def test_label_io():
"""Test IO of label files
"""
label = read_label(label_fname)
write_label('foo', label)
label2 = read_label('foo-lh.label')
for key in label.keys():
if key in ['comment', 'hemi']:
assert_true(label[key] == label2[key])
else:
assert_array_almost_equal(label[key], label2[key], 5)
def _merge_rois(mer_path, label_list):
"""
Function to merge a list of given labels.
Parameters
----------
mer_path: str
The directory for storing merged ROIs.
label_list: list
Labels to be merged
"""
class_list = []
class_list.append(label_list[0])
for test_fn in label_list[1:]:
test_label = mne.read_label(test_fn)
i = 0
belong = False
while (i < len(class_list)) and (belong is False):
class_label = mne.read_label(class_list[i])
label_name = class_label.name
if test_label.hemi != class_label.hemi:
i = i + 1
continue
overlapped = len(
np.intersect1d(test_label.vertices, class_label.vertices))
if overlapped > 0:
com_label = test_label + class_label
pre_test = test_label.name.split('_')[0]
pre_class = class_label.name.split('_')[0]
# label_name = pre_class + '_%s-%s' %(pre_test,class_label.name.split('-')[-1])
if pre_test != pre_class:
pre_class += ',%s' % pre_test
pre_class = list(set(pre_class.split(',')))
new_pre = ''
for pre in pre_class[:-1]:
new_pre += '%s,' % pre
new_pre = pre_class[-1]
label_name = '%s_' % (new_pre) + \
class_label.name.split('_')[-1]
os.remove(class_list[i])
os.remove(test_fn)
fn_newlabel = mer_path + '%s.label' % label_name
if os.path.isfile(fn_newlabel):
fn_newlabel = fn_newlabel[:fn_newlabel.rfind(
'_')] + '_new, %s' % fn_newlabel.split('_')[-1]
mne.write_label(fn_newlabel, com_label)
class_list[i] = fn_newlabel
belong = True
i = i + 1
if belong is False:
class_list.append(test_fn)
return len(class_list)
def _merge_rois(mer_path, label_list):
"""
Function to merge a list of given labels.
Parameters
----------
mer_path: str
The directory for storing merged ROIs.
label_list: list
Labels to be merged
"""
class_list = []
class_list.append(label_list[0])
for test_fn in label_list[1:]:
test_label = mne.read_label(test_fn)
i = 0
belong = False
while (i < len(class_list)) and (belong is False):
class_label = mne.read_label(class_list[i])
label_name = class_label.name
if test_label.hemi != class_label.hemi:
i = i + 1
continue
overlapped = len(np.intersect1d(test_label.vertices,
class_label.vertices))
if overlapped > 0:
com_label = test_label + class_label
pre_test = test_label.name.split('_')[0]
pre_class = class_label.name.split('_')[0]
# label_name = pre_class + '_%s-%s' %(pre_test,class_label.name.split('-')[-1])
if pre_test != pre_class:
pre_class += ',%s' % pre_test
pre_class = list(set(pre_class.split(',')))
new_pre = ''
for pre in pre_class[:-1]:
new_pre += '%s,' % pre
new_pre = pre_class[-1]
label_name = '%s_' % (new_pre) + \
class_label.name.split('_')[-1]
os.remove(class_list[i])
os.remove(test_fn)
fn_newlabel = mer_path + '%s.label' %label_name
if os.path.isfile(fn_newlabel):
fn_newlabel = fn_newlabel[:fn_newlabel.rfind('_')] + '_new, %s' % fn_newlabel.split('_')[-1]
mne.write_label(fn_newlabel, com_label)
class_list[i] = fn_newlabel
belong = True
i = i + 1
if belong is False:
class_list.append(test_fn)
return len(class_list)
def ROIs_Merging(subject):
import os, mne
import numpy as np
subject_path = subjects_dir + subject
list_dirs = os.walk(subject_path + '/func_labels/')
#list_dirs = os.walk(subjects_dir + subject)
#color = ['#990033', '#9900CC', '#FF6600', '#FF3333', '#00CC33']
#/home/qdong/freesurfer/subjects/101611/func_labels/stim_func_superiortemporal-lh.label
tri_list = ['']
res_list = ['']
for root, dirs, files in list_dirs:
for f in files:
label_fname = os.path.join(root, f)
if f[0:3] == 'tri':
tri_list.append(label_fname)
elif f[0:3] == 'res':
