def create_epi_t1_nonlinear_pipeline(name='epi_t1_nonlinear'):
"""Creates a pipeline that performs nonlinear EPI to T1 registration using
the antsRegistration tool. Beforehand, the T1 image has to be processed in
freesurfer and the EPI timeseries should be realigned.
Example
-------
>>> nipype_epi_t1_nonlin = create_epi_t1_nonlinear_pipeline('nipype_epi_t1_nonlin')
>>> nipype_epi_t1_nonlin.inputs.inputnode.fs_subject_id = '123456'
>>> nipype_epi_t1_nonlin.inputs.inputnode.fs_subjects_dir = '/project/data/freesurfer'
>>> nipype_epi_t1_nonlin.inputs.inputnode.realigned_epi = 'mcflirt.nii.gz'
>>> nipype_epi_t1_nonlin.run()
Inputs::
inputnode.fs_subject_id # subject id used in freesurfer
inputnode.fs_subjects_dir # path to freesurfer output
inputnode.realigned_epi # realigned EPI timeseries
Outputs::
outputnode.lin_epi2anat # ITK format
outputnode.lin_anat2epi # ITK format
outputnode.nonlin_epi2anat # ANTs specific 5D deformation field
outputnode.nonlin_anat2epi # ANTs specific 5D deformation field
"""
nonreg = Workflow(name='epi_t1_nonlinear')
# input
inputnode = Node(interface=util.IdentityInterface(
fields=['fs_subject_id', 'fs_subjects_dir', 'realigned_epi']),
name='inputnode')
# calculate the temporal mean image of the realigned timeseries
tmean = Node(interface=fsl.maths.MeanImage(dimension='T',
output_type='NIFTI_GZ'),
name='tmean')
nonreg.connect(inputnode, 'realigned_epi', tmean, 'in_file')
# import brain.mgz and ribbon.mgz from freesurfer directory
fs_import = Node(interface=nio.FreeSurferSource(),
name='freesurfer_import')
nonreg.connect(inputnode, 'fs_subjects_dir', fs_import, 'subjects_dir')
nonreg.connect(inputnode, 'fs_subject_id', fs_import, 'subject_id')
# convert brain.mgz to niigz
mriconvert = Node(interface=fs.MRIConvert(out_type='niigz'),
name='mriconvert')
nonreg.connect(fs_import, 'brain', mriconvert, 'in_file')
# calculate rigid transformation of mean epi to t1 with bbregister
bbregister = Node(interface=fs.BBRegister(init='fsl',
contrast_type='t2',
out_fsl_file=True),
name='bbregister')
nonreg.connect(inputnode, 'fs_subjects_dir', bbregister, 'subjects_dir')
nonreg.connect(inputnode, 'fs_subject_id', bbregister, 'subject_id')
nonreg.connect(tmean, 'out_file', bbregister, 'source_file')
# convert linear transformation to itk format compatible with ants
itk = Node(interface=c3.C3dAffineTool(fsl2ras=True,
itk_transform='epi2anat_affine.txt'),
name='itk')
nonreg.connect(tmean, 'out_file', itk, 'source_file')
nonreg.connect(mriconvert, 'out_file', itk, 'reference_file')
nonreg.connect(bbregister, 'out_fsl_file', itk, 'transform_file')
# get aparc aseg mask
# create brainmask from aparc+aseg
def get_aparc_aseg(files):
for name in files:
if 'aparc+aseg' in name:
return name
aparc_aseg_mask = Node(fs.Binarize(min=0.1,
dilate=10,
erode=7,
out_type='nii.gz',
binary_file='aparc_aseg_mask.nii.gz'),
name='aparc_aseg_mask')
# fill holes in mask
fillholes = Node(fsl.maths.MathsCommand(args='-fillh'), name='fillholes')
nonreg.connect([(fs_import, aparc_aseg_mask, [
(('aparc_aseg', get_aparc_aseg), 'in_file')
]), (aparc_aseg_mask, fillholes, [('binary_file', 'in_file')])])
#create bounding box mask and rigidly transform into anatomical (fs) space
fov = Node(interface=fs.model.Binarize(min=0.0, out_type='nii.gz'),
name='fov')
nonreg.connect(tmean, 'out_file', fov, 'in_file')
fov_trans = Node(interface=ants.resampling.ApplyTransforms(
dimension=3, interpolation='NearestNeighbor'),
name='fov_trans')
nonreg.connect(itk, ('itk_transform', filename_to_list), fov_trans,
'transforms')
nonreg.connect(fov, 'binary_file', fov_trans, 'input_image')
nonreg.connect(fillholes, 'out_file', fov_trans, 'reference_image')
#nonreg.connect(ribbon, 'binary_file', fov_trans, 'reference_image')
# intersect both masks
intersect = Node(interface=fsl.maths.BinaryMaths(operation='mul'),
name='intersect')
nonreg.connect(fillholes, 'out_file', intersect, 'in_file')
#nonreg.connect(ribbon, 'binary_file', intersect, 'in_file')
nonreg.connect(fov_trans, 'output_image', intersect, 'operand_file')
# inversly transform mask and mask original epi
mask_trans = Node(interface=ants.resampling.ApplyTransforms(
dimension=3,
interpolation='NearestNeighbor',
invert_transform_flags=[True]),
name='mask_trans')
nonreg.connect(itk, ('itk_transform', filename_to_list), mask_trans,
'transforms')
nonreg.connect(intersect, 'out_file', mask_trans, 'input_image')
nonreg.connect(tmean, 'out_file', mask_trans, 'reference_image')
maskepi = Node(interface=fs.utils.ApplyMask(), name='maskepi')
nonreg.connect(mask_trans, 'output_image', maskepi, 'mask_file')
nonreg.connect(tmean, 'out_file', maskepi, 'in_file')
# mask anatomical image (brain)
maskanat = Node(interface=fs.utils.ApplyMask(), name='maskanat')
nonreg.connect(intersect, 'out_file', maskanat, 'mask_file')
nonreg.connect(mriconvert, 'out_file', maskanat, 'in_file')
# invert masked anatomical image
anat_min_max = Node(interface=fsl.utils.ImageStats(op_string='-R'),
name='anat_min_max')
epi_min_max = Node(interface=fsl.utils.ImageStats(op_string='-r'),
name='epi_min_max')
nonreg.connect(maskanat, 'out_file', anat_min_max, 'in_file')
nonreg.connect(tmean, 'out_file', epi_min_max, 'in_file')
def calc_inversion(anat_min_max, epi_min_max):
mul = -(epi_min_max[1] - epi_min_max[0]) / (anat_min_max[1] -
anat_min_max[0])
add = abs(anat_min_max[1] * mul) + epi_min_max[0]
return mul, add
calcinv = Node(interface=Function(
input_names=['anat_min_max', 'epi_min_max'],
output_names=['mul', 'add'],
function=calc_inversion),
name='calcinv')
nonreg.connect(anat_min_max, 'out_stat', calcinv, 'anat_min_max')
nonreg.connect(epi_min_max, 'out_stat', calcinv, 'epi_min_max')
mulinv = Node(interface=fsl.maths.BinaryMaths(operation='mul'),
name='mulinv')
addinv = Node(interface=fsl.maths.BinaryMaths(operation='add'),
name='addinv')
nonreg.connect(maskanat, 'out_file', mulinv, 'in_file')
nonreg.connect(calcinv, 'mul', mulinv, 'operand_value')
nonreg.connect(mulinv, 'out_file', addinv, 'in_file')
nonreg.connect(calcinv, 'add', addinv, 'operand_value')
# nonlinear transformation of masked anat to masked epi with ants
antsreg = Node(interface=ants.registration.Registration(
dimension=3,
invert_initial_moving_transform=True,
metric=['CC'],
metric_weight=[1.0],
radius_or_number_of_bins=[4],
sampling_strategy=['None'],
transforms=['SyN'],
args='-g .1x1x.1',
transform_parameters=[(0.10, 3, 0)],
number_of_iterations=[[10, 5]],
convergence_threshold=[1e-06],
convergence_window_size=[10],
shrink_factors=[[2, 1]],
smoothing_sigmas=[[1, 0.5]],
sigma_units=['vox'],
use_estimate_learning_rate_once=[True],
use_histogram_matching=[True],
collapse_output_transforms=True,
output_inverse_warped_image=True,
output_warped_image=True),
name='antsreg')
nonreg.connect(itk, 'itk_transform', antsreg, 'initial_moving_transform')
nonreg.connect(maskepi, 'out_file', antsreg, 'fixed_image')
nonreg.connect(addinv, 'out_file', antsreg, 'moving_image')
# output
def second_element(file_list):
return file_list[1]
def first_element(file_list):
return file_list[0]
outputnode = Node(interface=util.IdentityInterface(fields=[
'lin_epi2anat', 'lin_anat2epi', 'nonlin_epi2anat', 'nonlin_anat2epi'
]),
name='outputnode')
nonreg.connect(itk, 'itk_transform', outputnode, 'lin_epi2anat')
nonreg.connect(antsreg, ('forward_transforms', first_element), outputnode,
'lin_anat2epi')
nonreg.connect(antsreg, ('forward_transforms', second_element), outputnode,
'nonlin_anat2epi')
nonreg.connect(antsreg, ('reverse_transforms', second_element), outputnode,
'nonlin_epi2anat')
return nonreg
def init_fsl_bbr_wf(use_bbr, bold2t1w_dof, name='fsl_bbr_wf'):
"""
This workflow uses FSL FLIRT to register a BOLD image to a T1-weighted
structural image, using a boundary-based registration (BBR) cost function.
It is a counterpart to :py:func:`~fmriprep.workflows.bold.registration.init_bbreg_wf`,
which performs the same task using FreeSurfer's ``bbregister``.
The ``use_bbr`` option permits a high degree of control over registration.
If ``False``, standard, rigid coregistration will be performed by FLIRT.
If ``True``, FLIRT-BBR will be seeded with the initial transform found by
the rigid coregistration.
If ``None``, after FLIRT-BBR is run, the resulting affine transform
will be compared to the initial transform found by FLIRT.
Excessive deviation will result in rejecting the BBR refinement and
accepting the original, affine registration.
.. workflow ::
:graph2use: orig
:simple_form: yes
from fmriprep.workflows.bold.registration import init_fsl_bbr_wf
wf = init_fsl_bbr_wf(use_bbr=True, bold2t1w_dof=9)
Parameters
use_bbr : bool or None
Enable/disable boundary-based registration refinement.
If ``None``, test BBR result for distortion before accepting.
bold2t1w_dof : 6, 9 or 12
Degrees-of-freedom for BOLD-T1w registration
name : str, optional
Workflow name (default: fsl_bbr_wf)
Inputs
in_file
Reference BOLD image to be registered
t1_brain
Skull-stripped T1-weighted structural image
t1_seg
FAST segmentation of ``t1_brain``
t1_2_fsnative_reverse_transform
Unused (see :py:func:`~fmriprep.workflows.util.init_bbreg_wf`)
subjects_dir
Unused (see :py:func:`~fmriprep.workflows.util.init_bbreg_wf`)
subject_id
Unused (see :py:func:`~fmriprep.workflows.util.init_bbreg_wf`)
Outputs
itk_bold_to_t1
Affine transform from ``ref_bold_brain`` to T1 space (ITK format)
itk_t1_to_bold
Affine transform from T1 space to BOLD space (ITK format)
out_report
Reportlet for assessing registration quality
fallback
Boolean indicating whether BBR was rejected (rigid FLIRT registration returned)
"""
workflow = Workflow(name=name)
workflow.__desc__ = """\
The BOLD reference was then co-registered to the T1w reference using
`flirt` [FSL {fsl_ver}, @flirt] with the boundary-based registration [@bbr]
cost-function.
Co-registration was configured with nine degrees of freedom to account
for distortions remaining in the BOLD reference.
""".format(fsl_ver=FLIRTRPT().version or '<ver>')
inputnode = pe.Node(
niu.IdentityInterface([
'in_file',
't1_2_fsnative_reverse_transform',
'subjects_dir',
'subject_id', # BBRegister
't1_seg',
't1_brain'
]), # FLIRT BBR
name='inputnode')
outputnode = pe.Node(niu.IdentityInterface(
['itk_bold_to_t1', 'itk_t1_to_bold', 'out_report', 'fallback']),
name='outputnode')
wm_mask = pe.Node(niu.Function(function=extract_wm), name='wm_mask')
flt_bbr_init = pe.Node(FLIRTRPT(dof=6,
generate_report=not use_bbr,
uses_qform=True),
name='flt_bbr_init')
invt_bbr = pe.Node(fsl.ConvertXFM(invert_xfm=True),
name='invt_bbr',
mem_gb=DEFAULT_MEMORY_MIN_GB)
