mask_img=mask_img)
if __name__ == '__main__':
prepare_derivatives(IBC)
protocols = ['mtt1']
for protocol in protocols:
jobfile = 'ini_files/IBC_preproc_%s.ini' % protocol
subject_session = get_subject_session(protocol)
subject_session = [('sub-08', 'ses-11')]
Parallel(n_jobs=1)(delayed(run_subject_glm)(
jobfile, protocol, subject, session, lowres=True, smooth=5)
for (subject, session) in subject_session)
smooth = 5
for protocol in protocols:
jobfile = 'ini_files/IBC_preproc_%s.ini' % protocol
Parallel(n_jobs=4)(delayed(run_subject_glm)(jobfile, protocol, subject,
session, smooth)
for (subject, session) in subject_session)
smooth = None
for protocol in protocols:
t_r=t_r,
use_3mm=use_3mm,
glm_mode=glm_mode)
cv_score = decode_inter(imgs, names, mask, pipeline, conf_matrix=conf_matrix)
return cv_score
# Model
svc = LinearSVC(max_iter=10000)
feature_selection = SelectPercentile(f_classif, percentile=25)
pipeline = Pipeline([('anova', feature_selection), ('svc', svc)])
# Subjects
session_1 = sorted(get_subject_session([task_list[0]]))
session_2 = sorted(get_subject_session([task_list[1]]))
session_list = [(sub1, (ses1, ses2)) for
(sub1, ses1), (sub2, ses2) in zip(session_1, session_2)]
# Classification
scores = make_decode_inter(session_list,
conditions,
data_dir=data_dir,
write_dir=write_dir,
mask=mask_gm,
t_r=t_r,
pipeline=pipeline,
use_3mm=False,
glm_mode=glm_mode,
list_subjects, params = _generate_preproc_pipeline(output_name)
# Preproc and Dump data
subject_data = do_subjects_preproc(output_name, report=True)
return subject_data
if __name__ == '__main__':
# correction of distortion_parameters
# custom solution, to be improved in the future
main_dir = '/neurospin/ibc/'
cache_dir = '/neurospin/tmp/ibc'
prepare_derivatives(main_dir)
do_topup = True
protocol = 'clips4' # 'clips1', 'clips2', 'clips3', 'clips4', 'archi', 'hcp2' 'tom' 'preferences'
subject_session = sorted(get_subject_session(protocol))[-1:]
if do_topup:
acq = None
if protocol in ['rs']:
acq = 'mb6'
elif protocol in ['mtt1', 'mtt2']:
acq = 'mb3'
apply_topup(main_dir, cache_dir, subject_session, acq=acq)
subject_data = []
for protocol in [protocol]:
jobfile = 'ini_files/IBC_preproc_%s.ini' % protocol
subject_data_ = Parallel(n_jobs=1)(
delayed(run_subject_preproc)(jobfile, subject, session)
for subject, session in subject_session)
# Read the jobfile
list_subjects, params = _generate_preproc_pipeline(output_name)
# Preproc and Dump data
subject_data = do_subjects_preproc(output_name, report=True)
return subject_data
if __name__ == '__main__':
# correction of distortion_parameters
# custom solution, to be improved in the future
main_dir = '/neurospin/ibc/'
cache_dir = '/neurospin/tmp/ibc'
prepare_derivatives(main_dir)
do_topup = True
protocol = 'BBT3'
subject_session = sorted(get_subject_session(['bbt3']))
# subject_session = [('sub-15', 'ses-29')]
if do_topup:
acq = None
if protocol in ['rs']:
acq = 'mb6'
elif protocol in ['mtt1', 'mtt2']:
acq = 'mb3'
apply_topup(main_dir, cache_dir, subject_session, acq=acq)
subject_data = []
jobfile = 'ini_files/IBC_preproc_%s.ini' % protocol
subject_data_ = Parallel(n_jobs=1)(
delayed(run_subject_preproc)(jobfile, subject, session)
for subject, session in subject_session)
subject_data = subject_data + subject_data_[0]
X = rs_masker.transform(rs_fmri)
seeds = roi_masker.transform(rs_fmri)
normalization = np.sqrt(X.shape[0])
X = (X.T - X.mean(1)).T
X = (X.T / X.std(1)).T / normalization
seeds = (seeds.T - seeds.mean(1)).T
seeds = (seeds.T / seeds.std(1)).T / normalization
correlation = np.dot(seeds.T, X)
return correlation
q = 0
correlations = []
for subject in SUBJECTS:
for task in ['mtt1', 'mtt2']:
subject_session = np.array(get_subject_session(task))
session = subject_session[subject_session[:, 0] == subject][0][1]
for acq in ['ap', 'pa']:
print(subject, session, acq)
correlation = compute_connectivity(subject, session, acq)
correlations.append(correlation)
if q == 0:
mean_correlation = correlation
else:
mean_correlation += correlation
q += 1
mean_correlation /= q
correlations = np.array(correlations)
X = np.rollaxis(correlations, 1)
import scipy.stats as sts