res_list.append(label_fname)
tri_list = tri_list[1:]
res_list = res_list[1:]
mer_path = subject_path + '/func_labels/merged/'
isExists = os.path.exists(mer_path)
if not isExists:
os.makedirs(mer_path)
com_list = ['']
for fn_tri in tri_list:
tri_label = mne.read_label(fn_tri)
com_label = tri_label.copy()
for fn_res in res_list:
res_label = mne.read_label(fn_res)
if tri_label.hemi != res_label.hemi:
continue
if len(np.intersect1d(tri_label.vertices, res_label.vertices)) > 0:
com_label = tri_label + res_label
tri_label.name += ',%s' % res_label.name
com_list.append(
fn_res) #Keep the overlapped ROIs related with res
mne.write_label(mer_path + '%s' % tri_label.name, com_label)
# save the independent res ROIs
com_list = com_list[1:]
ind_list = list(set(res_list) - set(com_list))
for fn_res in ind_list:
res_label = mne.read_label(fn_res)
res_label.save(mer_path + '%s' % res_label.name)
n_jobs=n_jobs,
seed=0)
print "Time elapsed : %s (s)" % (time() - t0)
clusters = [c.reshape(n_times, n_vertices).T for c in clusters]
#you get a cluster for every single thing that crossed the first-stage threshold
# stc_log_pv_cluster = copy.deepcopy(mean_stc1)
# stc_log_pv_cluster.data = np.zeros_like(stc_log_pv_cluster.data)
# for pv, c in zip(cluster_pv, clusters):
# stc_log_pv_cluster.data[c] = -np.log10(pv)
#
# stc_log_pv_cluster.save(prefix + 'clusters_pv_%s_%s' % (stat_name, t))
stc_cluster = copy.deepcopy(template_stc)
#you only write out a cluster to an stc file if it crosses the second-stage threshold
for k, c in enumerate(clusters):
stc_cluster._data = c
if cluster_pv[
k] < 0.3: ##This is the threshold for saving an stc file with cluster
stcFileName = '/cluster/kuperberg/SemPrMM/MEG/results/source_space/cluster_stats/' + prefix + '%d-%d_cluster%d_%s_thresh_%s_pv_%.3f' % (
args.t1 * 1000, args.t2 * 1000, k, stat_name, t, cluster_pv[k])
print stcFileName
stc_cluster.save(stcFileName)
#stc_cluster.save('/cluster/kuperberg/SemPrMM/MEG/results/source_space/cluster_stats/' + prefix + '%d-%d_cluster%d_%s_thresh_%s_pv_%.3f' % (args.t1*1000,args.t2*1000,k, stat_name, t, cluster_pv[k]))
labelArray = mne.stc_to_label(stc_cluster, 'fsaverage')
label = labelArray[0]
mne.write_label(stcFileName, label)
print 'pv : %s' % np.sort(cluster_pv)[:5]
''' Creates labels baased on the MNI teamplate descriptions that Philip sent me for the subcortical analysis '''
import mne
import numpy as np
import env
fname = '/Users/sudregp/Documents/surfaces/thalamus_right_morpho_labels_test.txt'
hemi = 'rh'
labels = np.genfromtxt(fname)
for label_num in np.unique(labels):
label_name = env.fsl + 'mni/label/' + hemi + '.' + str(int(label_num))
label = mne.Label(np.nonzero(labels == label_num)[0], hemi=hemi)
mne.write_label(label_name, label)
def save_label(self, filename):
""" save the label of the surface """
mne.write_label(filename, self.label)
belong = False
while (i < len(class_list)) and (belong == False):
class_label = mne.read_label(class_list[i])
label_name = class_label.name
com_label = class_label.copy()
if test_label.hemi != class_label.hemi:
i = i + 1
continue
if len(np.intersect1d(test_label.vertices, class_label.vertices)) > 0:
com_label = test_label + class_label
if test_label.name.split('_')[0] not in label_name:
label_name = '%s,' % (
test_label.name.split('_')[0]) + label_name
if os.path.dirname(class_list[i]) == mer_path:
os.remove(class_list[i])
mne.write_label(mer_path + '/%s' % label_name, com_label)
print label_name
class_list[i] = mer_path + '/%s.label' % label_name
belong = True
i = i + 1
if belong == False:
class_list.append(test_fn)
import shutil
com_path = subject_path + '/func_labels/common'
list_dirs = os.walk(subject_path + '/func_labels/merged/')
label_list = ['']
for root, dirs, files in list_dirs:
for f in files:
label_fname = os.path.join(root, f)
label_list.append(label_fname)
n_permutations=n_permutations,
tail=0,
stat_fun=stat_fun,
connectivity=connectivity,
n_jobs=n_jobs, seed=0)
print "Time elapsed : %s (s)" % (time() - t0)
clusters = [c.reshape(n_times, n_vertices).T for c in clusters]
#you get a cluster for every single thing that crossed the first-stage threshold
# stc_log_pv_cluster = copy.deepcopy(mean_stc1)
# stc_log_pv_cluster.data = np.zeros_like(stc_log_pv_cluster.data)
# for pv, c in zip(cluster_pv, clusters):
# stc_log_pv_cluster.data[c] = -np.log10(pv)
#
# stc_log_pv_cluster.save(prefix + 'clusters_pv_%s_%s' % (stat_name, t))
stc_cluster = copy.deepcopy(template_stc)
#you only write out a cluster to an stc file if it crosses the second-stage threshold
for k, c in enumerate(clusters):
stc_cluster.data = c
if cluster_pv[k] < 0.15: ##This is the threshold for saving an stc file with cluster
stcFileName = '/cluster/kuperberg/SemPrMM/MEG/results/source_space/cluster_stats/' + prefix + '%d-%d_cluster%d_%s_thresh_%s_pv_%.3f' % (args.t1*1000,args.t2*1000,k, stat_name, t, cluster_pv[k])
stc_cluster.save(stcFileName)
#stc_cluster.save('/cluster/kuperberg/SemPrMM/MEG/results/source_space/cluster_stats/' + prefix + '%d-%d_cluster%d_%s_thresh_%s_pv_%.3f' % (args.t1*1000,args.t2*1000,k, stat_name, t, cluster_pv[k]))
labelArray = mne.stc_to_label(stc_cluster, 'fsaverage')
label = labelArray[0]
mne.write_label(stcFileName, label)
print 'pv : %s' % np.sort(cluster_pv)[:5]
def create_source_models(subject, save=False, json_fname='default'):
""" Create cortical and subcortical source models
Pipeline for:
i) importing BrainVISA white meshes for positions
and MarsAtlas textures for areas
ii) create transformation file from BV to head coordinates
iii) create source spaces with cortical
and subcortical dipoles,
Parameters
----------
subject : str
Subject name
database :
To delete, database reference for trans file, useless from next version
save : bool | True
Allows to save source spaces and respective labels in the default directory
Returns
-------
surf_src : instance of SourceSpace
Cortical surface source spaces, lh and rh
surf_labels : instance of Labels
Cortical surfaces labels
vol_src : instance of VolSourceSpace
Subcortical volumes source space, lh and rh
vol_labels : instance of Labels
Subcortical volumes Labels
"""
if json_fname == 'default':
read_dir = op.join(op.abspath(__package__), 'config')
json_fname = op.join(read_dir, 'db_info.son')
database, project, db_mne, db_bv, db_fs = read_databases(json_fname)
raw_dir, prep_dir, trans_dir, mri_dir, src_dir, bem_dir, fwd_dir, hga_dir = read_directories(
json_fname)
###########################################################################
# -------------------------------------------------------------------------
# BrainVISA anatomical data
# -------------------------------------------------------------------------
# BV decimated white meshes (cortical sources)
fname_surf_L = op.join(db_bv, project, subject, 't1mri',
'default_acquisition', 'default_analysis',
'segmentation', 'mesh', 'surface_analysis',
'{0}_Lwhite_remeshed_hiphop.gii'.format(subject))
fname_surf_R = op.join(db_bv, project, subject, 't1mri',
'default_acquisition', 'default_analysis',
'segmentation', 'mesh', 'surface_analysis',
'{0}_Rwhite_remeshed_hiphop.gii'.format(subject))
# BV texture (MarsAtlas labels) for decimated white meshes
# (cortical sources)
fname_tex_L = op.join(db_bv, 'hiphop138-multiscale', 'Decimated', '4K',
'hiphop138_Lwhite_dec_4K_parcels_marsAtlas.gii')
fname_tex_R = op.join(db_bv, 'hiphop138-multiscale', 'Decimated', '4K',
'hiphop138_Rwhite_dec_4K_parcels_marsAtlas.gii')
# Labelling xls file
fname_atlas = op.join(db_mne, project, 'marsatlas',
'MarsAtlas_BV_2015.xls')