# BOLD to T1 transform matrix is from fsl, using c3 tools to convert to
# something ANTs will like.
fsl2itk_fwd = pe.Node(c3.C3dAffineTool(fsl2ras=True, itk_transform=True),
name='fsl2itk_fwd',
mem_gb=DEFAULT_MEMORY_MIN_GB)
fsl2itk_inv = pe.Node(c3.C3dAffineTool(fsl2ras=True, itk_transform=True),
name='fsl2itk_inv',
mem_gb=DEFAULT_MEMORY_MIN_GB)
workflow.connect([
(inputnode, flt_bbr_init, [('in_file', 'in_file'),
('t1_brain', 'reference')]),
(inputnode, fsl2itk_fwd, [('t1_brain', 'reference_file'),
('in_file', 'source_file')]),
(inputnode, fsl2itk_inv, [('in_file', 'reference_file'),
('t1_brain', 'source_file')]),
(invt_bbr, fsl2itk_inv, [('out_file', 'transform_file')]),
(fsl2itk_fwd, outputnode, [('itk_transform', 'itk_bold_to_t1')]),
(fsl2itk_inv, outputnode, [('itk_transform', 'itk_t1_to_bold')]),
])
# Short-circuit workflow building, use rigid registration
if use_bbr is False:
workflow.connect([
(flt_bbr_init, invt_bbr, [('out_matrix_file', 'in_file')]),
(flt_bbr_init, fsl2itk_fwd, [('out_matrix_file', 'transform_file')
]),
(flt_bbr_init, outputnode, [('out_report', 'out_report')]),
])
outputnode.inputs.fallback = True
return workflow
flt_bbr = pe.Node(FLIRTRPT(cost_func='bbr',
dof=bold2t1w_dof,
generate_report=True,
schedule=op.join(os.getenv('FSLDIR'),
'etc/flirtsch/bbr.sch')),
name='flt_bbr')
workflow.connect([
(inputnode, wm_mask, [('t1_seg', 'in_seg')]),
(inputnode, flt_bbr, [('in_file', 'in_file'),
('t1_brain', 'reference')]),
(flt_bbr_init, flt_bbr, [('out_matrix_file', 'in_matrix_file')]),
(wm_mask, flt_bbr, [('out', 'wm_seg')]),
])
# Short-circuit workflow building, use boundary-based registration
if use_bbr is True:
workflow.connect([
(flt_bbr, invt_bbr, [('out_matrix_file', 'in_file')]),
(flt_bbr, fsl2itk_fwd, [('out_matrix_file', 'transform_file')]),
(flt_bbr, outputnode, [('out_report', 'out_report')]),
])
outputnode.inputs.fallback = False
return workflow
transforms = pe.Node(niu.Merge(2),
run_without_submitting=True,
name='transforms')
reports = pe.Node(niu.Merge(2),
run_without_submitting=True,
name='reports')
compare_transforms = pe.Node(niu.Function(function=compare_xforms),
name='compare_transforms')
select_transform = pe.Node(niu.Select(),
run_without_submitting=True,
name='select_transform')
select_report = pe.Node(niu.Select(),
run_without_submitting=True,
name='select_report')
fsl_to_lta = pe.MapNode(fs.utils.LTAConvert(out_lta=True),
iterfield=['in_fsl'],
name='fsl_to_lta')
workflow.connect([
(flt_bbr, transforms, [('out_matrix_file', 'in1')]),
(flt_bbr_init, transforms, [('out_matrix_file', 'in2')]),
# Convert FSL transforms to LTA (RAS2RAS) transforms and compare
(inputnode, fsl_to_lta, [('in_file', 'source_file'),
('t1_brain', 'target_file')]),
(transforms, fsl_to_lta, [('out', 'in_fsl')]),
(fsl_to_lta, compare_transforms, [('out_lta', 'lta_list')]),
(compare_transforms, outputnode, [('out', 'fallback')]),
# Select output transform
(transforms, select_transform, [('out', 'inlist')]),
(compare_transforms, select_transform, [('out', 'index')]),
(select_transform, invt_bbr, [('out', 'in_file')]),
(select_transform, fsl2itk_fwd, [('out', 'transform_file')]),
(flt_bbr, reports, [('out_report', 'in1')]),
(flt_bbr_init, reports, [('out_report', 'in2')]),
(reports, select_report, [('out', 'inlist')]),
(compare_transforms, select_report, [('out', 'index')]),
(select_report, outputnode, [('out', 'out_report')]),
])
return workflow
maskT2 = pe.Node(interface=fsl.ImageMaths(op_string = '-mas', suffix = '_bet'), name='maskT2')
preproc.connect(t2tot1, 'out_file', maskT2, 'in_file')
preproc.connect(inflate_mask, 'out_file', maskT2, 'in_file2')
# apply inflated mask to t1
maskT1 = pe.Node(interface=fsl.ImageMaths(op_string = '-mas', suffix = '_bet'), name='maskT1')
preproc.connect(T1_to_standard, 'out_file', maskT1, 'in_file')
preproc.connect(inflate_mask, 'out_file', maskT1, 'in_file2')
# affine co-reg (flirt) of T1 to caltech atlas
init_t1_to_atlas_coreg = pe.Node(interface=fsl.FLIRT(dof=6, output_type = "NIFTI_GZ"), name='init_t1_to_atlas_coreg')
preproc.connect(maskT1, 'out_file', init_t1_to_atlas_coreg, 'in_file')
preproc.connect(downsample_atlas_t1, 'out_file', init_t1_to_atlas_coreg, 'reference')
# convert transform matrix to ants format (ITK)
fsl2ras = pe.Node(interface=c3.C3dAffineTool(fsl2ras = True, itk_transform=True), name='fsl2ras')
preproc.connect(downsample_atlas_t1, 'out_file', fsl2ras, 'reference_file')
preproc.connect(maskT1, 'out_file', fsl2ras, 'source_file')
preproc.connect(init_t1_to_atlas_coreg, 'out_matrix_file', fsl2ras, 'transform_file')
# merge fixed and moving images into list
merge_fixed = pe.Node(interface=util.Merge(2, axis='hstack'), name='merge_fixed')
preproc.connect(downsample_atlas_t1, 'out_file', merge_fixed, 'in1')
preproc.connect(downsample_atlas_t2, 'out_file', merge_fixed, 'in2')
merge_moving = pe.Node(interface=util.Merge(2, axis='hstack'), name='merge_moving')
preproc.connect(maskT1, 'out_file', merge_moving, 'in1')
preproc.connect(maskT2, 'out_file', merge_moving, 'in2')
# assemble the inputs for coregistration of t2/t1 to caltech atlas
structs_to_atlas_coreg_input = pe.Node(interface=util.IdentityInterface(fields=['fixed_image', 'moving_image', 'fixed_image_mask', 'moving_image_mask', 'initial_moving_transform']), name='structs_to_atlas_coreg_input')
preproc.connect(merge_fixed, ('out', unlist), structs_to_atlas_coreg_input, 'fixed_image')
def get_post_struct_norm_workflow(name='normalize_post_struct'):
""" Base post-structural workflow for normalization
Parameters
----------
name : name of workflow. Default = 'normalize_post_struct'
Inputs
------
inputspec.template_file :
inputspec.unwarped_brain :
inputspec.warp_field :
inputspec.affine_transformation :
inputspec.out_fsl_file :
inputspec.moving_image :
inputspec.mean_func :
inputspec.use_nearest:
Outputs
-------
outputspec.warped_image :
Returns
-------
workflow : post-structural normalization workflow
"""
#inputs to workflow
import nipype.interfaces.freesurfer as fs
import nipype.interfaces.ants as ants
import nipype.pipeline.engine as pe
import nipype.interfaces.utility as util
inputspec = pe.Node(util.IdentityInterface(fields=[
'template_file', 'unwarped_brain', 'warp_field',
'affine_transformation', 'out_fsl_file', 'moving_image', 'mean_func',
"use_nearest"
]),
name='inputspec')
#makes fsl-style coregistration ANTS compatible
fsl_reg_2_itk = pe.Node(c3.C3dAffineTool(fsl2ras=True),
name='fsl_reg_2_itk')
#collects series of transformations to be applied to the moving images
collect_transforms = pe.Node(util.Merge(3), name='collect_transforms')
#performs series of transformations on moving images
warp_images = pe.MapNode(ants.WarpTimeSeriesImageMultiTransform(),
name='warp_images',
iterfield=['input_image', 'dimension'])
#collects workflow outputs
outputspec = pe.Node(util.IdentityInterface(fields=['warped_image']),
name='outputspec')
#initializes and connects workflow nodes
normalize_post_struct = pe.Workflow(name=name)
normalize_post_struct.connect([
(inputspec, fsl_reg_2_itk, [('unwarped_brain', 'reference_file')]),
(inputspec, fsl_reg_2_itk, [('out_fsl_file', 'transform_file')]),
(inputspec, fsl_reg_2_itk, [('mean_func', 'source_file')]),
(fsl_reg_2_itk, collect_transforms, [('itk_transform', 'in3')]),
(inputspec, collect_transforms, [('warp_field', 'in1'),
('affine_transformation', 'in2')]),
(inputspec, warp_images, [('moving_image', 'input_image')]),
(inputspec, warp_images, [(('moving_image', get_image_dimensions),
'dimension')]),
(inputspec, warp_images, [('template_file', 'reference_image'),
('use_nearest', 'use_nearest')]),
(collect_transforms, warp_images, [('out', 'transformation_series')]),
(warp_images, outputspec, [('output_image', 'warped_image')])
])
return normalize_post_struct
def create_coreg_pipeline(name='coreg'):
# fsl output type
fsl.FSLCommand.set_default_output_type('NIFTI_GZ')
# initiate workflow
coreg = Workflow(name='coreg')
#inputnode
inputnode = Node(util.IdentityInterface(fields=[
'epi_median',
'fs_subjects_dir',
'fs_subject_id',
'uni_highres',
]),
name='inputnode')
# outputnode
outputnode = Node(util.IdentityInterface(fields=[
'uni_lowres', 'epi2lowres', 'epi2lowres_mat', 'epi2lowres_dat',
'highres2lowres', 'highres2lowres_mat', 'highres2lowres_dat',
'epi2highres_lin', 'epi2highres_lin_mat', 'epi2highres_lin_itk'
]),
name='outputnode')
# convert mgz head file for reference
fs_import = Node(interface=nio.FreeSurferSource(), name='fs_import')
brain_convert = Node(fs.MRIConvert(out_type='niigz',
out_file='uni_lowres.nii.gz'),
name='brain_convert')
coreg.connect([(inputnode, fs_import, [('fs_subjects_dir', 'subjects_dir'),
('fs_subject_id', 'subject_id')]),
(fs_import, brain_convert, [('brain', 'in_file')]),
(brain_convert, outputnode, [('out_file', 'uni_lowres')])])
# linear registration epi median to lowres mp2rage with bbregister
bbregister_epi = Node(fs.BBRegister(contrast_type='t2',
out_fsl_file='epi2lowres.mat',
out_reg_file='epi2lowres.dat',
registered_file='epi2lowres.nii.gz',
init='fsl',
epi_mask=True),
name='bbregister_epi')
coreg.connect([
(inputnode, bbregister_epi, [('fs_subjects_dir', 'subjects_dir'),
('fs_subject_id', 'subject_id'),
('epi_median', 'source_file')]),
(bbregister_epi, outputnode, [('out_fsl_file', 'epi2lowres_mat'),
('out_reg_file', 'epi2lowres_dat'),
('registered_file', 'epi2lowres')])
])
# linear register highres mp2rage to lowres mp2rage
bbregister_anat = Node(fs.BBRegister(
contrast_type='t1',
out_fsl_file='highres2lowres.mat',
out_reg_file='highres2lowres.dat',
registered_file='highres2lowres.nii.gz',
init='fsl'),
name='bbregister_anat')
coreg.connect([
(inputnode, bbregister_anat, [('fs_subjects_dir', 'subjects_dir'),
('fs_subject_id', 'subject_id'),
('uni_highres', 'source_file')]),
(bbregister_anat, outputnode, [('out_fsl_file', 'highres2lowres_mat'),
('out_reg_file', 'highres2lowres_dat'),
('registered_file', 'highres2lowres')])
])
# invert highres2lowres transform
invert = Node(fsl.ConvertXFM(invert_xfm=True), name='invert')
coreg.connect([(bbregister_anat, invert, [('out_fsl_file', 'in_file')])])
# concatenate epi2highres transforms
concat = Node(fsl.ConvertXFM(concat_xfm=True,
out_file='epi2highres_lin.mat'),
name='concat')
coreg.connect([(bbregister_epi, concat, [('out_fsl_file', 'in_file')]),
(invert, concat, [('out_file', 'in_file2')]),
(concat, outputnode, [('out_file', 'epi2higres_lin_mat')])])
# convert epi2highres transform into itk format
itk = Node(interface=c3.C3dAffineTool(fsl2ras=True,
itk_transform='epi2highres_lin.txt'),
name='itk')
coreg.connect([(inputnode, itk, [('epi_median', 'source_file'),
('uni_highres', 'reference_file')]),
(concat, itk, [('out_file', 'transform_file')]),
(itk, outputnode, [('itk_transform', 'epi2highres_lin_itk')
])])
# transform epi to highres
epi2highres = Node(ants.ApplyTransforms(
dimension=3,
output_image='epi2highres_lin.nii.gz',
interpolation='BSpline',
),
name='epi2highres')
coreg.connect([
(inputnode, epi2highres, [('uni_highres', 'reference_image'),
('epi_median', 'input_image')]),
(itk, epi2highres, [('itk_transform', 'transforms')]),
(epi2highres, outputnode, [('output_image', 'epi2highres_lin')])
])
return coreg
def ants_apply_warps_func_mni(workflow,
output_name,
func_key,
ref_key,
num_strat,
strat,
interpolation_method='LanczosWindowedSinc',
distcor=False,
map_node=False,
inverse=False,
symmetry='asymmetric',
input_image_type=0,
num_ants_cores=1):
"""
Applies previously calculated ANTS registration transforms to input
images. This workflow employs the antsApplyTransforms tool:
http://stnava.github.io/ANTs/
Parameters
----------
name : string, optional
Name of the workflow.
Returns
-------
apply_ants_warp_wf : nipype.pipeline.engine.Workflow
Notes
-----
Workflow Inputs::
workflow: Nipype workflow object
the workflow containing the resources involved
output_name: str
what the name of the warped functional should be when written to the
resource pool
func_key: string
resource pool key correspoding to the node containing the 3D or 4D
functional file to be written into MNI space, use 'leaf' for a
leaf node
ref_key: string
resource pool key correspoding to the file path to the template brain
used for functional-to-template registration
num_strat: int
the number of strategy objects
strat: C-PAC Strategy object
a strategy with one or more resource pools
interpolation_method: str
which interpolation to use when applying the warps, commonly used
options are 'Linear', 'Bspline', 'LanczosWindowedSinc' (default)
for derivatives and image data 'NearestNeighbor' for masks
distcor: boolean
indicates whether a distortion correction transformation should be
added to the transforms, this of course requires that a distortion
correction map exist in the resource pool
map_node: boolean
indicates whether a mapnode should be used, if TRUE func_key is
expected to correspond to a list of resources that should each
be written into standard space with the other parameters
inverse: boolean
writes the invrse of the transform, i.e. MNI->EPI instead of
EPI->MNI
input_image_type: int
argument taken by the ANTs apply warp tool; in this case, should be
0 for scalars (default) and 3 for 4D functional time-series
num_ants_cores: int
the number of CPU cores dedicated to ANTS anatomical-to-standard
registration
Workflow Outputs::
outputspec.output_image : string (nifti file)
Normalized output file
Workflow Graph:
.. image::
:width: 500
Detailed Workflow Graph:
.. image::
:width: 500
Apply the functional-to-structural and structural-to-template warps to
the 4D functional time-series to warp it to template space.
