def create_group_analysis(task_name):
def getContrasts(task_name):
from variables import design_parameters
return range(1, len(design_parameters[task_name]['contrasts'])+1)
from variables import subjects, exclude_subjects
if task_name in exclude_subjects.keys():
subjects = list(set(subjects) - set(exclude_subjects[task_name]))
pipeline = pe.Workflow(name="group_analysis_"+task_name)
pipeline.base_dir = working_dir
contrasts_infosource = pe.Node(interface=util.IdentityInterface(fields=['contrast_no']),
name="contrasts_infosource")
contrasts_infosource.iterables = ('contrast_no', getContrasts(task_name))
sessions_infosource = pe.Node(interface=util.IdentityInterface(fields=['session']),
name="sessions_infosource")
sessions_infosource.iterables = ('session', sessions)
datasource = pe.Node(interface=nio.DataGrabber(infields=['task_name', 'contrast_no'],
outfields=['con_images_first',
'con_images_second',
'spm_t_images_first',
'spm_t_images_second',
'mask_images']),
name = 'datasource', overwrite=True)
datasource.inputs.base_directory = "/home/filo/workspace/2010reliability/results/volumes/"
datasource.inputs.template = '%simages/_subject_id_%s/_session_%s/_task_name_%s/%s%04d.img'
datasource.inputs.field_template = dict(mask_images='mask_image/_subject_id_%s/_session_*/_task_name_%s/%s.img')
datasource.inputs.template_args = dict(con_images_first = [['con_', subjects, 'first', 'task_name', 'con_', 'contrast_no']],
con_images_second = [['con_', subjects, 'second', 'task_name', 'con_', 'contrast_no']],
spm_t_images_first = [['spmT_', subjects, 'first', 'task_name', 'spmT_','contrast_no']],
spm_t_images_second = [['spmT_', subjects, 'second', 'task_name', 'spmT_','contrast_no']],
mask_images = [[subjects, 'task_name', 'mask']])
datasource.inputs.sort_filelist = True
pipeline.connect(contrasts_infosource, "contrast_no", datasource, "contrast_no")
#datasource.inputs.session = "first"
datasource.inputs.task_name = task_name
thr_method_infosource = pe.Node(interface=util.IdentityInterface(fields=['thr_method']),
name="thr_method_infosource")
thr_method_infosource.iterables = ('thr_method', thr_methods)
overlap_datasource = pe.Node(interface=nio.DataGrabber(infields=['task_name', 'contrast_no'],
outfields=['overlap_map']),
name = 'overlap_datasource', overwrite=True)
overlap_datasource.inputs.base_directory = '/media/data/2010reliability/workdir_fmri/within_subjects_pipeline_noroi/'
overlap_datasource.inputs.template = '_task_name_%s/_masked_comparison_False/_subject_id_%s/_thr_method_%s/just_the_overlap/mapflow/_just_the_overlap%d/diff_thresh.nii'
overlap_datasource.inputs.template_args = dict(overlap_map = [['task_name', subjects, 'thr_method', 'contrast_no']])
overlap_datasource.inputs.sort_filelist = True
def minusOne(number):
return number -1
pipeline.connect(contrasts_infosource, ("contrast_no", minusOne), overlap_datasource, "contrast_no")
pipeline.connect(thr_method_infosource, 'thr_method', overlap_datasource, "thr_method")
#datasource.inputs.session = "first"
overlap_datasource.inputs.task_name = task_name
def stack_overlaps_func(file_list):
import nibabel as nb
import numpy as np
data_list = []
sum_max = 0
for file in file_list:
nii = nb.load(file)
data = nii.get_data()
data_list.append(data)
sum_max += data.max()
newdata = np.array(data_list).sum(axis=0)/sum_max
new_nii = nb.Nifti1Image(newdata, nii.get_affine(), nii.get_header())
new_file = "stack.nii"
nb.save(new_nii, new_file)
return new_file
stack_overlaps = pe.Node(interface=Function(function=stack_overlaps_func,
input_names=['file_list'],
output_names=['stack']), name="stack_overlaps")
pipeline.connect(overlap_datasource, "overlap_map", stack_overlaps, "file_list")
average = pe.MapNode(interface=fsl.maths.MultiImageMaths(op_string="-add %s -div 2"), name="average",
iterfield=['in_file', 'operand_files'])
pipeline.connect([(datasource, average, [('con_images_first', 'in_file'),
('con_images_second', 'operand_files')])
])
def pickCons(averages, firsts, seconds):
from variables import exceptions
new_list = []
for i,f in enumerate(firsts):
for exception in exceptions:
if f.find('task_name_'+exception["task_name"]) != -1 and f.find("subject_id_"+exception["subject_id"]) != -1:
if exception["which"] == "first":
new_list.append(f)
else:
new_list.append(seconds[i])
break
if len(new_list) == i:
new_list.append(averages[i])
assert len(averages) == len(new_list)
return new_list
pick_cons = pe.Node(interface=util.Function(function=pickCons,
input_names=['averages', 'firsts', 'seconds'],
output_names='out_list'),
name='pick_cons')
pipeline.connect([(average, pick_cons, [('out_file', 'averages')]),
(datasource, pick_cons, [('con_images_first', 'firsts'),
('con_images_second', 'seconds')])
])
#reslice = pe.MapNode(interface=spm.Coregister(jobtype="write"), name='reslice', iterfield=['source'])
def srtSubjects(in_files):