# Color palette (still used???)
fname_color = op.join(db_mne, project, 'marsatlas', 'MarsAtlas.ima')
# MarsAtlas volume parcellation
fname_vol = op.join(db_bv, project, subject, 't1mri',
'default_acquisition', 'default_analysis',
'segmentation', 'mesh', 'surface_analysis',
'{0}_parcellation.nii.gz'.format(subject))
# -------------------------------------------------------------------------
# Transformation files BV to FS
# -------------------------------------------------------------------------
# Referential file list
# (3 transformation files to transform BV meshes to FS space)
fname_trans_ref = op.join(db_mne, project, 'referential',
'referential.txt')
# This file contains the transformations for subject_01
fname_trans_out = op.join(db_mne, project, subject, 'ref',
'{0}-trans.trm'.format(subject))
name_lobe_vol = ['Subcortical']
# ---------------------------------------------------------------------
# Setting up the source space from BrainVISA results
# ---------------------------------------------------------------------
# http://martinos.org/mne/stable/manual/cookbook.html#source-localization
# Create .trm file transformation from BrainVisa to FreeSurfer needed
# for brain.py function for surface only
create_trans(subject, database, fname_trans_ref, fname_trans_out)
# Calculate cortical sources and MarsAtlas labels
print('\n---------- Cortical sources ----------\n')
surf_src, surf_labels = get_brain_surf_sources(subject, fname_surf_L,
fname_surf_R, fname_tex_L,
fname_tex_R,
fname_trans_out,
fname_atlas, fname_color)
if save == True:
print('\nSaving surface source space and labels.....')
mne.write_source_spaces(op.join(src_dir.format(subject),
'{0}_surf-src.fif'.format(subject)),
surf_src,
overwrite=True)
for sl in surf_labels:
mne.write_label(
op.join(src_dir.format(subject),
'{0}_surf-lab'.format(subject)), sl)
print('[done]')
# Create BEM model if needed
print('\nBEM model is needed for volume source space\n')
if not check_bem(json_fname, subject):
create_bem(json_fname, subject)
print('\n---------- Subcortical sources ----------\n')
vol_src, vol_labels = get_brain_vol_sources(subject,
fname_vol,
json_fname,
name_lobe_vol,
fname_trans_out,
fname_atlas,
space=5.)
if save == True:
print('Saving volume source space and labels.....')
mne.write_source_spaces(op.join(src_dir.format(subject),
'{0}_vol-src.fif'.format(subject)),
vol_src,
overwrite=True)
for vl in vol_labels:
mne.write_label(
op.join(src_dir.format(subject),
'{0}_vol-lab'.format(subject)), vl)
print('[done]')
#
print('\n---------- Sources Completed ----------\n')
return surf_src, surf_labels, vol_src, vol_labels
for fn_label1 in label_list:
label1 = mne.read_label(fn_label1)
com_label = label1.copy()
label_name = label1.name
for fn_label2 in label_list:
label2 = mne.read_label(fn_label2)
if label1.hemi != label2.hemi:
continue
if len(np.intersect1d(label1.vertices, label2.vertices)) > 0:
com_label = label1 + label2
if label2.name.split('_')[0] not in label_name:
label_name = '%s,' %(label2.name.split('_')[0]) + label_name
subjects = (label_name.split('_')[0]).split(',')
if len(set(subjects)) >= 4:
mne.write_label(mer_path + '%s' %label_name, com_label)
# mne.write_label(mer_path + '%s' %label1.name, com_label)
list_dirs = os.walk(mer_path)
label_list = ['']
for root, dirs, files in list_dirs:
for f in files:
label_fname = os.path.join(root, f)
label_list.append(label_fname)
label_list=label_list[1:]
com_path = subject_path+'/func_labels/common/'
isExists=os.path.exists(com_path)
if not isExists:
os.makedirs(com_path)
def create_source_models(subject,
bem_dir=None,
trans_dir=None,
src_dir=None,
json_fname='default',
decim='orig',
save=False):
""" Create cortical and subcortical source models
Pipeline for:
i) importing BrainVISA white meshes for positions
and MarsAtlas textures for areas
ii) create transformation file from BV to head coordinates
