class T1Study(MRIStudy, metaclass=StudyMetaClass):
add_data_specs = [
DatasetSpec('fs_recon_all', freesurfer_recon_all_format,
'freesurfer_pipeline'),
DatasetSpec('brain', nifti_gz_format, 'brain_extraction_pipeline')]
add_parameter_specs = [
ParameterSpec('bet_robust', True),
ParameterSpec('bet_f_threshold', 0.57),
ParameterSpec('bet_g_threshold', -0.1)]
add_switch_specs = [
SwitchSpec('bet_method', 'optibet',
choices=MRIStudy.switch_spec('bet_method').choices)]
def freesurfer_pipeline(self, **kwargs):
"""
Segments grey matter, white matter and CSF from T1 images using
SPM "NewSegment" function.
NB: Default values come from the W2MHS toolbox
"""
pipeline = self.create_pipeline(
name='segmentation',
inputs=[DatasetSpec('primary', nifti_gz_format)],
outputs=[DatasetSpec('fs_recon_all',
freesurfer_recon_all_format)],
desc="Segment white/grey matter and csf",
version=1,
citations=copy(freesurfer_cites),
**kwargs)
# FS ReconAll node
recon_all = pipeline.create_node(
interface=ReconAll(), name='recon_all',
requirements=[freesurfer_req], wall_time=2000)
recon_all.inputs.directive = 'all'
recon_all.inputs.openmp = self.runner.num_processes
# Wrapper around os.path.join
join = pipeline.create_node(interface=JoinPath(), name='join')
pipeline.connect(recon_all, 'subjects_dir', join, 'dirname')
pipeline.connect(recon_all, 'subject_id', join, 'filename')
# Connect inputs/outputs
pipeline.connect_input('primary', recon_all, 'T1_files')
pipeline.connect_output('fs_recon_all', join, 'path')
return pipeline
def segmentation_pipeline(self, **kwargs):
pipeline = super(T1Study, self).segmentation_pipeline(img_type=1,
**kwargs)
return pipeline
class EPIStudy(MRIStudy, metaclass=StudyMetaClass):
add_data_specs = [
DatasetSpec('coreg_ref_preproc', nifti_gz_format, optional=True),
DatasetSpec('coreg_ref_wmseg', nifti_gz_format, optional=True),
DatasetSpec('reverse_phase', nifti_gz_format, optional=True),
DatasetSpec('field_map_mag', nifti_gz_format, optional=True),
DatasetSpec('field_map_phase', nifti_gz_format, optional=True),
DatasetSpec('moco', nifti_gz_format, 'intrascan_alignment_pipeline'),
DatasetSpec('align_mats', directory_format,
'intrascan_alignment_pipeline'),
DatasetSpec('moco_par', par_format, 'intrascan_alignment_pipeline'),
FieldSpec('field_map_delta_te', float, 'field_map_time_info_pipeline')
]
add_parameter_specs = [
ParameterSpec('bet_robust', True),
ParameterSpec('bet_f_threshold', 0.2),
ParameterSpec('bet_reduce_bias', False),
ParameterSpec('fugue_echo_spacing', 0.000275)
]
add_switch_specs = [
SwitchSpec('linear_reg_method',
'epireg',
choices=('flirt', 'spm', 'ants', 'epireg'))
]
def linear_coregistration_pipeline(self, **kwargs):
if self.branch('linear_reg_method', 'epireg'):
return self._epireg_linear_coregistration_pipeline(**kwargs)
else:
return super(EPIStudy,
self).linear_coregistration_pipeline(**kwargs)
def _epireg_linear_coregistration_pipeline(self, **kwargs):
pipeline = self.create_pipeline(
name='linear_coreg',
inputs=[
DatasetSpec('brain', nifti_gz_format),
DatasetSpec('coreg_ref_brain', nifti_gz_format),
DatasetSpec('coreg_ref_preproc', nifti_gz_format),
DatasetSpec('coreg_ref_wmseg', nifti_gz_format)
],
outputs=[
DatasetSpec('coreg_brain', nifti_gz_format),
DatasetSpec('coreg_matrix', text_matrix_format)
],