Parameters
----------
"""
# if the input is a string, assume that it is resource pool key,
# if it is a tuple, assume that it is a node, outfile pair,
# otherwise, something funky is going on
if isinstance(func_key, str):
if func_key == "leaf":
input_node, input_out = strat.get_leaf_properties()
else:
input_node, input_out = strat[func_key]
elif isinstance(func_key, tuple):
input_node, input_out = func_key
if isinstance(ref_key, str):
ref_node, ref_out = strat[ref_key]
elif isinstance(ref_key, tuple):
ref_node, ref_out = func_key
# when inverse is enabled, we want to update the name of various
# nodes so that we know they were inverted
inverse_string = ''
if inverse is True:
inverse_string = '_inverse'
# make sure that resource pool has some required resources before proceeding
if 'fsl_mat_as_itk' not in strat:
fsl_reg_2_itk = pe.Node(c3.C3dAffineTool(),
name='fsl_reg_2_itk_{0}'.format(num_strat))
fsl_reg_2_itk.inputs.itk_transform = True
fsl_reg_2_itk.inputs.fsl2ras = True
# convert the .mat from linear Func->Anat to
# ANTS format
node, out_file = strat['functional_to_anat_linear_xfm']
workflow.connect(node, out_file, fsl_reg_2_itk, 'transform_file')
node, out_file = strat['anatomical_brain']
workflow.connect(node, out_file, fsl_reg_2_itk, 'reference_file')
ref_node, ref_out = strat['mean_functional']
workflow.connect(ref_node, ref_out, fsl_reg_2_itk, 'source_file')
itk_imports = ['import os']
change_transform = pe.Node(
util.Function(input_names=['input_affine_file'],
output_names=['updated_affine_file'],
function=change_itk_transform_type,
imports=itk_imports),
name='change_transform_type_{0}'.format(num_strat))
workflow.connect(fsl_reg_2_itk, 'itk_transform', change_transform,
'input_affine_file')
strat.update_resource_pool(
{'fsl_mat_as_itk': (change_transform, 'updated_affine_file')})
strat.append_name(fsl_reg_2_itk.name)
# stack of transforms to be combined to acheive the desired transformation
num_transforms = 5
collect_transforms_key = \
'collect_transforms{0}'.format(inverse_string)
if distcor is True:
num_transforms = 6
collect_transforms_key = \
'collect_transforms{0}{1}'.format('_distcor',
inverse_string)
if collect_transforms_key not in strat:
# handle both symmetric and asymmetric transforms
ants_transformation_dict = {
'asymmetric': {
'anatomical_to_mni_nonlinear_xfm':
'anatomical_to_mni_nonlinear_xfm',
'ants_affine_xfm': 'ants_affine_xfm',
'ants_rigid_xfm': 'ants_rigid_xfm',
'ants_initial_xfm': 'ants_initial_xfm',
'blip_warp': 'blip_warp',
'blip_warp_inverse': 'blip_warp_inverse',
'fsl_mat_as_itk': 'fsl_mat_as_itk',
},
'symmetric': {
'anatomical_to_mni_nonlinear_xfm': 'ants_symm_warp_field',
'ants_affine_xfm': 'ants_symm_affine_xfm',
'ants_rigid_xfm': 'ants_symm_rigid_xfm',
'ants_initial_xfm': 'ants_symm_initial_xfm',
'blip_warp': 'blip_warp',
'blip_warp_inverse': 'blip_warp_inverse',
'fsl_mat_as_itk': 'fsl_mat_as_itk',
}
}
# transforms to be concatenated, the first element of each tuple is
# the resource pool key related to the resource that should be
# connected in, and the second element is the input to which it
# should be connected
if inverse is True:
if distcor is True:
# Field file from anatomical nonlinear registration
transforms_to_combine = [\
('anatomical_to_mni_nonlinear_xfm', 'in6'),
('ants_affine_xfm', 'in5'),
('ants_rigid_xfm', 'in4'),
('ants_initial_xfm', 'in3'),
('fsl_mat_as_itk', 'in2'),
('blip_warp_inverse', 'in1')]
else:
transforms_to_combine = [\
('anatomical_to_mni_nonlinear_xfm', 'in5'),
('ants_affine_xfm', 'in4'),
('ants_rigid_xfm', 'in3'),
('ants_initial_xfm', 'in2'),
('fsl_mat_as_itk', 'in1')]
else:
transforms_to_combine = [\
('anatomical_to_mni_nonlinear_xfm', 'in1'),
('ants_affine_xfm', 'in2'),
('ants_rigid_xfm', 'in3'),
('ants_initial_xfm', 'in4'),
('fsl_mat_as_itk', 'in5')]
if distcor is True:
transforms_to_combine.append(('blip_warp', 'in6'))
# define the node
collect_transforms = pe.Node(util.Merge(num_transforms),
name='collect_transforms{0}_{1}'.format(
inverse_string, num_strat))
# wire in the various tranformations
for transform_key, input_port in transforms_to_combine:
node, out_file = strat[ants_transformation_dict[symmetry]
[transform_key]]
workflow.connect(node, out_file, collect_transforms, input_port)
# set the output
strat.update_resource_pool(
{collect_transforms_key: (collect_transforms, 'out')})
strat.append_name(collect_transforms.name)
#### now we add in the apply ants warps node
if map_node:
apply_ants_warp = pe.MapNode(
interface=ants.ApplyTransforms(),
name='apply_ants_warp_{0}_mapnode{1}_{2}'.format(
output_name, inverse_string, num_strat),
iterfield=['input_image'],
mem_gb=1.5)
else:
apply_ants_warp = pe.Node(interface=ants.ApplyTransforms(),
name='apply_ants_warp_{0}{1}_{2}'.format(
output_name, inverse_string, num_strat),
mem_gb=1.5)
apply_ants_warp.inputs.out_postfix = '_antswarp'
apply_ants_warp.interface.num_threads = int(num_ants_cores)
if inverse is True:
apply_ants_warp.inputs.invert_transform_flags = \
[True, True, True, True, False]
# input_image_type:
# (0 or 1 or 2 or 3)
# Option specifying the input image type of scalar
# (default), vector, tensor, or time series.
apply_ants_warp.inputs.input_image_type = input_image_type
apply_ants_warp.inputs.dimension = 3
apply_ants_warp.inputs.interpolation = interpolation_method
node, out_file = strat[ref_key]
workflow.connect(node, out_file, apply_ants_warp, 'reference_image')
collect_node, collect_out = strat[collect_transforms_key]
workflow.connect(collect_node, collect_out, apply_ants_warp, 'transforms')
workflow.connect(input_node, input_out, apply_ants_warp, 'input_image')
strat.update_resource_pool(
{output_name: (apply_ants_warp, 'output_image')})
strat.append_name(apply_ants_warp.name)
return workflow
def init_epi_reg_wf(freesurfer, bold2t1w_dof,
bold_file_size_gb, output_spaces, output_dir,
name='epi_reg_wf', use_fieldwarp=False):
"""
Uses FSL FLIRT with the BBR cost function to find the transform that
maps the EPI space into the T1-space
"""
workflow = pe.Workflow(name=name)
inputnode = pe.Node(
niu.IdentityInterface(fields=['name_source', 'ref_epi_brain', 'ref_epi_mask',
't1_preproc', 't1_brain', 't1_mask',
't1_seg', 'epi_split', 'hmc_xforms',
'subjects_dir', 'subject_id', 'fs_2_t1_transform',
'fieldwarp']),
name='inputnode'
)
outputnode = pe.Node(
niu.IdentityInterface(fields=['mat_epi_to_t1', 'mat_t1_to_epi',
'itk_epi_to_t1', 'itk_t1_to_epi',
'epi_t1', 'epi_mask_t1', 'fs_reg_file',
'out_report']),
name='outputnode'
)
if freesurfer:
bbregister = pe.Node(
BBRegisterRPT(
dof=bold2t1w_dof,
contrast_type='t2',
init='coreg',
registered_file=True,
out_fsl_file=True,
generate_report=True),
name='bbregister'
)
def apply_fs_transform(fs_2_t1_transform, bbreg_transform):
import os
import numpy as np
out_file = os.path.abspath('transform.mat')
fs_xfm = np.loadtxt(fs_2_t1_transform)
bbrxfm = np.loadtxt(bbreg_transform)
out_xfm = fs_xfm.dot(bbrxfm)
assert np.allclose(out_xfm[3], [0, 0, 0, 1])
out_xfm[3] = [0, 0, 0, 1]
np.savetxt(out_file, out_xfm, fmt='%.12g')
return out_file
transformer = pe.Node(niu.Function(function=apply_fs_transform),
name='transformer', run_without_submitting=True)
else:
wm_mask = pe.Node(niu.Function(function=_extract_wm), name='wm_mask')
flt_bbr_init = pe.Node(FLIRTRPT(generate_report=True, dof=6), name='flt_bbr_init')
flt_bbr = pe.Node(
FLIRTRPT(generate_report=True, cost_func='bbr',
dof=bold2t1w_dof),
name='flt_bbr')
flt_bbr.inputs.schedule = op.join(os.getenv('FSLDIR'),
'etc/flirtsch/bbr.sch')
# make equivalent warp fields
invt_bbr = pe.Node(fsl.ConvertXFM(invert_xfm=True), name='invt_bbr')
# EPI to T1 transform matrix is from fsl, using c3 tools to convert to
# something ANTs will like.
fsl2itk_fwd = pe.Node(c3.C3dAffineTool(fsl2ras=True, itk_transform=True),
name='fsl2itk_fwd')
fsl2itk_inv = pe.Node(c3.C3dAffineTool(fsl2ras=True, itk_transform=True),
name='fsl2itk_inv')
workflow.connect([
(inputnode, fsl2itk_fwd, [('t1_preproc', 'reference_file'),
('ref_epi_brain', 'source_file')]),
(inputnode, fsl2itk_inv, [('ref_epi_brain', 'reference_file'),
('t1_preproc', 'source_file')]),
(invt_bbr, outputnode, [('out_file', 'mat_t1_to_epi')]),
(invt_bbr, fsl2itk_inv, [('out_file', 'transform_file')]),
(fsl2itk_fwd, outputnode, [('itk_transform', 'itk_epi_to_t1')]),
(fsl2itk_inv, outputnode, [('itk_transform', 'itk_t1_to_epi')]),
])
gen_ref = pe.Node(GenerateSamplingReference(), name='gen_ref')
mask_t1w_tfm = pe.Node(
ants.ApplyTransforms(interpolation='NearestNeighbor',
float=True),
name='mask_t1w_tfm'
)
workflow.connect([
(inputnode, gen_ref, [('ref_epi_brain', 'moving_image'),
('t1_brain', 'fixed_image')]),
(gen_ref, mask_t1w_tfm, [('out_file', 'reference_image')]),
(fsl2itk_fwd, mask_t1w_tfm, [('itk_transform', 'transforms')]),
(inputnode, mask_t1w_tfm, [('ref_epi_mask', 'input_image')]),
(mask_t1w_tfm, outputnode, [('output_image', 'epi_mask_t1')])
])
if use_fieldwarp:
merge_transforms = pe.MapNode(niu.Merge(3), iterfield=['in3'],
name='merge_transforms', run_without_submitting=True)
workflow.connect([
(inputnode, merge_transforms, [('fieldwarp', 'in2'),
('hmc_xforms', 'in3')])
])
else:
merge_transforms = pe.MapNode(niu.Merge(2), iterfield=['in2'],
name='merge_transforms', run_without_submitting=True)
workflow.connect([
(inputnode, merge_transforms, [('hmc_xforms', 'in2')])
])
merge = pe.Node(Merge(), name='merge')
merge.interface.estimated_memory_gb = bold_file_size_gb * 3
epi_to_t1w_transform = pe.MapNode(
ants.ApplyTransforms(interpolation="LanczosWindowedSinc",
float=True),
iterfield=['input_image', 'transforms'],
name='epi_to_t1w_transform')
epi_to_t1w_transform.terminal_output = 'file'
workflow.connect([
(fsl2itk_fwd, merge_transforms, [('itk_transform', 'in1')]),
(merge_transforms, epi_to_t1w_transform, [('out', 'transforms')]),
(epi_to_t1w_transform, merge, [('output_image', 'in_files')]),
(inputnode, merge, [('name_source', 'header_source')]),
(merge, outputnode, [('out_file', 'epi_t1')]),
(inputnode, epi_to_t1w_transform, [('epi_split', 'input_image')]),
(gen_ref, epi_to_t1w_transform, [('out_file', 'reference_image')]),
])
if freesurfer:
workflow.connect([
(inputnode, bbregister, [('subjects_dir', 'subjects_dir'),
('subject_id', 'subject_id')]),
(inputnode, bbregister, [('ref_epi_brain', 'source_file')]),
(inputnode, transformer, [('fs_2_t1_transform', 'fs_2_t1_transform')]),
(bbregister, transformer, [('out_fsl_file', 'bbreg_transform')]),
(transformer, invt_bbr, [('out', 'in_file')]),
(transformer, fsl2itk_fwd, [('out', 'transform_file')]),
(transformer, outputnode, [('out', 'mat_epi_to_t1')]),
(bbregister, outputnode, [('out_reg_file', 'fs_reg_file'),
('out_report', 'out_report')]),
])
else:
workflow.connect([
(inputnode, wm_mask, [('t1_seg', 'in_file')]),
(inputnode, flt_bbr_init, [('ref_epi_brain', 'in_file')]),
(inputnode, flt_bbr_init, [('t1_brain', 'reference')]),
(flt_bbr_init, flt_bbr, [('out_matrix_file', 'in_matrix_file')]),
(inputnode, flt_bbr, [('t1_brain', 'reference')]),
(inputnode, flt_bbr, [('ref_epi_brain', 'in_file')]),
(wm_mask, flt_bbr, [('out', 'wm_seg')]),
(flt_bbr, invt_bbr, [('out_matrix_file', 'in_file')]),
(flt_bbr, fsl2itk_fwd, [('out_matrix_file', 'transform_file')]),
(flt_bbr, outputnode, [('out_matrix_file', 'mat_epi_to_t1'),
('out_report', 'out_report')]),
])
return workflow
def create_wf_c3d_fsl_to_itk(map_node, input_image_type=0,
name='create_wf_c3d_fsl_to_itk'):
"""
Converts an FSL-format output matrix to an ITK-format (ANTS) matrix
for use with ANTS registration tools.
Parameters
----------
name : string, optional
Name of the workflow.
Returns
-------
fsl_to_itk_conversion : nipype.pipeline.engine.Workflow
Notes
-----
Workflow Inputs::
inputspec.affine_file : string (nifti file)
Output matrix of FSL-based functional to anatomical registration
inputspec.reference_file : string (nifti file)
File of skull-stripped anatomical brain to be used in affine
conversion
inputspec.source_file : string (nifti file)
Should match the input of the apply warp (in_file) unless you are
applying the warp to a 4-d file, in which case this file should
be a mean_functional file
Workflow Outputs::
outputspec.itk_transform : string (nifti file)
Converted affine transform in ITK format usable with ANTS
"""
import nipype.interfaces.c3 as c3
from nipype.interfaces.utility import Function
from CPAC.registration.utils import change_itk_transform_type
from nipype.interfaces.afni import preprocess
fsl_to_itk_conversion = pe.Workflow(name=name)
inputspec = pe.Node(util.IdentityInterface(fields=['affine_file',
'reference_file', 'source_file']), name='inputspec')
# converts FSL-format .mat affine xfm into ANTS-format .txt
# .mat affine comes from Func->Anat registration
if map_node == 0:
fsl_reg_2_itk = pe.Node(c3.C3dAffineTool(), name='fsl_reg_2_itk')
elif map_node == 1:
fsl_reg_2_itk = pe.MapNode(c3.C3dAffineTool(),
name='fsl_reg_2_itk_mapnode', iterfield=['source_file'])
fsl_reg_2_itk.inputs.itk_transform = True
fsl_reg_2_itk.inputs.fsl2ras = True
itk_imports = ['import os']
if map_node == 0:
change_transform = pe.Node(util.Function(
input_names=['input_affine_file'],
output_names=['updated_affine_file'],
function=change_itk_transform_type,
imports=itk_imports),
name='change_transform_type')
elif map_node == 1:
change_transform = pe.MapNode(util.Function(
input_names=['input_affine_file'],
output_names=['updated_affine_file'],
function=change_itk_transform_type,
imports=itk_imports),
name='change_transform_type', iterfield=['input_affine_file'])
outputspec = pe.Node(util.IdentityInterface(fields=['itk_transform']),
name='outputspec')
fsl_to_itk_conversion.connect(inputspec, 'affine_file', fsl_reg_2_itk,
'transform_file')
fsl_to_itk_conversion.connect(inputspec, 'reference_file', fsl_reg_2_itk,
'reference_file')
# source_file input of the conversion must be a 3D file, so if the source
# file is 4D (input_image_type=3), average it into a 3D file first
if input_image_type == 0:
fsl_to_itk_conversion.connect(inputspec, 'source_file', fsl_reg_2_itk,
'source_file')
elif input_image_type == 3:
try:
tstat_source = pe.Node(interface=preprocess.TStat(),
name='fsl_to_itk_tcat_source')
except AttributeError:
from nipype.interfaces.afni import utils as afni_utils
tstat_source = pe.Node(interface=afni_utils.TStat(),
name='fsl_to_itk_tcat_source')
tstat_source.inputs.outputtype = 'NIFTI_GZ'
tstat_source.inputs.options = '-mean'
fsl_to_itk_conversion.connect(inputspec, 'source_file', tstat_source,
'in_file')
fsl_to_itk_conversion.connect(tstat_source, 'out_file', fsl_reg_2_itk,
'source_file')
fsl_to_itk_conversion.connect(fsl_reg_2_itk, 'itk_transform',
change_transform, 'input_affine_file')
fsl_to_itk_conversion.connect(change_transform, 'updated_affine_file',
outputspec, 'itk_transform')
return fsl_to_itk_conversion
if c in valid_chars) + "_roi.nii.gz"
sphere = pe.Node(fsl.ImageMaths(), name="sphere")
sphere.inputs.out_data_type = "float"
sphere.inputs.op_string = "-kernel sphere 6 -fmean -bin"
wf.connect(point, "out_file", sphere, "in_file")
wf.connect(roi_infosource, ("roi", format_filename), sphere, "out_file")
# mask2func = pe.Node(fsl.ApplyWarp(), name="mask2func")
# mask2func.inputs.datatype = "float"
# wf.connect(normalize, 'invert_warp.inverted_warp_file', mask2func, "field_file")
# wf.connect(datagrabber, "preprocessed_epi", mask2func, "ref_file")
# wf.connect(sphere, "out_file", mask2func, "in_file")
# wf.connect(inv_func2anat, 'out_file', mask2func, "postmat")
func2anat_fsl2itk = pe.Node(c3.C3dAffineTool(), name="func2anat_fsl2itk")
func2anat_fsl2itk.inputs.itk_transform = True
func2anat_fsl2itk.inputs.fsl2ras = True
wf.connect(datagrabber, "func2anat_transform", func2anat_fsl2itk,
"transform_file")
wf.connect(datagrabber, "epi_mask", func2anat_fsl2itk, "source_file")
wf.connect(ants_normalize, "brain_2nii.out_file", func2anat_fsl2itk,
"reference_file")
collect_transforms = pe.Node(util.Merge(3), name='collect_transforms')
wf.connect(func2anat_fsl2itk, "itk_transform", collect_transforms, "in1")
wf.connect(ants_normalize, "outputspec.affine_transformation",
collect_transforms, "in2")
wf.connect(ants_normalize, "outputspec.inverse_warp", collect_transforms,
"in3")
def init_epi_hmc_wf(metadata, bold_file_size_gb, ignore,
name='epi_hmc_wf'):
"""
Performs :abbr:`HMC (head motion correction)` over the input
:abbr:`EPI (echo-planar imaging)` image.