from variables import subjects
import re
mapping = [(subject, dict(scans=['',''],conds=[1,2])) for subject in subjects]
tmp_dict = dict(mapping)
print tmp_dict
for file in in_files:
subject_id = re.findall(r'_subject_id_([0-9a-z\-]+)',file)[0]
session_no = int((re.findall(r'_session_([a-z]+)',file)[0] == 'second'))
print subject_id
tmp_dict[subject_id]['scans'][session_no] = file
return tmp_dict.values()
def mask_union(list_of_masks):
import numpy as np
import nibabel as nb
import os
n_list_of_masks = []
for pair in list_of_masks:
for mask in pair:
n_list_of_masks.append(mask)
list_of_masks = n_list_of_masks
nii = nb.load(list_of_masks[0])
final_mask = np.ones(nii.get_shape())
for mask in list_of_masks:
print mask
data = nb.load(mask).get_data()
final_mask = np.logical_and(final_mask, np.logical_not(np.logical_or(data == 0, np.isnan(data))))
out_file = os.path.abspath("union_mask.nii")
nb.save(nb.Nifti1Image(final_mask, nii.get_affine(), nii.get_header()), out_file)
return out_file
union_mask = pe.Node(Function(function=mask_union,
input_names=['list_of_masks'],
output_names=['union_mask']),
name='union_mask')
pipeline.connect([(datasource, union_mask, [('mask_images', 'list_of_masks')]),
])
design = pe.Node(interface=spm.OneSampleTTestDesign(), name="design")
pipeline.connect([(pick_cons, design, [('out_list', 'in_files')]),
(union_mask, design, [('union_mask', 'explicit_mask_file')])
])
#
# if subset == "all":
# pipeline.connect([(datasource, onesamplettestdes, [('con_images', 'in_files')])
# ])
# elif subset == "righties":
# def pick_subjects(l):
# from variables import subjects, lefties
# righties = list(set(subjects) - set(lefties))
# l_out = []
# for item in l:
# for subject in righties:
# if subject in item:
# l_out.append(item)
# return l_out
# pipeline.connect([(datasource, onesamplettestdes, [(('con_images', pick_subjects), 'in_files')])
# ])
# elif subset == "lefties":
# def pick_subjects(l):
# from variables import subjects, lefties
# l_out = []
# for item in l:
# for subject in lefties:
# if subject in item:
# l_out.append(item)
# return l_out
# pipeline.connect([(datasource, onesamplettestdes, [(('con_images', pick_subjects), 'in_files')])
# ])
#reslice.inputs.target = '/media/data/2010reliability/normsize.nii'
#pipeline.connect(reslice, "coregistered_source", onesamplettestdes, "in_files")
l2estimate = pe.Node(interface=spm.EstimateModel(), name="level2estimate")
l2estimate.inputs.estimation_method = {'Classical' : 1}
pipeline.connect(design, "spm_mat_file", l2estimate, "spm_mat_file")
icc = pe.Node(interface=neuroutils.ICC(), name="icc")
pipeline.connect(l2estimate, "mask_image", icc, "mask")
pipeline.connect(datasource, "spm_t_images_first", icc, "first_session_t_maps")
pipeline.connect(datasource, "spm_t_images_second", icc, "second_session_t_maps")
l2conestimate = pe.Node(interface = spm.EstimateContrast(), name="level2conestimate")
cont1 = ('Group','T', ['mean'],[1])
l2conestimate.inputs.contrasts = [cont1]
l2conestimate.inputs.group_contrast = True
pipeline.connect(l2estimate, "spm_mat_file", l2conestimate, "spm_mat_file")
pipeline.connect(l2estimate, "beta_images", l2conestimate, "beta_images")
pipeline.connect(l2estimate, "residual_image", l2conestimate, "residual_image")
threshold = pe.Node(interface=spm.Threshold(), name="threshold")
threshold.inputs.contrast_index = 1
threshold.inputs.use_fwe_correction = False
threshold.inputs.height_threshold = 0.001
pipeline.connect(l2conestimate, 'spm_mat_file', threshold, 'spm_mat_file')
pipeline.connect(l2conestimate, 'spmT_images', threshold, 'stat_image')
return pipeline
# metric="jaccard",
# filename = os.path.join(results_dir, "compare within with between", "%s_%s_%s_roi_%s_jaccard.pdf"%(task,contrast, thr_method, roi)))
# compare covert with overt
for roi in rois:
for thr_method in ['topo_ggmm', 'topo_fdr']:
if roi:
masked_comparisons = [True]
else:
masked_comparisons = [False, True]
for masked_comparison in masked_comparisons:
exclude = []
for task in ["covert_verb_generation", "overt_verb_generation"]:
for exception in exceptions:
if exception['task_name'] == task:
exclude.append(exception['subject_id'])
if task in exclude_subjects.keys():
exclude += exclude_subjects[task]
plot_between_session_box_plot(where=[('thr_method',thr_method), ('roi', roi), ('masked_comparison', masked_comparison)],
metric="dice",
compare= ('task_name',["overt_verb_generation", "covert_verb_generation"]),
exclude=('subject_id', exclude),
filename = os.path.join(results_dir, "compare covert_verb_generation with overt_verb_generation", "%s_roi_%s_masked_%s_dice.pdf"%(thr_method, roi, masked_comparison)))
#for roi in rois:
# for thr_method in ['topo_ggmm', 'topo_fdr']:
# if roi:
# masked_comparisons = [True]
# else:
# masked_comparisons = [False, True]
# for masked_comparison in masked_comparisons:
# exclude = []