desc=("Intra-subjects epi registration improved using white "
"matter boundaries."),
version=1,
citations=[fsl_cite],
**kwargs)
epireg = pipeline.create_node(fsl.epi.EpiReg(),
name='epireg',
requirements=[fsl509_req])
epireg.inputs.out_base = 'epireg2ref'
pipeline.connect_input('brain', epireg, 'epi')
pipeline.connect_input('coreg_ref_brain', epireg, 't1_brain')
pipeline.connect_input('coreg_ref_preproc', epireg, 't1_head')
pipeline.connect_input('coreg_ref_wmseg', epireg, 'wmseg')
pipeline.connect_output('coreg_brain', epireg, 'out_file')
pipeline.connect_output('coreg_matrix', epireg, 'epi2str_mat')
return pipeline
def intrascan_alignment_pipeline(self, **kwargs):
pipeline = self.create_pipeline(
name='MCFLIRT_pipeline',
inputs=[DatasetSpec('preproc', nifti_gz_format)],
outputs=[
DatasetSpec('moco', nifti_gz_format),
DatasetSpec('align_mats', directory_format),
DatasetSpec('moco_par', par_format)
],
desc=("Intra-epi volumes alignment."),
version=1,
citations=[fsl_cite],
**kwargs)
mcflirt = pipeline.create_node(fsl.MCFLIRT(),
name='mcflirt',
requirements=[fsl509_req])
mcflirt.inputs.ref_vol = 0
mcflirt.inputs.save_mats = True
mcflirt.inputs.save_plots = True
mcflirt.inputs.output_type = 'NIFTI_GZ'
mcflirt.inputs.out_file = 'moco.nii.gz'
pipeline.connect_input('preproc', mcflirt, 'in_file')
pipeline.connect_output('moco', mcflirt, 'out_file')
pipeline.connect_output('moco_par', mcflirt, 'par_file')
merge = pipeline.create_node(MergeListMotionMat(), name='merge')
pipeline.connect(mcflirt, 'mat_file', merge, 'file_list')
pipeline.connect_output('align_mats', merge, 'out_dir')
return pipeline
def field_map_time_info_pipeline(self, **kwargs):
pipeline = self.create_pipeline(
name='field_map_time_info_pipeline',
inputs=[DatasetSpec('field_map_mag', dicom_format)],
outputs=[FieldSpec('field_map_delta_te', float)],
desc=("Pipeline to extract delta TE from field map "
"images, if provided"),
version=1,
citations=[fsl_cite],
**kwargs)
delta_te = pipeline.create_node(FieldMapTimeInfo(),
name='extract_delta_te')
pipeline.connect_input('field_map_mag', delta_te, 'fm_mag')
pipeline.connect_output('field_map_delta_te', delta_te, 'delta_te')
return pipeline
def preproc_pipeline(self, **kwargs):
if ('field_map_phase' in self.input_names
and 'field_map_mag' in self.input_names):
return self._fugue_pipeline(**kwargs)
elif 'reverse_phase' in self.input_names:
return self._topup_pipeline(**kwargs)
else:
return super(EPIStudy, self).preproc_pipeline(**kwargs)
def _topup_pipeline(self, **kwargs):
pipeline = self.create_pipeline(
name='preproc_pipeline',
inputs=[
DatasetSpec('primary', nifti_gz_format),
DatasetSpec('reverse_phase', nifti_gz_format),
FieldSpec('ped', str),
FieldSpec('pe_angle', str)
],
outputs=[DatasetSpec('preproc', nifti_gz_format)],
desc=("Topup distortion correction pipeline"),
version=1,
citations=[fsl_cite],
**kwargs)
reorient_epi_in = pipeline.create_node(fsl.utils.Reorient2Std(),
name='reorient_epi_in',
requirements=[fsl509_req])
pipeline.connect_input('primary', reorient_epi_in, 'in_file')
reorient_epi_opposite = pipeline.create_node(
fsl.utils.Reorient2Std(),
name='reorient_epi_opposite',
requirements=[fsl509_req])
pipeline.connect_input('reverse_phase', reorient_epi_opposite,
'in_file')
prep_dwi = pipeline.create_node(PrepareDWI(), name='prepare_dwi')