"""
workflow = pe.Workflow(name=name)
inputnode = pe.Node(niu.IdentityInterface(fields=['epi']),
name='inputnode')
outputnode = pe.Node(niu.IdentityInterface(
fields=['xforms', 'epi_hmc', 'epi_split', 'epi_mask', 'ref_image',
'ref_image_brain', 'movpar_file', 'n_volumes_to_discard',
'epi_mask_report']), name='outputnode')
def normalize_motion_func(in_file, format):
import os
import numpy as np
from nipype.utils.misc import normalize_mc_params
mpars = np.loadtxt(in_file) # mpars is N_t x 6
mpars = np.apply_along_axis(func1d=normalize_mc_params,
axis=1, arr=mpars,
source=format)
np.savetxt("motion_params.txt", mpars)
return os.path.abspath("motion_params.txt")
normalize_motion = pe.Node(niu.Function(function=normalize_motion_func),
name="normalize_motion", run_without_submitting=True)
normalize_motion.inputs.format = "FSL"
# Head motion correction (hmc)
hmc = pe.Node(fsl.MCFLIRT(
save_mats=True, save_plots=True), name='EPI_hmc')
hmc.interface.estimated_memory_gb = bold_file_size_gb * 3
hcm2itk = pe.MapNode(c3.C3dAffineTool(fsl2ras=True, itk_transform=True),
iterfield=['transform_file'], name='hcm2itk')
enhance_and_skullstrip_epi_wf = init_enhance_and_skullstrip_epi_wf()
gen_ref = pe.Node(EstimateReferenceImage(), name="gen_ref")
workflow.connect([
(inputnode, gen_ref, [('epi', 'in_file')]),
(gen_ref, enhance_and_skullstrip_epi_wf, [('ref_image', 'inputnode.in_file')]),
(gen_ref, hmc, [('ref_image', 'ref_file')]),
(enhance_and_skullstrip_epi_wf, outputnode, [('outputnode.bias_corrected_file', 'ref_image'),
('outputnode.mask_file', 'epi_mask'),
('outputnode.out_report', 'epi_mask_report'),
('outputnode.skull_stripped_file', 'ref_image_brain')]),
])
split = pe.Node(fsl.Split(dimension='t'), name='split')
split.interface.estimated_memory_gb = bold_file_size_gb * 3
if "SliceTiming" in metadata and 'slicetiming' not in ignore:
LOGGER.info('Slice-timing correction will be included.')
def create_custom_slice_timing_file_func(metadata):
import os
slice_timings = metadata["SliceTiming"]
slice_timings_ms = [str(t) for t in slice_timings]
out_file = "timings.1D"
with open("timings.1D", "w") as fp:
fp.write("\t".join(slice_timings_ms))
return os.path.abspath(out_file)
create_custom_slice_timing_file = pe.Node(
niu.Function(function=create_custom_slice_timing_file_func),
name="create_custom_slice_timing_file", run_without_submitting=True)
create_custom_slice_timing_file.inputs.metadata = metadata
slice_timing_correction = pe.Node(interface=afni.TShift(),
name='slice_timing_correction')
slice_timing_correction.inputs.outputtype = 'NIFTI_GZ'
slice_timing_correction.inputs.tr = str(metadata["RepetitionTime"]) + "s"
def prefix_at(x):
return "@" + x
workflow.connect([
(inputnode, slice_timing_correction, [('epi', 'in_file')]),
(gen_ref, slice_timing_correction, [('n_volumes_to_discard', 'ignore')]),
(create_custom_slice_timing_file, slice_timing_correction, [
(('out', prefix_at), 'tpattern')]),
(slice_timing_correction, hmc, [('out_file', 'in_file')])
])
else:
workflow.connect([
(inputnode, hmc, [('epi', 'in_file')])
])
workflow.connect([
(hmc, hcm2itk, [('mat_file', 'transform_file')]),
(gen_ref, hcm2itk, [('ref_image', 'source_file'),
('ref_image', 'reference_file')]),
(hcm2itk, outputnode, [('itk_transform', 'xforms')]),
(hmc, normalize_motion, [('par_file', 'in_file')]),
(normalize_motion, outputnode, [('out', 'movpar_file')]),
(inputnode, split, [('epi', 'in_file')]),
(split, outputnode, [('out_files', 'epi_split')]),
])
return workflow
def epi_pipeline(name='susceptibility_distortion_correction_using_t1'):
"""
This workflow allows to correct for echo-planareinduced susceptibility artifacts without fieldmap
(e.g. ADNI Database) by elastically register DWIs to their respective baseline T1-weighted
structural scans using an inverse consistent registration algorithm with a mutual information cost
function (SyN algorithm). This workflow allows also a coregistration of DWIs with their respective
baseline T1-weighted structural scans in order to latter combine tracks and cortex parcelation.
.. warning:: This workflow rotates the `b`-vectors'
.. References
.. Nir et al. (Neurobiology of Aging 2015)- Connectivity network measures predict volumetric atrophy in mild cognitive impairment
Leow et al. (IEEE Trans Med Imaging 2007)- Statistical Properties of Jacobian Maps and the Realization of Unbiased Large Deformation Nonlinear Image Registration
Example
-------
>>> epi = epi_pipeline()
>>> epi.inputs.inputnode.DWI = 'DWI.nii'
>>> epi.inputs.inputnode.bvec = 'bvec.txt'
>>> epi.inputs.inputnode.T1 = 'T1.nii'
>>> epi.run() # doctest: +SKIP
"""
from clinica.pipelines.dwi_preprocessing_using_t1.dwi_preprocessing_using_t1_utils import create_jacobian_determinant_image, change_itk_transform_type, expend_matrix_list, rotate_bvecs, ants_registration_syn_quick, ants_warp_image_multi_transform, ants_combin_transform
import nipype.pipeline.engine as pe
import nipype.interfaces.utility as niu
import nipype.interfaces.fsl as fsl
import nipype.interfaces.c3 as c3
inputnode = pe.Node(niu.IdentityInterface(fields=['T1', 'DWI', 'bvec']), name='inputnode')
split = pe.Node(fsl.Split(dimension='t'), name='SplitDWIs')
pick_ref = pe.Node(niu.Select(), name='Pick_b0')
pick_ref.inputs.index = [0]
flirt_b0_2_T1 = pe.Node(interface=fsl.FLIRT(dof=6), name='flirt_B0_2_T1')
flirt_b0_2_T1.inputs.interp = "spline"
flirt_b0_2_T1.inputs.cost = 'normmi'
flirt_b0_2_T1.inputs.cost_func = 'normmi'
apply_xfm = pe.Node(interface=fsl.preprocess.ApplyXFM(), name='apply_xfm')
apply_xfm.inputs.apply_xfm = True
expend_matrix = pe.Node(
interface=niu.Function(
input_names=['in_matrix', 'in_bvec'],
output_names=['out_matrix_list'],
function=expend_matrix_list),
name='expend_matrix')
rot_bvec = pe.Node(
niu.Function(
input_names=['in_matrix', 'in_bvec'],
output_names=['out_file'],
function=rotate_bvecs),
name='Rotate_Bvec')
antsRegistrationSyNQuick = pe.Node(
interface=niu.Function(
input_names=['fix_image', 'moving_image'],
output_names=['image_warped',
'affine_matrix',
'warp',
'inverse_warped',
'inverse_warp'],
function=ants_registration_syn_quick),
name='antsRegistrationSyNQuick')
c3d_flirt2ants = pe.Node(c3.C3dAffineTool(), name='fsl_reg_2_itk')
c3d_flirt2ants.inputs.itk_transform = True
c3d_flirt2ants.inputs.fsl2ras = True
change_transform = pe.Node(niu.Function(
input_names=['input_affine_file'],
output_names=['updated_affine_file'],
function=change_itk_transform_type),
name='change_transform_type')
merge_transform = pe.Node(niu.Merge(3), name='MergeTransforms')
apply_transform = pe.MapNode(interface=niu.Function(input_names=['fix_image', 'moving_image', 'ants_warp_affine'],
output_names=['out_warp_field', 'out_warped'],
function=ants_combin_transform),
iterfield=['moving_image'],
name='warp_filed')
jacobian = pe.MapNode(interface=niu.Function(input_names=['imageDimension', 'deformationField', 'outputImage'],
output_names=['outputImage'],
function=create_jacobian_determinant_image),
iterfield=['deformationField'],
name='jacobian')
jacobian.inputs.imageDimension = 3
jacobian.inputs.outputImage = 'Jacobian_image.nii.gz'
jacmult = pe.MapNode(fsl.MultiImageMaths(op_string='-mul %s'),
iterfield=['in_file', 'operand_files'],
name='ModulateDWIs')
thres = pe.MapNode(fsl.Threshold(thresh=0.0), iterfield=['in_file'],
name='RemoveNegative')
merge = pe.Node(fsl.Merge(dimension='t'), name='MergeDWIs')
outputnode = pe.Node(niu.IdentityInterface(fields=['DWI_2_T1_Coregistration_matrix',
'epi_correction_deformation_field',
'epi_correction_affine_transform',
'epi_correction_image_warped',
'DWIs_epicorrected',
'warp_epi',
'out_bvec'
]), name='outputnode')
wf = pe.Workflow(name='epi_pipeline')
wf.connect([(inputnode, split, [('DWI', 'in_file')])])
wf.connect([(split, pick_ref, [('out_files', 'inlist')])])
wf.connect([(pick_ref, flirt_b0_2_T1, [('out', 'in_file')])])
wf.connect([(inputnode, flirt_b0_2_T1, [('T1', 'reference')])])
wf.connect([(inputnode, rot_bvec, [('bvec', 'in_bvec')])])
wf.connect([(flirt_b0_2_T1, expend_matrix, [('out_matrix_file', 'in_matrix')])])
wf.connect([(inputnode, expend_matrix, [('bvec', 'in_bvec')])])
wf.connect([(expend_matrix, rot_bvec, [('out_matrix_list', 'in_matrix')])])
wf.connect([(inputnode, antsRegistrationSyNQuick, [('T1', 'fix_image')])])
wf.connect([(flirt_b0_2_T1, antsRegistrationSyNQuick, [('out_file', 'moving_image')])])
wf.connect([(inputnode, c3d_flirt2ants, [('T1', 'reference_file')])])
wf.connect([(pick_ref, c3d_flirt2ants, [('out', 'source_file')])])
wf.connect([(flirt_b0_2_T1, c3d_flirt2ants, [('out_matrix_file', 'transform_file')])])
wf.connect([(c3d_flirt2ants, change_transform, [('itk_transform', 'input_affine_file')])])
wf.connect([(antsRegistrationSyNQuick, merge_transform, [('warp', 'in1')])])
wf.connect([(antsRegistrationSyNQuick, merge_transform, [('affine_matrix', 'in2')])])
wf.connect([(change_transform, merge_transform, [('updated_affine_file', 'in3')])])
wf.connect([(inputnode, apply_transform, [('T1', 'fix_image')])])
wf.connect([(split, apply_transform, [('out_files', 'moving_image')])])
wf.connect([(merge_transform, apply_transform, [('out', 'ants_warp_affine')])])
wf.connect([(apply_transform, jacobian, [('out_warp_field', 'deformationField')])])
wf.connect([(apply_transform, jacmult, [('out_warped', 'operand_files')])])
wf.connect([(jacobian, jacmult, [('outputImage', 'in_file')])])
wf.connect([(jacmult, thres, [('out_file', 'in_file')])])
wf.connect([(thres, merge, [('out_file', 'in_files')])])
wf.connect([(merge, outputnode, [('merged_file', 'DWIs_epicorrected')])])
wf.connect([(flirt_b0_2_T1, outputnode, [('out_matrix_file', 'DWI_2_T1_Coregistration_matrix')])])
wf.connect([(antsRegistrationSyNQuick, outputnode, [('warp', 'epi_correction_deformation_field'),
('affine_matrix', 'epi_correction_affine_transform'),
('image_warped', 'epi_correction_image_warped')])])
wf.connect([(merge_transform, outputnode, [('out', 'warp_epi')])])
wf.connect([(rot_bvec, outputnode, [('out_file', 'out_bvec')])])
return wf
(mosaic_suvr_pvc, datasink, [('mosaic', 'QC.@mosaic_suvr_pvc')]),
])
av1451_workflow.connect([
(SUVR_pvc_workflow, MNI_workflow, [('SUVR.out_file',
'transform_suvr_pvc.in_file')]),
])
MNI_workflow.write_graph('MNI.dot', graph2use='colored', simple_form=True)
# 7. MNI SPACE -- deformable alignment onto study-specific template
# placeholder
pet = Node(interface=IdentityInterface(fields=['pet']), name="pet")
# convert the affine transform from PET to MRI obtained via FSL to ANTs format
convert2itk = Node(interface=c3.C3dAffineTool(fsl2ras=True,
itk_transform=True),
name="convert2itk")
# merge affine transform to MRI with composite warp to study-specific template in MNI space
# first should be composite
# second should be pet-to-mri
merge_list = Node(Merge(2), name='merge_list')
merge_list.inputs.in1 = args.mnitransform
warp_pet = Node(
interface=ants.ApplyTransforms(reference_image=blsa_template_mni),
name="warp_pet")
warp_suvr = warp_pet.clone(name='warp_suvr')
mask_pet = Node(interface=fsl.ImageMaths(op_string=' -mul ',
suffix='_mul',
def init_fsl_gebbr_wf(use_bbr,
asl2t1w_dof,
asl2t1w_init,
sloppy=False,
name='fsl_bbr_wf'):
"""
"""
from ...niworkflows.engine.workflows import LiterateWorkflow as Workflow
from ...niworkflows.utils.images import dseg_label as _dseg_label
from ...niworkflows.interfaces.freesurfer import PatchedLTAConvert as LTAConvert
from ...niworkflows.interfaces.registration import FLIRTRPT
workflow = Workflow(name=name)
workflow.__desc__ = """\
The ASL reference was then co-registered to the T1w reference using
`flirt` [FSL {fsl_ver}, @flirt] with the boundary-based registration [@bbr]
cost-function.
Co-registration was configured with nine degrees of freedom to account
for distortions remaining in the ASL reference.
