def _first_level(out_dir, data, design_matrices, contrasts,
glm_model='ar1', mask='compute', verbose=1):
if verbose:
print '%s:' % out_dir
data = check_niimgs(data)
design_matrices = check_design_matrices(design_matrices)
contrasts = check_contrasts(contrasts)
glm = FMRILinearModel(data, design_matrices, mask=mask)
glm.fit(do_scaling=True, model=glm_model)
for i, contrast_id in enumerate(contrasts):
if verbose:
print ' %s/%s - %s ' % (i, len(contrasts), contrast_id)
con_val = []
for session_con, session_dm in zip(contrasts[contrast_id],
design_matrices):
con = np.zeros(session_dm.shape[1])
con[:len(session_con)] = session_con
con_val.append(con)
z_map, t_map, c_map, var_map = glm.contrast(
con_val,
con_id=contrast_id,
output_z=True,
output_stat=True,
output_effects=True,
output_variance=True,)
for dtype, img in zip(['z', 't', 'c', 'var'],
[z_map, t_map, c_map, var_map]):
map_dir = os.path.join(out_dir, '%s_maps' % dtype)
if not os.path.exists(map_dir):
os.makedirs(map_dir)
path = os.path.join(
map_dir, '%s.nii.gz' % remove_special(contrast_id))
nb.save(img, path)
nb.save(glm.mask, os.path.join(out_dir, 'mask.nii.gz'))
def _filename(self, ext='json'):
return str(self.number) + ' - ' + remove_special(self.title) + '.' + ext
def _openfmri_intra(out_dir, doc, metadata=None, verbose=1):
"""
Parameters
----------
metadata: dict
- condition_key
https://openfmri.org/content/metadata-condition-key
Examples
--------
{'condition_key': {'task001 cond001': 'task',
'task001 cond002': 'parametric gain'}}
"""
if 'study_id' in doc:
study_dir = os.path.join(out_dir, doc['study_id'])
else:
study_dir = out_dir
if verbose > 0:
print '%s@%s: dumping stats intra' % (doc['subject_id'],
doc['study_id'])
subject_dir = os.path.join(study_dir, doc['subject_id'])
model_dir = os.path.join(study_dir, 'models', 'model001')
if not os.path.exists(model_dir):
os.makedirs(model_dir)
# conditions specification
conditions_spec = []
for key, val in sorted(metadata['condition_key'].iteritems()):
for i, name in enumerate(val):
conditions_spec.append(
'%s cond%03i %s\n' % (key.split(' ')[0], i + 1, name))
with open(os.path.join(model_dir, 'condition_key.txt'), 'wb') as f:
f.write(''.join(sorted(set(conditions_spec))))
# contrasts specification
contrasts_spec = []
for key, val in doc['task_contrasts'].iteritems():
if 'task_contrasts' in metadata:
key = doc['task_contrasts'][key]
for i, session_contrast in enumerate(val):
task_id = metadata['run_key'][i].split(' ')[0]
# check not null and 1d
if (np.abs(session_contrast).sum() > 0
and len(np.array(session_contrast).shape) == 1):
con = ' '.join(np.array(session_contrast).astype('|S32'))
contrasts_spec.append('%s %s %s\n' % (task_id, key, con))
with open(os.path.join(model_dir, 'task_contrasts.txt'), 'wb') as f:
f.write(''.join(sorted(set(contrasts_spec))))
# dump onsets
model_dir = os.path.join(subject_dir, 'model', 'model001')
onsets_dir = os.path.join(model_dir, 'onsets')
for onsets, run_key in zip(doc['onsets'], metadata['run_key']):
run_dir = os.path.join(onsets_dir, run_key.replace(' ', '_'))
if not os.path.exists(run_dir):
os.makedirs(run_dir)
for condition_id, values in sorted(onsets.iteritems()):
cond = os.path.join(run_dir, '%s.txt' % condition_id)
with open(cond, 'wb') as f:
for timepoint in values:
f.write('%s %s %s\n' % timepoint)
# analyses
for dtype in ['c_maps', 't_maps']:
data_dir = os.path.join(model_dir, dtype)
if not os.path.exists(data_dir):
os.makedirs(data_dir)
if isinstance(doc[dtype], dict):
for contrast_id in doc[dtype].keys():
fname = remove_special(contrast_id)
img = nb.load(doc[dtype][contrast_id])
nb.save(img, os.path.join(data_dir, '%s.nii.gz' % fname))
# general data for analysis
img = nb.load(doc['mask'])
nb.save(img, os.path.join(model_dir, 'mask.nii.gz'))
json.dump(doc, open(os.path.join(model_dir, 'SPM.json'), 'wb'))
def _filename(self, ext='json'):
return str(self.number) + ' - ' + remove_special(
self.title) + '.' + ext