prep_dwi.inputs.topup = True
pipeline.connect_input('ped', prep_dwi, 'pe_dir')
pipeline.connect_input('pe_angle', prep_dwi, 'ped_polarity')
pipeline.connect(reorient_epi_in, 'out_file', prep_dwi, 'dwi')
pipeline.connect(reorient_epi_opposite, 'out_file', prep_dwi, 'dwi1')
ped = pipeline.create_node(GenTopupConfigFiles(), name='gen_config')
pipeline.connect(prep_dwi, 'pe', ped, 'ped')
merge_outputs = pipeline.create_node(merge_lists(2),
name='merge_files')
pipeline.connect(prep_dwi, 'main', merge_outputs, 'in1')
pipeline.connect(prep_dwi, 'secondary', merge_outputs, 'in2')
merge = pipeline.create_node(fsl_merge(),
name='fsl_merge',
requirements=[fsl509_req])
merge.inputs.dimension = 't'
pipeline.connect(merge_outputs, 'out', merge, 'in_files')
topup = pipeline.create_node(TOPUP(),
name='topup',
requirements=[fsl509_req])
pipeline.connect(merge, 'merged_file', topup, 'in_file')
pipeline.connect(ped, 'config_file', topup, 'encoding_file')
in_apply_tp = pipeline.create_node(merge_lists(1), name='in_apply_tp')
pipeline.connect(reorient_epi_in, 'out_file', in_apply_tp, 'in1')
apply_topup = pipeline.create_node(ApplyTOPUP(),
name='applytopup',
requirements=[fsl509_req])
apply_topup.inputs.method = 'jac'
apply_topup.inputs.in_index = [1]
pipeline.connect(in_apply_tp, 'out', apply_topup, 'in_files')
pipeline.connect(ped, 'apply_topup_config', apply_topup,
'encoding_file')
pipeline.connect(topup, 'out_movpar', apply_topup, 'in_topup_movpar')
pipeline.connect(topup, 'out_fieldcoef', apply_topup,
'in_topup_fieldcoef')
pipeline.connect_output('preproc', apply_topup, 'out_corrected')
return pipeline
def _fugue_pipeline(self, **kwargs):
pipeline = self.create_pipeline(
name='preproc_pipeline',
inputs=[
DatasetSpec('primary', nifti_gz_format),
DatasetSpec('field_map_mag', nifti_gz_format),
DatasetSpec('field_map_phase', nifti_gz_format),
FieldSpec('field_map_delta_te', float)
],
outputs=[DatasetSpec('preproc', nifti_gz_format)],
desc=("Fugue distortion correction pipeline"),
version=1,
citations=[fsl_cite],
**kwargs)
reorient_epi_in = pipeline.create_node(fsl.utils.Reorient2Std(),
name='reorient_epi_in',
requirements=[fsl509_req])
pipeline.connect_input('primary', reorient_epi_in, 'in_file')
fm_mag_reorient = pipeline.create_node(fsl.utils.Reorient2Std(),
name='reorient_fm_mag',
requirements=[fsl509_req])
pipeline.connect_input('field_map_mag', fm_mag_reorient, 'in_file')
fm_phase_reorient = pipeline.create_node(fsl.utils.Reorient2Std(),
name='reorient_fm_phase',
requirements=[fsl509_req])
pipeline.connect_input('field_map_phase', fm_phase_reorient, 'in_file')
bet = pipeline.create_node(BET(),
name="bet",
wall_time=5,
requirements=[fsl509_req])
bet.inputs.robust = True
pipeline.connect(fm_mag_reorient, 'out_file', bet, 'in_file')
create_fmap = pipeline.create_node(PrepareFieldmap(),
name="prepfmap",
wall_time=5,
requirements=[fsl509_req])
# create_fmap.inputs.delta_TE = 2.46
pipeline.connect_input('field_map_delta_te', create_fmap, 'delta_TE')
pipeline.connect(bet, "out_file", create_fmap, "in_magnitude")
pipeline.connect(fm_phase_reorient, 'out_file', create_fmap,
'in_phase')
fugue = pipeline.create_node(FUGUE(),
name='fugue',
wall_time=5,
requirements=[fsl509_req])
fugue.inputs.unwarp_direction = 'x'
fugue.inputs.dwell_time = self.parameter('fugue_echo_spacing')