""".format(fsl_ver=FLIRTRPT().version or '<ver>')
inputnode = pe.Node(
niu.IdentityInterface(['in_file', 't1w_dseg',
't1w_brain']), # FLIRT BBR
name='inputnode')
outputnode = pe.Node(niu.IdentityInterface(
['itk_asl_to_t1', 'itk_t1_to_asl', 'out_report', 'fallback']),
name='outputnode')
wm_mask = pe.Node(niu.Function(function=_dseg_label), name='wm_mask')
wm_mask.inputs.label = 2 # BIDS default is WM=2
flt_bbr_init = pe.Node(FLIRTRPT(dof=6,
generate_report=not use_bbr,
uses_qform=True),
name='flt_bbr_init')
if asl2t1w_init not in ("register", "header"):
raise ValueError(
f"Unknown ASL-T1w initialization option: {asl2t1w_init}")
if asl2t1w_init == "header":
raise NotImplementedError(
"Header-based registration initialization not supported for FSL")
invt_bbr = pe.Node(fsl.ConvertXFM(invert_xfm=True),
name='invt_bbr',
mem_gb=DEFAULT_MEMORY_MIN_GB)
# ASL to T1 transform matrix is from fsl, using c3 tools to convert to
# something ANTs will like.
fsl2itk_fwd = pe.Node(c3.C3dAffineTool(fsl2ras=True, itk_transform=True),
name='fsl2itk_fwd',
mem_gb=DEFAULT_MEMORY_MIN_GB)
fsl2itk_inv = pe.Node(c3.C3dAffineTool(fsl2ras=True, itk_transform=True),
name='fsl2itk_inv',
mem_gb=DEFAULT_MEMORY_MIN_GB)
workflow.connect([
(inputnode, flt_bbr_init, [('in_file', 'in_file'),
('t1w_brain', 'reference')]),
(inputnode, fsl2itk_fwd, [('t1w_brain', 'reference_file'),
('in_file', 'source_file')]),
(inputnode, fsl2itk_inv, [('in_file', 'reference_file'),
('t1w_brain', 'source_file')]),
(invt_bbr, fsl2itk_inv, [('out_file', 'transform_file')]),
(fsl2itk_fwd, outputnode, [('itk_transform', 'itk_asl_to_t1')]),
(fsl2itk_inv, outputnode, [('itk_transform', 'itk_t1_to_asl')]),
])
outputnode.inputs.fallback = True
# Short-circuit workflow building, use rigid registration
if use_bbr is False:
workflow.connect([
(flt_bbr_init, invt_bbr, [('out_matrix_file', 'in_file')]),
(flt_bbr_init, fsl2itk_fwd, [('out_matrix_file', 'transform_file')
]),
(flt_bbr_init, outputnode, [('out_report', 'out_report')]),
])
outputnode.inputs.fallback = True
return workflow
flt_bbr = pe.Node(FLIRTRPT(cost_func='bbr',
dof=asl2t1w_dof,
generate_report=True),
name='flt_bbr')
FSLDIR = os.getenv('FSLDIR')
if FSLDIR:
flt_bbr.inputs.schedule = op.join(FSLDIR, 'etc/flirtsch/bbr.sch')
else:
# Should mostly be hit while building docs
LOGGER.warning("FSLDIR unset - using packaged BBR schedule")
flt_bbr.inputs.schedule = pkgr.resource_filename(
'aslprep', 'data/flirtsch/bbr.sch')
workflow.connect([
(inputnode, wm_mask, [('t1w_dseg', 'in_seg')]),
(inputnode, flt_bbr, [('in_file', 'in_file')]),
(flt_bbr_init, flt_bbr, [('out_matrix_file', 'in_matrix_file')]),
])
if sloppy is True:
downsample = pe.Node(niu.Function(
function=_conditional_downsampling,
output_names=["out_file", "out_mask"]),
name='downsample')
workflow.connect([
(inputnode, downsample, [("t1w_brain", "in_file")]),
(wm_mask, downsample, [("out", "in_mask")]),
(downsample, flt_bbr, [('out_file', 'reference'),
('out_mask', 'wm_seg')]),
])
else:
workflow.connect([
(inputnode, flt_bbr, [('t1w_brain', 'reference')]),
(wm_mask, flt_bbr, [('out', 'wm_seg')]),
])
# Short-circuit workflow building, use boundary-based registration
if use_bbr is True:
workflow.connect([
(flt_bbr, invt_bbr, [('out_matrix_file', 'in_file')]),
(flt_bbr, fsl2itk_fwd, [('out_matrix_file', 'transform_file')]),
(flt_bbr, outputnode, [('out_report', 'out_report')]),
])
outputnode.inputs.fallback = False
return workflow
def init_bbreg_wf(use_bbr, bold2t1w_dof, omp_nthreads, name='bbreg_wf'):
"""
Build a workflow to run FreeSurfer's ``bbregister``.
This workflow uses FreeSurfer's ``bbregister`` to register a BOLD image to
a T1-weighted structural image.
It is a counterpart to :py:func:`~fmriprep.workflows.bold.registration.init_fsl_bbr_wf`,
which performs the same task using FSL's FLIRT with a BBR cost function.
The ``use_bbr`` option permits a high degree of control over registration.
If ``False``, standard, affine coregistration will be performed using
FreeSurfer's ``mri_coreg`` tool.
If ``True``, ``bbregister`` will be seeded with the initial transform found
by ``mri_coreg`` (equivalent to running ``bbregister --init-coreg``).
If ``None``, after ``bbregister`` is run, the resulting affine transform
will be compared to the initial transform found by ``mri_coreg``.
Excessive deviation will result in rejecting the BBR refinement and
accepting the original, affine registration.
Workflow Graph
.. workflow ::
:graph2use: orig
:simple_form: yes
from fmriprep.workflows.bold.registration import init_bbreg_wf
wf = init_bbreg_wf(use_bbr=True, bold2t1w_dof=9, omp_nthreads=1)
Parameters
----------
use_bbr : bool or None
Enable/disable boundary-based registration refinement.
If ``None``, test BBR result for distortion before accepting.
bold2t1w_dof : 6, 9 or 12
Degrees-of-freedom for BOLD-T1w registration
name : str, optional
Workflow name (default: bbreg_wf)
Inputs
------
in_file
Reference BOLD image to be registered
fsnative2t1w_xfm
FSL-style affine matrix translating from FreeSurfer T1.mgz to T1w
subjects_dir
FreeSurfer SUBJECTS_DIR
subject_id
FreeSurfer subject ID (must have folder in SUBJECTS_DIR)
t1w_brain
Unused (see :py:func:`~fmriprep.workflows.bold.registration.init_fsl_bbr_wf`)
t1w_dseg
Unused (see :py:func:`~fmriprep.workflows.bold.registration.init_fsl_bbr_wf`)
Outputs
-------
itk_bold_to_t1
Affine transform from ``ref_bold_brain`` to T1 space (ITK format)
itk_t1_to_bold
Affine transform from T1 space to BOLD space (ITK format)
out_report
Reportlet for assessing registration quality
fallback
Boolean indicating whether BBR was rejected (mri_coreg registration returned)
"""
workflow = Workflow(name=name)
workflow.__desc__ = """\
The BOLD reference was then co-registered to the T1w reference using
`bbregister` (FreeSurfer) which implements boundary-based registration [@bbr].
Co-registration was configured with {dof} degrees of freedom{reason}.
""".format(dof={6: 'six', 9: 'nine', 12: 'twelve'}[bold2t1w_dof],
reason='' if bold2t1w_dof == 6 else
'to account for distortions remaining in the BOLD reference')
inputnode = pe.Node(
niu.IdentityInterface([
'in_file',
'fsnative2t1w_xfm', 'subjects_dir', 'subject_id', # BBRegister
't1w_dseg', 't1w_brain']), # FLIRT BBR
name='inputnode')
outputnode = pe.Node(
niu.IdentityInterface(['itk_bold_to_t1', 'itk_t1_to_bold', 'out_report', 'fallback']),
name='outputnode')
mri_coreg = pe.Node(
MRICoregRPT(dof=bold2t1w_dof, sep=[4], ftol=0.0001, linmintol=0.01,
generate_report=not use_bbr),
name='mri_coreg', n_procs=omp_nthreads, mem_gb=5)
lta_concat = pe.Node(ConcatenateLTA(out_file='out.lta'), name='lta_concat')
# XXX LTA-FSL-ITK may ultimately be able to be replaced with a straightforward
# LTA-ITK transform, but right now the translation parameters are off.
lta2fsl_fwd = pe.Node(LTAConvert(out_fsl=True), name='lta2fsl_fwd')
lta2fsl_inv = pe.Node(LTAConvert(out_fsl=True, invert=True), name='lta2fsl_inv')
fsl2itk_fwd = pe.Node(c3.C3dAffineTool(fsl2ras=True, itk_transform=True),
name='fsl2itk_fwd', mem_gb=DEFAULT_MEMORY_MIN_GB)
fsl2itk_inv = pe.Node(c3.C3dAffineTool(fsl2ras=True, itk_transform=True),
name='fsl2itk_inv', mem_gb=DEFAULT_MEMORY_MIN_GB)
workflow.connect([
(inputnode, mri_coreg, [('subjects_dir', 'subjects_dir'),
('subject_id', 'subject_id'),
('in_file', 'source_file')]),
# Output ITK transforms
(inputnode, lta_concat, [('fsnative2t1w_xfm', 'in_lta2')]),
(lta_concat, lta2fsl_fwd, [('out_file', 'in_lta')]),
(lta_concat, lta2fsl_inv, [('out_file', 'in_lta')]),
(inputnode, fsl2itk_fwd, [('t1w_brain', 'reference_file'),
('in_file', 'source_file')]),
(inputnode, fsl2itk_inv, [('in_file', 'reference_file'),
('t1w_brain', 'source_file')]),
(lta2fsl_fwd, fsl2itk_fwd, [('out_fsl', 'transform_file')]),
(lta2fsl_inv, fsl2itk_inv, [('out_fsl', 'transform_file')]),
(fsl2itk_fwd, outputnode, [('itk_transform', 'itk_bold_to_t1')]),
(fsl2itk_inv, outputnode, [('itk_transform', 'itk_t1_to_bold')]),
])
# Short-circuit workflow building, use initial registration
if use_bbr is False:
workflow.connect([
(mri_coreg, outputnode, [('out_report', 'out_report')]),
(mri_coreg, lta_concat, [('out_lta_file', 'in_lta1')])])
outputnode.inputs.fallback = True
return workflow
bbregister = pe.Node(
BBRegisterRPT(dof=bold2t1w_dof, contrast_type='t2', registered_file=True,
out_lta_file=True, generate_report=True),
name='bbregister', mem_gb=12)
workflow.connect([
(inputnode, bbregister, [('subjects_dir', 'subjects_dir'),
('subject_id', 'subject_id'),
('in_file', 'source_file')]),
(mri_coreg, bbregister, [('out_lta_file', 'init_reg_file')]),
])
# Short-circuit workflow building, use boundary-based registration
if use_bbr is True:
workflow.connect([
(bbregister, outputnode, [('out_report', 'out_report')]),
(bbregister, lta_concat, [('out_lta_file', 'in_lta1')])])
outputnode.inputs.fallback = False
return workflow
transforms = pe.Node(niu.Merge(2), run_without_submitting=True, name='transforms')
reports = pe.Node(niu.Merge(2), run_without_submitting=True, name='reports')
lta_ras2ras = pe.MapNode(LTAConvert(out_lta=True), iterfield=['in_lta'],
name='lta_ras2ras', mem_gb=2)
compare_transforms = pe.Node(niu.Function(function=compare_xforms), name='compare_transforms')
select_transform = pe.Node(niu.Select(), run_without_submitting=True, name='select_transform')
select_report = pe.Node(niu.Select(), run_without_submitting=True, name='select_report')
workflow.connect([
(bbregister, transforms, [('out_lta_file', 'in1')]),
(mri_coreg, transforms, [('out_lta_file', 'in2')]),
# Normalize LTA transforms to RAS2RAS (inputs are VOX2VOX) and compare
(transforms, lta_ras2ras, [('out', 'in_lta')]),
(lta_ras2ras, compare_transforms, [('out_lta', 'lta_list')]),
(compare_transforms, outputnode, [('out', 'fallback')]),
# Select output transform
(transforms, select_transform, [('out', 'inlist')]),
(compare_transforms, select_transform, [('out', 'index')]),
(select_transform, lta_concat, [('out', 'in_lta1')]),
# Select output report
(bbregister, reports, [('out_report', 'in1')]),
(mri_coreg, reports, [('out_report', 'in2')]),
(reports, select_report, [('out', 'inlist')]),
(compare_transforms, select_report, [('out', 'index')]),
(select_report, outputnode, [('out', 'out_report')]),
])
return workflow
def ants_apply_warps_func_mni(workflow,
output_name,
func_key,
ref_key,
num_strat,
strat,
interpolation_method='LanczosWindowedSinc',
distcor=False,
map_node=False,
inverse=False,
symmetry='asymmetric',
input_image_type=0,
num_ants_cores=1,
registration_template='t1',
func_type='non-ica-aroma',
num_cpus=1):
"""
Applies previously calculated ANTS registration transforms to input
images. This workflow employs the antsApplyTransforms tool:
http://stnava.github.io/ANTs/
Parameters
----------
name : string, optional
Name of the workflow.
Returns
-------
apply_ants_warp_wf : nipype.pipeline.engine.Workflow
Notes
-----
Workflow Inputs::
workflow: Nipype workflow object
the workflow containing the resources involved
output_name: str
what the name of the warped functional should be when written to the
resource pool
func_key: string
resource pool key correspoding to the node containing the 3D or 4D
functional file to be written into MNI space, use 'leaf' for a
leaf node
ref_key: string
resource pool key correspoding to the file path to the template brain
used for functional-to-template registration
num_strat: int
the number of strategy objects
strat: C-PAC Strategy object
a strategy with one or more resource pools
interpolation_method: str
which interpolation to use when applying the warps, commonly used
options are 'Linear', 'Bspline', 'LanczosWindowedSinc' (default)
for derivatives and image data 'NearestNeighbor' for masks
distcor: boolean
indicates whether a distortion correction transformation should be
added to the transforms, this of course requires that a distortion
correction map exist in the resource pool
map_node: boolean
indicates whether a mapnode should be used, if TRUE func_key is
expected to correspond to a list of resources that should each
be written into standard space with the other parameters
inverse: boolean
writes the invrse of the transform, i.e. MNI->EPI instead of
EPI->MNI
input_image_type: int
argument taken by the ANTs apply warp tool; in this case, should be
0 for scalars (default) and 3 for 4D functional time-series
num_ants_cores: int
the number of CPU cores dedicated to ANTS anatomical-to-standard
registration
registration_template: str
which template to use as a target for the apply warps ('t1' or 'epi'),
should be the same as the target used in the warp calculation
(registration)
func_type: str
'non-ica-aroma' or 'ica-aroma' - how to handle the functional time series
based on the particular demands of ICA-AROMA processed time series
num_cpus: int
the number of CPUs dedicated to each participant workflow - this is
used to determine how to parallelize the warp application step
Workflow Outputs::
outputspec.output_image : string (nifti file)
Normalized output file
Workflow Graph:
.. image::
:width: 500
Detailed Workflow Graph:
.. image::
:width: 500
Apply the functional-to-structural and structural-to-template warps to
the 4D functional time-series to warp it to template space.