iii) create source spaces with cortical
and subcortical dipoles,
Parameters
----------
subject : str
Subject name
database :
To delete, database reference for trans file, useless from next version
save : bool | True
Allows to save source spaces and respective labels in the default directory
Returns
-------
surf_src : instance of SourceSpace
Cortical surface source spaces, lh and rh
surf_labels : instance of Labels
Cortical surfaces labels
vol_src : instance of VolSourceSpace
Subcortical volumes source space, lh and rh
vol_labels : instance of Labels
Subcortical volumes Labels
"""
# if json_fname == 'default':
# read_dir = op.join(op.abspath(__package__), 'config')
# json_fname = op.join(read_dir, 'db_coords.json')
#
# database, project, db_mne, db_bv, db_fs = read_databases(json_fname)
# raw_dir, prep_dir, trans_dir, mri_dir, src_dir, bem_dir, fwd_dir, hga_dir = read_directories(json_fname)
###########################################################################
# -------------------------------------------------------------------------
# BrainVISA anatomical data
# -------------------------------------------------------------------------
db_fs, db_bv, db_mne = read_db_coords(json_fname)
if db_mne != None:
_, _, trans_dir, _, src_dir, bem_dir, _, _ = mne_directories(db_mne)
trans_dir = trans_dir.format(subject)
src_dir = src_dir.format(subject)
bem_dir = bem_dir.format(subject)
assert decim in [
'orig', '1K', '2K', '3K', '4K'
], 'decim should be one of this string: orig, 1K, 2K, 3K, 4K'
assert op.exists(bem_dir), 'bem_dir is not an existing folder'
assert op.exists(trans_dir), 'trans_dir is not an existing folder'
assert op.exists(src_dir), 'src_dir is not an existing folder'
# BV decimated white meshes (cortical sources)
# fname_surf_L = op.join(db_bv, project, subject, 't1mri', 'default_acquisition', 'default_analysis', 'segmentation',
# 'mesh', 'surface_analysis', '{0}_Lwhite_remeshed_hiphop.gii'.format(subject))
#
# fname_surf_R = op.join(db_bv, project, subject, 't1mri', 'default_acquisition', 'default_analysis', 'segmentation',
# 'mesh', 'surface_analysis', '{0}_Rwhite_remeshed_hiphop.gii'.format(subject))
# BV texture (MarsAtlas labels) for decimated white meshes
# (cortical sources)
if decim == 'orig':
# BV decimated white meshes (cortical sources)
fname_surf_L = op.join(db_bv, subject, 't1mri', 'default_acquisition',
'default_analysis', 'segmentation', 'mesh',
'{0}_Lwhite.gii'.format(subject))
fname_surf_R = op.join(db_bv, subject, 't1mri', 'default_acquisition',
'default_analysis', 'segmentation', 'mesh',
'{0}_Rwhite.gii'.format(subject))
# BV texture (MarsAtlas labels) for decimated white meshes
fname_tex_L = op.join(
db_bv, subject, 't1mr', 'default_acquisition', 'default_analysis',
'segmentation', 'mesh', 'surface_analysis',
'{0}_Lwhite_parcels_marsAtlas.gii'.format(subject))
fname_tex_R = op.join(
db_bv, subject, 't1mr', 'default_acquisition', 'default_analysis',
'segmentation', 'mesh', 'surface_analysis',
'{0}_Rwhite_parcels_marsAtlas.gii'.format(subject))
else:
if not op.exists(op.join(db_bv, 'hiphop138-multiscale')):
get_decimated(db_bv)
# BV decimated white meshes (cortical sources)
fname_surf_L = op.join(
db_bv, subject, 't1mri', 'default_acquisition', 'default_analysis',
'segmentation', 'mesh', 'surface_analysis',
'{0}_Lwhite_remeshed_hiphop.gii'.format(subject))
fname_surf_R = op.join(
db_bv, subject, 't1mri', 'default_acquisition', 'default_analysis',
'segmentation', 'mesh', 'surface_analysis',
'{0}_Rwhite_remeshed_hiphop.gii'.format(subject))
# BV texture (MarsAtlas labels) for decimated white meshes
fname_tex_L = op.join(
db_bv, 'hiphop138-multiscale', 'Decimated', decim,
'hiphop138_Lwhite_dec_{0}_parcels_marsAtlas.gii'.format(decim))
fname_tex_R = op.join(
db_bv, 'hiphop138-multiscale', 'Decimated', decim,
'hiphop138_Rwhite_dec_{0}_parcels_marsAtlas.gii'.format(decim))