Parameters
----------
"""
# if the input is a string, assume that it is resource pool key,
# if it is a tuple, assume that it is a node, outfile pair,
# otherwise, something funky is going on
if isinstance(func_key, str):
if func_key == "leaf":
input_node, input_out = strat.get_leaf_properties()
else:
input_node, input_out = strat[func_key]
elif isinstance(func_key, tuple):
input_node, input_out = func_key
if isinstance(ref_key, str):
ref_node, ref_out = strat[ref_key]
elif isinstance(ref_key, tuple):
ref_node, ref_out = func_key
# when inverse is enabled, we want to update the name of various
# nodes so that we know they were inverted
inverse_string = ''
if inverse is True:
inverse_string = '_inverse'
# make sure that resource pool has some required resources before proceeding
if 'fsl_mat_as_itk' not in strat and registration_template == 't1':
fsl_reg_2_itk = pe.Node(c3.C3dAffineTool(),
name='fsl_reg_2_itk_{0}'.format(num_strat))
fsl_reg_2_itk.inputs.itk_transform = True
fsl_reg_2_itk.inputs.fsl2ras = True
# convert the .mat from linear Func->Anat to
# ANTS format
node, out_file = strat['functional_to_anat_linear_xfm']
workflow.connect(node, out_file, fsl_reg_2_itk, 'transform_file')
node, out_file = strat['anatomical_brain']
workflow.connect(node, out_file, fsl_reg_2_itk, 'reference_file')
ref_node, ref_out = strat['mean_functional']
workflow.connect(ref_node, ref_out, fsl_reg_2_itk, 'source_file')
itk_imports = ['import os']
change_transform = pe.Node(
util.Function(input_names=['input_affine_file'],
output_names=['updated_affine_file'],
function=change_itk_transform_type,
imports=itk_imports),
name='change_transform_type_{0}'.format(num_strat))
workflow.connect(fsl_reg_2_itk, 'itk_transform', change_transform,
'input_affine_file')
strat.update_resource_pool(
{'fsl_mat_as_itk': (change_transform, 'updated_affine_file')})
strat.append_name(fsl_reg_2_itk.name)
# stack of transforms to be combined to acheive the desired transformation
num_transforms = 5
collect_transforms_key = \
'collect_transforms{0}'.format(inverse_string)
if distcor is True and func_type not in 'ica-aroma':
num_transforms = 6
collect_transforms_key = \
'collect_transforms{0}{1}'.format('_distcor',
inverse_string)
if collect_transforms_key not in strat:
if registration_template == 't1':
# handle both symmetric and asymmetric transforms
ants_transformation_dict = {
'asymmetric': {
'anatomical_to_mni_nonlinear_xfm':
'anatomical_to_mni_nonlinear_xfm',
'mni_to_anatomical_nonlinear_xfm':
'mni_to_anatomical_nonlinear_xfm',
'ants_affine_xfm': 'ants_affine_xfm',
'ants_rigid_xfm': 'ants_rigid_xfm',
'ants_initial_xfm': 'ants_initial_xfm',
'blip_warp': 'blip_warp',
'blip_warp_inverse': 'blip_warp_inverse',
'fsl_mat_as_itk': 'fsl_mat_as_itk',
},
'symmetric': {
'anatomical_to_mni_nonlinear_xfm':
'anatomical_to_symmetric_mni_nonlinear_xfm',
'mni_to_anatomical_nonlinear_xfm':
'symmetric_mni_to_anatomical_nonlinear_xfm',
'ants_affine_xfm': 'ants_symmetric_affine_xfm',
'ants_rigid_xfm': 'ants_symmetric_rigid_xfm',
'ants_initial_xfm': 'ants_symmetric_initial_xfm',
'blip_warp': 'blip_warp',
'blip_warp_inverse': 'blip_warp_inverse',
'fsl_mat_as_itk': 'fsl_mat_as_itk',
}
}
# transforms to be concatenated, the first element of each tuple is
# the resource pool key related to the resource that should be
# connected in, and the second element is the input to which it
# should be connected
if inverse is True:
if distcor is True and func_type not in 'ica-aroma':
# Field file from anatomical nonlinear registration
transforms_to_combine = [\
('mni_to_anatomical_nonlinear_xfm', 'in6'),
('ants_affine_xfm', 'in5'),
('ants_rigid_xfm', 'in4'),
('ants_initial_xfm', 'in3'),
('fsl_mat_as_itk', 'in2'),
('blip_warp_inverse', 'in1')]
else:
transforms_to_combine = [\
('mni_to_anatomical_nonlinear_xfm', 'in5'),
('ants_affine_xfm', 'in4'),
('ants_rigid_xfm', 'in3'),
('ants_initial_xfm', 'in2'),
('fsl_mat_as_itk', 'in1')]
else:
transforms_to_combine = [\
('anatomical_to_mni_nonlinear_xfm', 'in1'),
('ants_affine_xfm', 'in2'),
('ants_rigid_xfm', 'in3'),
('ants_initial_xfm', 'in4'),
('fsl_mat_as_itk', 'in5')]
if distcor is True and func_type not in 'ica-aroma':
transforms_to_combine.append(('blip_warp', 'in6'))
if registration_template == 'epi':
# handle both symmetric and asymmetric transforms
ants_transformation_dict = {
'asymmetric': {
'func_to_epi_nonlinear_xfm': 'func_to_epi_nonlinear_xfm',
'epi_to_func_nonlinear_xfm': 'epi_to_func_nonlinear_xfm',
'func_to_epi_ants_affine_xfm':
'func_to_epi_ants_affine_xfm',
'func_to_epi_ants_rigid_xfm': 'func_to_epi_ants_rigid_xfm',
'func_to_epi_ants_initial_xfm':
'func_to_epi_ants_initial_xfm',
# 'blip_warp': 'blip_warp',
# 'blip_warp_inverse': 'blip_warp_inverse',
# 'fsl_mat_as_itk': 'fsl_mat_as_itk',
},
# 'symmetric': {
# 'func_to_epi_nonlinear_xfm': 'anatomical_to_mni_nonlinear_xfm',
# 'func_to_epi_ants_affine_xfm': 'func_to_epi_ants_affine_xfm',
# 'func_to_epi_ants_rigid_xfm': 'func_to_epi_ants_rigid_xfm',
# 'func_to_epi_ants_initial_xfm': 'ants_initial_xfm',
# 'blip_warp': 'blip_warp',
# 'blip_warp_inverse': 'blip_warp_inverse',
# 'fsl_mat_as_itk': 'fsl_mat_as_itk',
# }
}
# transforms to be concatenated, the first element of each tuple is
# the resource pool key related to the resource that should be
# connected in, and the second element is the input to which it
# should be connected
if inverse is True:
if distcor is True and func_type not in 'ica-aroma':
# Field file from anatomical nonlinear registration
transforms_to_combine = [\
('epi_to_func_nonlinear_xfm', 'in4'),
('func_to_epi_ants_affine_xfm', 'in3'),
('func_to_epi_ants_rigid_xfm', 'in2'),
('func_to_epi_ants_initial_xfm', 'in1')]
else:
transforms_to_combine = [\
('epi_to_func_nonlinear_xfm', 'in4'),
('func_to_epi_ants_affine_xfm', 'in3'),
('func_to_epi_ants_rigid_xfm', 'in2'),
('func_to_epi_ants_initial_xfm', 'in1')]
else:
transforms_to_combine = [\
('func_to_epi_nonlinear_xfm', 'in1'),
('func_to_epi_ants_affine_xfm', 'in2'),
('func_to_epi_ants_rigid_xfm', 'in3'),
('func_to_epi_ants_initial_xfm', 'in4')]
# define the node
collect_transforms = pe.Node(
util.Merge(num_transforms),
name='collect_transforms_{0}_{1}_{2}_{3}'.format(
output_name, inverse_string, registration_template, num_strat))
# wire in the various transformations
for transform_key, input_port in transforms_to_combine:
try:
node, out_file = strat[ants_transformation_dict[symmetry]
[transform_key]]
except KeyError:
raise Exception(locals())
workflow.connect(node, out_file, collect_transforms, input_port)
# check transform list (if missing any init/rig/affine) and exclude Nonetype
check_transform = pe.Node(
util.Function(
input_names=['transform_list'],
output_names=['checked_transform_list', 'list_length'],
function=check_transforms),
name='check_transforms{0}_{1}_{2}_{3}'.format(
output_name, inverse_string, registration_template, num_strat))
workflow.connect(collect_transforms, 'out', check_transform,
'transform_list')
# generate inverse transform flags, which depends on the number of transforms
inverse_transform_flags = pe.Node(
util.Function(input_names=['transform_list'],
output_names=['inverse_transform_flags'],
function=generate_inverse_transform_flags),
name='inverse_transform_flags_{0}_{1}_{2}_{3}'.format(
output_name, inverse_string, registration_template, num_strat))
workflow.connect(check_transform, 'checked_transform_list',
inverse_transform_flags, 'transform_list')
# set the output
strat.update_resource_pool({
collect_transforms_key: (check_transform, 'checked_transform_list')
})
strat.append_name(check_transform.name)
strat.append_name(inverse_transform_flags.name)
#### now we add in the apply ants warps node
if int(num_cpus) > 1 and input_image_type == 3:
# parallelize time series warp application
map_node = True
if map_node:
apply_ants_warp = pe.MapNode(
interface=ants.ApplyTransforms(),
name='apply_ants_warp_{0}_mapnode_{1}_{2}_{3}'.format(
output_name, inverse_string, registration_template, num_strat),
iterfield=['input_image'],
mem_gb=10.0)
else:
apply_ants_warp = pe.Node(
interface=ants.ApplyTransforms(),
name='apply_ants_warp_{0}_{1}_{2}_{3}'.format(
output_name, inverse_string, registration_template, num_strat),
mem_gb=10.0)
apply_ants_warp.inputs.out_postfix = '_antswarp'
apply_ants_warp.interface.num_threads = int(num_ants_cores)
if inverse is True:
workflow.connect(inverse_transform_flags, 'inverse_transform_flags',
apply_ants_warp, 'invert_transform_flags')
# input_image_type:
# (0 or 1 or 2 or 3)
# Option specifying the input image type of scalar
# (default), vector, tensor, or time series.
apply_ants_warp.inputs.input_image_type = input_image_type
apply_ants_warp.inputs.dimension = 3
apply_ants_warp.inputs.interpolation = interpolation_method
node, out_file = strat[ref_key]
workflow.connect(node, out_file, apply_ants_warp, 'reference_image')
collect_node, collect_out = strat[collect_transforms_key]
workflow.connect(collect_node, collect_out, apply_ants_warp, 'transforms')
if output_name == "functional_to_standard":
# write out the composite functional to standard transforms
write_composite_xfm = pe.Node(
interface=ants.ApplyTransforms(),
name='write_composite_xfm_{0}_{1}_{2}_{3}'.format(
output_name, inverse_string, registration_template, num_strat),
mem_gb=8.0)
write_composite_xfm.inputs.print_out_composite_warp_file = True
write_composite_xfm.inputs.output_image = "func_to_standard_xfm.nii.gz"
workflow.connect(input_node, input_out, write_composite_xfm,
'input_image')
write_composite_xfm.inputs.input_image_type = input_image_type
write_composite_xfm.inputs.dimension = 3
write_composite_xfm.inputs.interpolation = interpolation_method
node, out_file = strat[ref_key]
workflow.connect(node, out_file, write_composite_xfm,
'reference_image')
collect_node, collect_out = strat[collect_transforms_key]
workflow.connect(collect_node, collect_out, write_composite_xfm,
'transforms')
# write_composite_inv_xfm = pe.Node(
# interface=ants.ApplyTransforms(),
# name='write_composite_xfm_{0}_{1}_{2}_{3}'.format(output_name,
# '_inverse', registration_template, num_strat), mem_gb=1.5)
# write_composite_inv_xfm.inputs.print_out_composite_warp_file = True
# write_composite_inv_xfm.inputs.output_image = "func_to_standard_inverse-xfm.nii.gz"
#
# workflow.connect(input_node, input_out,
# write_composite_inv_xfm, 'input_image')
#
# workflow.connect(inverse_transform_flags, 'inverse_transform_flags',
# write_composite_inv_xfm, 'invert_transform_flags')
#
#
# write_composite_inv_xfm.inputs.input_image_type = input_image_type
# write_composite_inv_xfm.inputs.dimension = 3
# write_composite_inv_xfm.inputs.interpolation = interpolation_method
#
# node, out_file = strat[ref_key]
# workflow.connect(node, out_file,
# write_composite_inv_xfm, 'reference_image')
#
# collect_node, collect_out = strat[collect_transforms_key]
# workflow.connect(collect_node, collect_out,
# write_composite_inv_xfm, 'transforms')
strat.update_resource_pool({
"functional_to_standard_xfm": (write_composite_xfm, 'output_image')
})
#"functional_to_standard_inverse-xfm": (write_composite_inv_xfm, 'output_image')
#})
# parallelize the apply warp, if multiple CPUs, and it's a time series!
if int(num_cpus) > 1 and input_image_type == 3:
node_id = f'_{output_name}_{inverse_string}_{registration_template}_{num_strat}'
chunk_imports = ['import nibabel as nb']
chunk = pe.Node(Function(input_names=['func_file', 'n_cpus'],
output_names=['TR_ranges'],
function=chunk_ts,
imports=chunk_imports),
name=f'chunk_{node_id}')
chunk.inputs.n_cpus = int(num_cpus)
workflow.connect(input_node, input_out, chunk, 'func_file')
split_imports = ['import os', 'import subprocess']
split = pe.Node(Function(input_names=['func_file', 'tr_ranges'],
output_names=['split_funcs'],
function=split_ts_chunks,
imports=split_imports),
name=f'split_{node_id}')
workflow.connect(input_node, input_out, split, 'func_file')
workflow.connect(chunk, 'TR_ranges', split, 'tr_ranges')
workflow.connect(split, 'split_funcs', apply_ants_warp, 'input_image')
func_concat = pe.Node(interface=afni_utils.TCat(),
name=f'func_concat_{node_id}')
func_concat.inputs.outputtype = 'NIFTI_GZ'
workflow.connect(apply_ants_warp, 'output_image', func_concat,
'in_files')
strat.update_resource_pool({output_name: (func_concat, 'out_file')})
else:
workflow.connect(input_node, input_out, apply_ants_warp, 'input_image')
strat.update_resource_pool(
{output_name: (apply_ants_warp, 'output_image')})
strat.append_name(apply_ants_warp.name)
return workflow
def epi_pipeline(name="susceptibility_distortion_correction_using_t1"):
"""
This workflow allows to correct for echo-planareinduced susceptibility artifacts without fieldmap
(e.g. ADNI Database) by elastically register DWIs to their respective baseline T1-weighted
structural scans using an inverse consistent registration algorithm with a mutual information cost
function (SyN algorithm). This workflow allows also a coregistration of DWIs with their respective
baseline T1-weighted structural scans in order to latter combine tracks and cortex parcelation.
.. warning:: This workflow rotates the `b`-vectors'
.. References
.. Nir et al. (Neurobiology of Aging 2015)- Connectivity network measures predict volumetric atrophy in mild cognitive impairment
Leow et al. (IEEE Trans Med Imaging 2007)- Statistical Properties of Jacobian Maps and the Realization of Unbiased Large Deformation Nonlinear Image Registration
Example
-------
>>> epi = epi_pipeline()
>>> epi.inputs.inputnode.DWI = 'DWI.nii'
>>> epi.inputs.inputnode.bvec = 'bvec.txt'
>>> epi.inputs.inputnode.T1 = 'T1.nii'
>>> epi.run() # doctest: +SKIP
"""
import nipype.interfaces.c3 as c3
import nipype.interfaces.fsl as fsl
import nipype.interfaces.utility as niu
import nipype.pipeline.engine as pe
from clinica.pipelines.dwi_preprocessing_using_t1.dwi_preprocessing_using_t1_utils import (
ants_combin_transform,
ants_registration_syn_quick,
ants_warp_image_multi_transform,
change_itk_transform_type,
create_jacobian_determinant_image,
expend_matrix_list,
rotate_bvecs,
)
inputnode = pe.Node(niu.IdentityInterface(fields=["T1", "DWI", "bvec"]),
name="inputnode")
split = pe.Node(fsl.Split(dimension="t"), name="SplitDWIs")
pick_ref = pe.Node(niu.Select(), name="Pick_b0")
pick_ref.inputs.index = [0]
flirt_b0_2_T1 = pe.Node(interface=fsl.FLIRT(dof=6), name="flirt_B0_2_T1")
flirt_b0_2_T1.inputs.interp = "spline"
flirt_b0_2_T1.inputs.cost = "normmi"
flirt_b0_2_T1.inputs.cost_func = "normmi"
apply_xfm = pe.Node(interface=fsl.preprocess.ApplyXFM(), name="apply_xfm")
apply_xfm.inputs.apply_xfm = True
expend_matrix = pe.Node(
interface=niu.Function(
input_names=["in_matrix", "in_bvec"],
output_names=["out_matrix_list"],
function=expend_matrix_list,
),
name="expend_matrix",
)
rot_bvec = pe.Node(
niu.Function(
input_names=["in_matrix", "in_bvec"],
output_names=["out_file"],
function=rotate_bvecs,
),
name="Rotate_Bvec",
)
antsRegistrationSyNQuick = pe.Node(
interface=niu.Function(
input_names=["fix_image", "moving_image"],
output_names=[
"image_warped",
"affine_matrix",
"warp",
"inverse_warped",
"inverse_warp",
],
function=ants_registration_syn_quick,
),
name="antsRegistrationSyNQuick",
)
c3d_flirt2ants = pe.Node(c3.C3dAffineTool(), name="fsl_reg_2_itk")
c3d_flirt2ants.inputs.itk_transform = True
c3d_flirt2ants.inputs.fsl2ras = True
change_transform = pe.Node(
niu.Function(
input_names=["input_affine_file"],
output_names=["updated_affine_file"],
function=change_itk_transform_type,
),
name="change_transform_type",
)
merge_transform = pe.Node(niu.Merge(3), name="MergeTransforms")
apply_transform = pe.MapNode(
interface=niu.Function(
input_names=["fix_image", "moving_image", "ants_warp_affine"],
output_names=["out_warp_field", "out_warped"],
function=ants_combin_transform,
),
iterfield=["moving_image"],
name="warp_filed",
)
jacobian = pe.MapNode(
interface=niu.Function(
input_names=["imageDimension", "deformationField", "outputImage"],
output_names=["outputImage"],
function=create_jacobian_determinant_image,
),
iterfield=["deformationField"],
name="jacobian",
)
jacobian.inputs.imageDimension = 3
jacobian.inputs.outputImage = "Jacobian_image.nii.gz"
jacmult = pe.MapNode(
fsl.MultiImageMaths(op_string="-mul %s"),
iterfield=["in_file", "operand_files"],
name="ModulateDWIs",
)
thres = pe.MapNode(fsl.Threshold(thresh=0.0),
iterfield=["in_file"],
name="RemoveNegative")
merge = pe.Node(fsl.Merge(dimension="t"), name="MergeDWIs")
outputnode = pe.Node(
niu.IdentityInterface(fields=[
"DWI_2_T1_Coregistration_matrix",
"epi_correction_deformation_field",
"epi_correction_affine_transform",
"epi_correction_image_warped",
"DWIs_epicorrected",
"warp_epi",
"out_bvec",
]),
name="outputnode",
)
wf = pe.Workflow(name="epi_pipeline")
wf.connect([(inputnode, split, [("DWI", "in_file")])])
wf.connect([(split, pick_ref, [("out_files", "inlist")])])
wf.connect([(pick_ref, flirt_b0_2_T1, [("out", "in_file")])])
wf.connect([(inputnode, flirt_b0_2_T1, [("T1", "reference")])])
wf.connect([(inputnode, rot_bvec, [("bvec", "in_bvec")])])
wf.connect([(flirt_b0_2_T1, expend_matrix, [("out_matrix_file",
"in_matrix")])])
wf.connect([(inputnode, expend_matrix, [("bvec", "in_bvec")])])
wf.connect([(expend_matrix, rot_bvec, [("out_matrix_list", "in_matrix")])])
wf.connect([(inputnode, antsRegistrationSyNQuick, [("T1", "fix_image")])])
wf.connect([(flirt_b0_2_T1, antsRegistrationSyNQuick, [("out_file",
"moving_image")])])
wf.connect([(inputnode, c3d_flirt2ants, [("T1", "reference_file")])])
wf.connect([(pick_ref, c3d_flirt2ants, [("out", "source_file")])])
wf.connect([(flirt_b0_2_T1, c3d_flirt2ants, [("out_matrix_file",
"transform_file")])])
wf.connect([(c3d_flirt2ants, change_transform, [("itk_transform",
"input_affine_file")])])
wf.connect([(antsRegistrationSyNQuick, merge_transform, [("warp", "in1")])
])
wf.connect([(antsRegistrationSyNQuick, merge_transform, [("affine_matrix",
"in2")])])
wf.connect([(change_transform, merge_transform, [("updated_affine_file",
"in3")])])
wf.connect([(inputnode, apply_transform, [("T1", "fix_image")])])
wf.connect([(split, apply_transform, [("out_files", "moving_image")])])
wf.connect([(merge_transform, apply_transform, [("out", "ants_warp_affine")
])])
wf.connect([(apply_transform, jacobian, [("out_warp_field",
"deformationField")])])
wf.connect([(apply_transform, jacmult, [("out_warped", "operand_files")])])
wf.connect([(jacobian, jacmult, [("outputImage", "in_file")])])
wf.connect([(jacmult, thres, [("out_file", "in_file")])])
wf.connect([(thres, merge, [("out_file", "in_files")])])
wf.connect([(merge, outputnode, [("merged_file", "DWIs_epicorrected")])])
wf.connect([(
flirt_b0_2_T1,
outputnode,
[("out_matrix_file", "DWI_2_T1_Coregistration_matrix")],
)])
wf.connect([(
antsRegistrationSyNQuick,
outputnode,
[
("warp", "epi_correction_deformation_field"),
("affine_matrix", "epi_correction_affine_transform"),
("image_warped", "epi_correction_image_warped"),
],
)])
wf.connect([(merge_transform, outputnode, [("out", "warp_epi")])])
wf.connect([(rot_bvec, outputnode, [("out_file", "out_bvec")])])
return wf
def epi_mean_t1_registration(name='EPIMeanNormalization', settings=None):
"""
Uses FSL FLIRT with the BBR cost function to find the transform that
maps the EPI space into the T1-space
"""
workflow = pe.Workflow(name=name)
inputnode = pe.Node(niu.IdentityInterface(
fields=['epi', 'epi_mean', 't1_brain', 't1_seg']),
name='inputnode')
outputnode = pe.Node(niu.IdentityInterface(fields=['mat_epi_to_t1']),
name='outputnode')
# Extract wm mask from segmentation
wm_mask = pe.Node(niu.Function(input_names=['in_file'],
output_names=['out_file'],
function=_extract_wm),
name='WM_mask')
flt_bbr_init = pe.Node(fsl.FLIRT(dof=6, out_matrix_file='init.mat'),
name='Flirt_BBR_init')
flt_bbr = pe.Node(fsl.FLIRT(dof=6, cost_func='bbr'), name='Flirt_BBR')
flt_bbr.inputs.schedule = op.join(os.getenv('FSLDIR'),
'etc/flirtsch/bbr.sch')
# make equivalent warp fields
invt_bbr = pe.Node(fsl.ConvertXFM(invert_xfm=True), name='Flirt_BBR_Inv')
# EPI to T1 transform matrix is from fsl, using c3 tools to convert to
# something ANTs will like.
fsl2itk_fwd = pe.Node(c3.C3dAffineTool(fsl2ras=True, itk_transform=True),
name='fsl2itk_fwd')
fsl2itk_inv = pe.Node(c3.C3dAffineTool(fsl2ras=True, itk_transform=True),
name='fsl2itk_inv')
# Write EPI mean in T1w space
ds_t1w = pe.Node(DerivativesDataSink(base_directory=settings['output_dir'],
suffix='hmc_t1'),
name='DerivHMC_T1w')
# Write registrated file in the designated output dir
ds_tfm_fwd = pe.Node(DerivativesDataSink(
base_directory=settings['output_dir'], suffix='epi2t1w_affine'),
name='DerivEPI_to_T1w_fwd')
ds_tfm_inv = pe.Node(DerivativesDataSink(
base_directory=settings['output_dir'], suffix='t1w2epi_affine'),
name='DerivEPI_to_T1w_inv')
workflow.connect([
(inputnode, wm_mask, [('t1_seg', 'in_file')]),
(inputnode, flt_bbr_init, [('t1_brain', 'reference')]),
(inputnode, fsl2itk_fwd, [('t1_brain', 'reference_file'),
('epi_mean', 'source_file')]),
(inputnode, fsl2itk_inv, [('epi_mean', 'reference_file'),
('t1_brain', 'source_file')]),
(inputnode, flt_bbr_init, [('epi_mean', 'in_file')]),
(flt_bbr_init, flt_bbr, [('out_matrix_file', 'in_matrix_file')]),
(inputnode, flt_bbr, [('t1_brain', 'reference')]),
(inputnode, flt_bbr, [('epi_mean', 'in_file')]),
(wm_mask, flt_bbr, [('out_file', 'wm_seg')]),
(flt_bbr, invt_bbr, [('out_matrix_file', 'in_file')]),
(flt_bbr, fsl2itk_fwd, [('out_matrix_file', 'transform_file')]),
(invt_bbr, fsl2itk_inv, [('out_file', 'transform_file')]),
(fsl2itk_fwd, outputnode, [('itk_transform', 'mat_epi_to_t1')]),
(fsl2itk_inv, outputnode, [('itk_transform', 'mat_t1_to_epi')]),
(inputnode, ds_tfm_fwd, [('epi', 'source_file')]),
(inputnode, ds_tfm_inv, [('epi', 'source_file')]),
(inputnode, ds_t1w, [('epi', 'source_file')]),
(fsl2itk_fwd, ds_tfm_fwd, [('itk_transform', 'in_file')]),
(fsl2itk_inv, ds_tfm_inv, [('itk_transform', 'in_file')]),
(flt_bbr, ds_t1w, [('out_file', 'in_file')])
])
# Plots for report
png_sbref_t1 = pe.Node(niu.Function(
input_names=['in_file', 'overlay_file', 'out_file'],
output_names=['out_file'],
function=stripped_brain_overlay),
name='PNG_sbref_t1')
png_sbref_t1.inputs.out_file = 'sbref_to_t1.png'
# Write corrected file in the designated output dir
ds_png = pe.Node(DerivativesDataSink(base_directory=settings['output_dir'],
suffix='epi_to_t1'),
name='DerivativesPNG')
workflow.connect([(flt_bbr, png_sbref_t1, [('out_file', 'overlay_file')]),
(inputnode, png_sbref_t1, [('t1_seg', 'in_file')]),
(inputnode, ds_png, [('epi', 'source_file')]),
(png_sbref_t1, ds_png, [('out_file', 'in_file')])])
return workflow
def create_fsl_to_itk_conversion(mapnode, name='fsl_to_itk_conversion'):
"""
Converts an FSL-format output matrix to an ITK-format (ANTS) matrix
for use with ANTS registration tools.
Parameters
----------
name : string, optional
Name of the workflow.
Returns
-------
fsl_to_itk_conversion : nipype.pipeline.engine.Workflow
Notes
-----
Workflow Inputs::
inputspec.transform_file : string (nifti file)
Output matrix of FSL-based functional to anatomical registration
inputspec.reference_file : string (nifti file)
File of skull-stripped anatomical brain to be used in affine conversion
inputspec.source_file : string (nifti file)
Should match the input of the apply warp (in_file) unless you are
applying the warp to a 4-d file, in which case this file should
be a mean_functional file
Workflow Outputs::
outputspec.itk_transform : string (nifti file)
Converted affine transform in ITK format usable with ANTS
"""
fsl_to_itk_conversion = pe.Workflow(name=name)
inputspec = pe.Node(util.IdentityInterface(
fields=['transform_file', 'reference_file', 'source_file']),
name='inputspec')
# converts FSL-format .mat affine xfm into ANTS-format .txt
# .mat affine comes from Func->Anat registration
if mapnode == 0:
fsl_reg_2_itk = pe.Node(c3.C3dAffineTool(), name='fsl_reg_2_itk')
elif mapnode == 1:
fsl_reg_2_itk = pe.MapNode(interface=c3.C3dAffineTool(),
name='fsl_reg_2_itk',
iterfield=['source_file'])
fsl_reg_2_itk.inputs.itk_transform = True
fsl_reg_2_itk.inputs.fsl2ras = True
outputspec = pe.Node(util.IdentityInterface(fields=['itk_transform']),
name='outputspec')
fsl_to_itk_conversion.connect(inputspec, 'transform_file', fsl_reg_2_itk,
'transform_file')
fsl_to_itk_conversion.connect(inputspec, 'reference_file', fsl_reg_2_itk,
'reference_file')
fsl_to_itk_conversion.connect(inputspec, 'source_file', fsl_reg_2_itk,
'source_file')
fsl_to_itk_conversion.connect(fsl_reg_2_itk, 'itk_transform', outputspec,
'itk_transform')
return fsl_to_itk_conversion
def epi_hmc(name='EPI_HMC', settings=None):
"""
Performs :abbr:`HMC (head motion correction)` over the input
:abbr:`EPI (echo-planar imaging)` image.
"""
workflow = pe.Workflow(name=name)
inputnode = pe.Node(niu.IdentityInterface(fields=['epi']),
name='inputnode')
outputnode = pe.Node(niu.IdentityInterface(
fields=['epi_brain', 'xforms', 'epi_mask', 'epi_mean']),
name='outputnode')
bet = pe.Node(fsl.BET(functional=True, frac=0.6), name='EPI_bet')
# Head motion correction (hmc)
hmc = pe.Node(fsl.MCFLIRT(save_mats=True, save_plots=True, mean_vol=True),
name='EPI_hmc')
pick_1st = pe.Node(fsl.ExtractROI(t_min=0, t_size=1), name='EPIPickFirst')
hcm2itk = pe.MapNode(c3.C3dAffineTool(fsl2ras=True, itk_transform=True),
iterfield=['transform_file'],
name='hcm2itk')
avscale = pe.MapNode(fsl.utils.AvScale(all_param=True),
name='AvScale',
iterfield=['mat_file'])
avs_format = pe.Node(FormatHMCParam(), name='AVScale_Format')
# Calculate EPI mask on the average after HMC
bet_hmc = pe.Node(fsl.BET(mask=True, frac=0.6), name='EPI_hmc_bet')
workflow.connect([(inputnode, pick_1st, [('epi', 'in_file')]),
(inputnode, bet, [('epi', 'in_file')]),
(bet, hmc, [('out_file', 'in_file')]),
(hmc, hcm2itk, [('mat_file', 'transform_file')]),
(pick_1st, hcm2itk, [('roi_file', 'source_file'),
('roi_file', 'reference_file')]),
(hcm2itk, outputnode, [('itk_transform', 'xforms')]),
(hmc, outputnode, [('out_file', 'epi_brain')]),
(hmc, avscale, [('mat_file', 'mat_file')]),
(avscale, avs_format, [('translations', 'translations'),
('rot_angles', 'rot_angles')]),
(hmc, bet_hmc, [('mean_img', 'in_file')]),
(hmc, avscale, [('mean_img', 'ref_file')]),
(bet_hmc, outputnode, [('mask_file', 'epi_mask'),
('out_file', 'epi_mean')])])
# Write corrected file in the designated output dir
ds_hmc = pe.Node(DerivativesDataSink(base_directory=settings['output_dir'],
suffix='hmc'),
name='DerivativesHMC')
ds_mats = pe.Node(DerivativesDataSink(
base_directory=settings['output_dir'], suffix='hmc'),
name='DerivativesHMCmats')
ds_mask = pe.Node(DerivativesDataSink(
base_directory=settings['output_dir'], suffix='hmc_bmask'),
name='DerivativesEPImask')
ds_motion = pe.Node(DerivativesDataSink(
base_directory=settings['output_dir'], suffix='hmc'),
name='DerivativesParamsHMC')
workflow.connect([(inputnode, ds_hmc, [('epi', 'source_file')]),
(inputnode, ds_mats, [('epi', 'source_file')]),
(inputnode, ds_mask, [('epi', 'source_file')]),
(inputnode, ds_motion, [('epi', 'source_file')]),
(hmc, ds_hmc, [('out_file', 'in_file')]),
(hcm2itk, ds_mats, [('itk_transform', 'in_file')]),
(bet_hmc, ds_mask, [('mask_file', 'in_file')]),
(avs_format, ds_motion, [('out_file', 'in_file')])])
return workflow
def init_fsl_bbr_wf(use_bbr,
bold2t1w_dof,
bold2t1w_init,
sloppy=False,
name='fsl_bbr_wf'):
"""
Build a workflow to run FSL's ``flirt``.