# Labelling xls file
# fname_atlas = op.join(db_mne, project, 'marsatlas', 'MarsAtlas_BV_2015.xls')
# Color palette (still used???)
# fname_color = op.join(db_mne, project, 'marsatlas', 'MarsAtlas.ima')
# MarsAtlas volume parcellation
# fname_vol = op.join(db_bv, subject, 't1mri', 'default_acquisition', 'default_analysis', 'segmentation',
# 'mesh', 'surface_analysis', '{0}_parcellation.nii.gz'.format(subject))
# -------------------------------------------------------------------------
# Transformation files BV to FS
# -------------------------------------------------------------------------
# Referential file list
# (3 transformation files to transform BV meshes to FS space)
# fname_trans_ref = op.join(db_mne, project, 'referential', 'referential.txt')
# This file contains the transformations
trans_out = op.join(trans_dir, '{0}-trans.trm'.format(subject))
# if trans_out is None:
# trans_out = op.join(db_bv, project, subject, 't1mri', 'default_acquisition', 'default_analysis',
# 'segmentation', 'mesh', '{0}-trans.trm'.format(subject))
# elif type(trans_out) == str:
# trans_out = op.join(trans_out, '{0}-trans.trm'.format(subject))
# name_lobe_vol = ['Subcortical']
# ---------------------------------------------------------------------
# Setting up the source space from BrainVISA results
# ---------------------------------------------------------------------
# http://martinos.org/mne/stable/manual/cookbook.html#source-localization
# Create .trm file transformation from BrainVisa to FreeSurfer needed
# for brain.py function for surface only
create_trans(subject, db_fs, db_bv, trans_out)
# Calculate cortical sources and MarsAtlas labels
print('\n---------- Cortical sources ----------\n')
surf_src, surf_labels = get_brain_surf_sources(subject, fname_surf_L,
fname_surf_R, fname_tex_L,
fname_tex_R, trans_out)
if save == True:
# src_dir = op.join(db_mne, project, subject, 'src')
# if not op.exists(src_dir):
# os.mkdir(src_dir)
# print('\nSaving surface source space and labels in {0}'.format(src_dir))
# mne.write_source_spaces(op.join(src_dir.format(subject), '{0}_surf-src.fif'.format(subject)), surf_src, overwrite=True)
# for sl in surf_labels:
# mne.write_label(op.join(src_dir.format(subject), '{0}_surf-lab'.format(subject)), sl)
# print('[done]')
print(
'\nSaving surface source space and labels in {0}'.format(src_dir))
mne.write_source_spaces(op.join(src_dir,
'{0}_surf-src.fif'.format(subject)),
surf_src,
overwrite=True)
for sl in surf_labels:
mne.write_label(op.join(src_dir, '{0}_surf-lab'.format(subject)),
sl)
print('[done]')
# Create BEM model if needed
print('\nBEM model is needed for volume source space\n')
if not check_bem(subject, bem_dir):
print('BEM model not found... creating bem model')
create_bem(subject, bem_dir, json_fname)
print('\n---------- Subcortical sources ----------\n')
vol_src, vol_labels = get_brain_vol_sources(subject,
bem_dir,
space=5.,
json_fname=json_fname)
if save == True:
print('Saving volume source space and labels in {0}'.format(src_dir))
mne.write_source_spaces(op.join(src_dir,
'{0}_vol-src.fif'.format(subject)),
vol_src,
overwrite=True)
for vl in vol_labels:
mne.write_label(op.join(src_dir, '{0}_vol-lab'.format(subject)),
vl)
print('[done]')
#
print('\n---------- Sources Completed ----------\n')
return surf_src, surf_labels, vol_src, vol_labels