This workflow uses FSL FLIRT to register a BOLD image to a T1-weighted
structural image, using a boundary-based registration (BBR) cost function.
It is a counterpart to :py:func:`~fmriprep.workflows.bold.registration.init_bbreg_wf`,
which performs the same task using FreeSurfer's ``bbregister``.
The ``use_bbr`` option permits a high degree of control over registration.
If ``False``, standard, rigid coregistration will be performed by FLIRT.
If ``True``, FLIRT-BBR will be seeded with the initial transform found by
the rigid coregistration.
If ``None``, after FLIRT-BBR is run, the resulting affine transform
will be compared to the initial transform found by FLIRT.
Excessive deviation will result in rejecting the BBR refinement and
accepting the original, affine registration.
Workflow Graph
.. workflow ::
:graph2use: orig
:simple_form: yes
from fmriprep.workflows.bold.registration import init_fsl_bbr_wf
wf = init_fsl_bbr_wf(use_bbr=True, bold2t1w_dof=9, bold2t1w_init='register')
Parameters
----------
use_bbr : :obj:`bool` or None
Enable/disable boundary-based registration refinement.
If ``None``, test BBR result for distortion before accepting.
bold2t1w_dof : 6, 9 or 12
Degrees-of-freedom for BOLD-T1w registration
bold2t1w_init : str, 'header' or 'register'
If ``'header'``, use header information for initialization of BOLD and T1 images.
If ``'register'``, align volumes by their centers.
name : :obj:`str`, optional
Workflow name (default: fsl_bbr_wf)
Inputs
------
in_file
Reference BOLD image to be registered
t1w_brain
Skull-stripped T1-weighted structural image
t1w_dseg
FAST segmentation of ``t1w_brain``
fsnative2t1w_xfm
Unused (see :py:func:`~fmriprep.workflows.bold.registration.init_bbreg_wf`)
subjects_dir
Unused (see :py:func:`~fmriprep.workflows.bold.registration.init_bbreg_wf`)
subject_id
Unused (see :py:func:`~fmriprep.workflows.bold.registration.init_bbreg_wf`)
Outputs
-------
itk_bold_to_t1
Affine transform from ``ref_bold_brain`` to T1w space (ITK format)
itk_t1_to_bold
Affine transform from T1 space to BOLD space (ITK format)
out_report
Reportlet for assessing registration quality
fallback
Boolean indicating whether BBR was rejected (rigid FLIRT registration returned)
"""
from niworkflows.engine.workflows import LiterateWorkflow as Workflow
from niworkflows.utils.images import dseg_label as _dseg_label
from niworkflows.interfaces.freesurfer import PatchedLTAConvert as LTAConvert
from niworkflows.interfaces.registration import FLIRTRPT
workflow = Workflow(name=name)
workflow.__desc__ = """\
The BOLD reference was then co-registered to the T1w reference using
`flirt` [FSL {fsl_ver}, @flirt] with the boundary-based registration [@bbr]
cost-function.
Co-registration was configured with nine degrees of freedom to account
for distortions remaining in the BOLD reference.
""".format(fsl_ver=FLIRTRPT().version or '<ver>')
inputnode = pe.Node(
niu.IdentityInterface([
'in_file',
'fsnative2t1w_xfm',
'subjects_dir',
'subject_id', # BBRegister
't1w_dseg',
't1w_brain'
]), # FLIRT BBR
name='inputnode')
outputnode = pe.Node(niu.IdentityInterface(
['itk_bold_to_t1', 'itk_t1_to_bold', 'out_report', 'fallback']),
name='outputnode')
wm_mask = pe.Node(niu.Function(function=_dseg_label), name='wm_mask')
wm_mask.inputs.label = 2 # BIDS default is WM=2
flt_bbr_init = pe.Node(FLIRTRPT(dof=6,
generate_report=not use_bbr,
uses_qform=True),
name='flt_bbr_init')
if bold2t1w_init not in ("register", "header"):
raise ValueError(
f"Unknown BOLD-T1w initialization option: {bold2t1w_init}")
if bold2t1w_init == "header":
raise NotImplementedError(
"Header-based registration initialization not supported for FSL")
invt_bbr = pe.Node(fsl.ConvertXFM(invert_xfm=True),
name='invt_bbr',
mem_gb=DEFAULT_MEMORY_MIN_GB)
# BOLD to T1 transform matrix is from fsl, using c3 tools to convert to
# something ANTs will like.
fsl2itk_fwd = pe.Node(c3.C3dAffineTool(fsl2ras=True, itk_transform=True),
name='fsl2itk_fwd',
mem_gb=DEFAULT_MEMORY_MIN_GB)
fsl2itk_inv = pe.Node(c3.C3dAffineTool(fsl2ras=True, itk_transform=True),
name='fsl2itk_inv',
mem_gb=DEFAULT_MEMORY_MIN_GB)
workflow.connect([
(inputnode, flt_bbr_init, [('in_file', 'in_file'),
('t1w_brain', 'reference')]),
(inputnode, fsl2itk_fwd, [('t1w_brain', 'reference_file'),
('in_file', 'source_file')]),
(inputnode, fsl2itk_inv, [('in_file', 'reference_file'),
('t1w_brain', 'source_file')]),
(invt_bbr, fsl2itk_inv, [('out_file', 'transform_file')]),
(fsl2itk_fwd, outputnode, [('itk_transform', 'itk_bold_to_t1')]),
(fsl2itk_inv, outputnode, [('itk_transform', 'itk_t1_to_bold')]),
])
# Short-circuit workflow building, use rigid registration
if use_bbr is False:
workflow.connect([
(flt_bbr_init, invt_bbr, [('out_matrix_file', 'in_file')]),
(flt_bbr_init, fsl2itk_fwd, [('out_matrix_file', 'transform_file')
]),
(flt_bbr_init, outputnode, [('out_report', 'out_report')]),
])
outputnode.inputs.fallback = True
return workflow
flt_bbr = pe.Node(FLIRTRPT(cost_func='bbr',
dof=bold2t1w_dof,
generate_report=True),
name='flt_bbr')
FSLDIR = os.getenv('FSLDIR')
if FSLDIR:
flt_bbr.inputs.schedule = op.join(FSLDIR, 'etc/flirtsch/bbr.sch')
else:
# Should mostly be hit while building docs
LOGGER.warning("FSLDIR unset - using packaged BBR schedule")
flt_bbr.inputs.schedule = pkgr.resource_filename(
'fmriprep', 'data/flirtsch/bbr.sch')
workflow.connect([
(inputnode, wm_mask, [('t1w_dseg', 'in_seg')]),
(inputnode, flt_bbr, [('in_file', 'in_file')]),
(flt_bbr_init, flt_bbr, [('out_matrix_file', 'in_matrix_file')]),
])
if sloppy is True:
downsample = pe.Node(niu.Function(
function=_conditional_downsampling,
output_names=["out_file", "out_mask"]),
name='downsample')
workflow.connect([
(inputnode, downsample, [("t1w_brain", "in_file")]),
(wm_mask, downsample, [("out", "in_mask")]),
(downsample, flt_bbr, [('out_file', 'reference'),
('out_mask', 'wm_seg')]),
])
else:
workflow.connect([
(inputnode, flt_bbr, [('t1w_brain', 'reference')]),
(wm_mask, flt_bbr, [('out', 'wm_seg')]),
])
# Short-circuit workflow building, use boundary-based registration
if use_bbr is True:
workflow.connect([
(flt_bbr, invt_bbr, [('out_matrix_file', 'in_file')]),
(flt_bbr, fsl2itk_fwd, [('out_matrix_file', 'transform_file')]),
(flt_bbr, outputnode, [('out_report', 'out_report')]),
])
outputnode.inputs.fallback = False
return workflow
transforms = pe.Node(niu.Merge(2),
run_without_submitting=True,
name='transforms')
reports = pe.Node(niu.Merge(2),
run_without_submitting=True,
name='reports')
compare_transforms = pe.Node(niu.Function(function=compare_xforms),
name='compare_transforms')
select_transform = pe.Node(niu.Select(),
run_without_submitting=True,
name='select_transform')
select_report = pe.Node(niu.Select(),
run_without_submitting=True,
name='select_report')
fsl_to_lta = pe.MapNode(LTAConvert(out_lta=True),
iterfield=['in_fsl'],
name='fsl_to_lta')
workflow.connect([
(flt_bbr, transforms, [('out_matrix_file', 'in1')]),
(flt_bbr_init, transforms, [('out_matrix_file', 'in2')]),
# Convert FSL transforms to LTA (RAS2RAS) transforms and compare
(inputnode, fsl_to_lta, [('in_file', 'source_file'),
('t1w_brain', 'target_file')]),
(transforms, fsl_to_lta, [('out', 'in_fsl')]),
(fsl_to_lta, compare_transforms, [('out_lta', 'lta_list')]),
(compare_transforms, outputnode, [('out', 'fallback')]),
# Select output transform
(transforms, select_transform, [('out', 'inlist')]),
(compare_transforms, select_transform, [('out', 'index')]),
(select_transform, invt_bbr, [('out', 'in_file')]),
(select_transform, fsl2itk_fwd, [('out', 'transform_file')]),
(flt_bbr, reports, [('out_report', 'in1')]),
(flt_bbr_init, reports, [('out_report', 'in2')]),
(reports, select_report, [('out', 'inlist')]),
(compare_transforms, select_report, [('out', 'index')]),
(select_report, outputnode, [('out', 'out_report')]),
])
return workflow
def init_bold_reg_wf(
bold2t1w_dof,
bold2t1w_init,
mem_gb,
omp_nthreads,
name="bold_reg_wf",
use_compression=True,
write_report=True,
):
"""
Build a workflow to run same-subject, BOLD-to-T1w image-registration.
Calculates the registration between a reference BOLD image and T1w-space.
Workflow Graph
.. workflow::
:graph2use: orig
:simple_form: yes
from fprodents.workflows.bold.registration import init_bold_reg_wf
wf = init_bold_reg_wf(mem_gb=3,
omp_nthreads=1,
bold2t1w_dof=9,
bold2t1w_init='register')
Parameters
----------
bold2t1w_dof : 6, 9 or 12
Degrees-of-freedom for BOLD-T1w registration
bold2t1w_init : str, 'header' or 'register'
If ``'header'``, use header information for initialization of BOLD and T1 images.
If ``'register'``, align volumes by their centers.
mem_gb : :obj:`float`
Size of BOLD file in GB
omp_nthreads : :obj:`int`
Maximum number of threads an individual process may use
name : :obj:`str`
Name of workflow (default: ``bold_reg_wf``)
use_compression : :obj:`bool`
Save registered BOLD series as ``.nii.gz``
use_fieldwarp : :obj:`bool`
Include SDC warp in single-shot transform from BOLD to T1
write_report : :obj:`bool`
Whether a reportlet should be stored
Inputs
------
ref_bold_brain
Reference image to which BOLD series is aligned
If ``fieldwarp == True``, ``ref_bold_brain`` should be unwarped
t1w_brain
Skull-stripped ``t1w_preproc``
Outputs
-------
bold2anat
Affine transform from ``ref_bold_brain`` to T1 space (ITK format)
anat2bold
Affine transform from T1 space to BOLD space (ITK format)
out_report
Reportlet for assessing registration quality
"""
from niworkflows.engine.workflows import LiterateWorkflow as Workflow
from niworkflows.interfaces.reportlets.registration import FLIRTRPT
workflow = Workflow(name=name)
workflow.__desc__ = """\
The BOLD reference was then co-registered to the T2w reference using
`flirt` [FSL {fsl_ver}, @flirt].
Co-registration was configured with six degrees of freedom.
""".format(fsl_ver=FLIRTRPT().version or "<ver>")
inputnode = pe.Node(
niu.IdentityInterface(fields=["ref_bold_brain", "t1w_brain"]),
name="inputnode")
outputnode = pe.Node(
niu.IdentityInterface(fields=["bold2anat", "anat2bold", "out_report"]),
name="outputnode",
)
coreg = pe.Node(FLIRTRPT(dof=bold2t1w_dof,
generate_report=True,
uses_qform=True,
args="-basescale 1"),
name="coreg")
if bold2t1w_init not in ("register", "header"):
raise ValueError(
f"Unknown BOLD-T1w initialization option: {bold2t1w_init}")
if bold2t1w_init == "header":
raise NotImplementedError(
"Header-based registration initialization not supported for FSL")
invt_xfm = pe.Node(fsl.ConvertXFM(invert_xfm=True),
name="invt_xfm",
mem_gb=DEFAULT_MEMORY_MIN_GB)
# BOLD to T1 transform matrix is from fsl, using c3 tools to convert to
# something ANTs will like.
fsl2itk_fwd = pe.Node(
c3.C3dAffineTool(fsl2ras=True, itk_transform=True),
name="fsl2itk_fwd",
mem_gb=DEFAULT_MEMORY_MIN_GB,
)
fsl2itk_inv = pe.Node(
c3.C3dAffineTool(fsl2ras=True, itk_transform=True),
name="fsl2itk_inv",
mem_gb=DEFAULT_MEMORY_MIN_GB,
)
# fmt:off
workflow.connect([
(inputnode, coreg, [('ref_bold_brain', 'in_file'),
('t1w_brain', 'reference')]),
(coreg, invt_xfm, [('out_matrix_file', 'in_file')]),
(coreg, fsl2itk_fwd, [('out_matrix_file', 'transform_file')]),
(coreg, outputnode, [('out_report', 'out_report')]),
(inputnode, fsl2itk_fwd, [('t1w_brain', 'reference_file'),
('ref_bold_brain', 'source_file')]),
(inputnode, fsl2itk_inv, [('ref_bold_brain', 'reference_file'),
('t1w_brain', 'source_file')]),
(invt_xfm, fsl2itk_inv, [('out_file', 'transform_file')]),
(fsl2itk_fwd, outputnode, [('itk_transform', 'bold2anat')]),
(fsl2itk_inv, outputnode, [('itk_transform', 'anat2bold')]),
])
# fmt:on
if write_report:
ds_report_reg = pe.Node(
DerivativesDataSink(datatype="figures",
dismiss_entities=("echo", )),
name="ds_report_reg",
run_without_submitting=True,
desc="coreg",
mem_gb=DEFAULT_MEMORY_MIN_GB,
)
# fmt:off
workflow.connect([
(outputnode, ds_report_reg, [('out_report', 'in_file')]),
])
# fmt:on
return workflow