fugue.inputs.unwarped_file = 'example_func.nii.gz'
pipeline.connect(create_fmap, 'out_fieldmap', fugue, 'fmap_in_file')
pipeline.connect(reorient_epi_in, 'out_file', fugue, 'in_file')
pipeline.connect_output('preproc', fugue, 'unwarped_file')
return pipeline
def motion_mat_pipeline(self, **kwargs):
inputs = [
DatasetSpec('coreg_matrix', text_matrix_format),
DatasetSpec('qform_mat', text_matrix_format)
]
if 'reverse_phase' not in self.input_names:
inputs.append(DatasetSpec('align_mats', directory_format))
pipeline = self.create_pipeline(
name='motion_mat_calculation',
inputs=inputs,
outputs=[DatasetSpec('motion_mats', motion_mats_format)],
desc=("Motion matrices calculation"),
version=1,
citations=[fsl_cite],
**kwargs)
mm = pipeline.create_node(MotionMatCalculation(), name='motion_mats')
pipeline.connect_input('coreg_matrix', mm, 'reg_mat')
pipeline.connect_input('qform_mat', mm, 'qform_mat')
if 'reverse_phase' not in self.input_names:
pipeline.connect_input('align_mats', mm, 'align_mats')
pipeline.connect_output('motion_mats', mm, 'motion_mats')
return pipeline
class FunctionalMRIStudy(EPIStudy, metaclass=StudyMetaClass):
add_parameter_specs = [
ParameterSpec('component_threshold', 20),
ParameterSpec('motion_reg', True),
ParameterSpec('highpass', 0.01),
ParameterSpec('brain_thresh_percent', 5),
ParameterSpec('MNI_template',
os.path.join(atlas_path, 'MNI152_T1_2mm.nii.gz')),
ParameterSpec(
'MNI_template_mask',
os.path.join(atlas_path, 'MNI152_T1_2mm_brain_mask.nii.gz')),
ParameterSpec('group_ica_components', 15)
]
add_data_specs = [
DatasetSpec('train_data',
rfile_format,
optional=True,
frequency='per_project'),
DatasetSpec('hand_label_noise', text_format,
'fix_preparation_pipeline'),
DatasetSpec('labelled_components', text_format,
'fix_classification_pipeline'),
DatasetSpec('cleaned_file', nifti_gz_format,
'fix_regression_pipeline'),
DatasetSpec('filtered_data', nifti_gz_format,
'rsfMRI_filtering_pipeline'),
DatasetSpec('mc_par', par_format, 'rsfMRI_filtering_pipeline'),
DatasetSpec('melodic_ica', zip_format,
'single_subject_melodic_pipeline'),
DatasetSpec('fix_dir', zip_format, 'fix_preparation_pipeline'),
DatasetSpec('normalized_ts', nifti_gz_format,
'timeseries_normalization_to_atlas_pipeline'),
DatasetSpec('smoothed_ts', nifti_gz_format, 'smoothing_pipeline')
]
add_switch_specs = [
SwitchSpec('linear_reg_method',
'ants',
choices=('flirt', 'spm', 'ants', 'epireg'))
]
def rsfMRI_filtering_pipeline(self, **kwargs):
pipeline = self.create_pipeline(
name='rsfMRI_filtering',
inputs=[
DatasetSpec('preproc', nifti_gz_format),
DatasetSpec('brain_mask', nifti_gz_format),
DatasetSpec('coreg_ref_brain', nifti_gz_format),
FieldSpec('tr', float)
],
outputs=[
DatasetSpec('filtered_data', nifti_gz_format),
DatasetSpec('mc_par', par_format)
],
desc=("Spatial and temporal rsfMRI filtering"),
version=1,
citations=[fsl_cite],
**kwargs)
afni_mc = pipeline.create_node(Volreg(),
name='AFNI_MC',
wall_time=5,
requirements=[afni_req])
afni_mc.inputs.zpad = 1
afni_mc.inputs.out_file = 'rsfmri_mc.nii.gz'
afni_mc.inputs.oned_file = 'prefiltered_func_data_mcf.par'
pipeline.connect_input('preproc', afni_mc, 'in_file')
filt = pipeline.create_node(Tproject(),
name='Tproject',
wall_time=5,
requirements=[afni_req])
filt.inputs.stopband = (0, 0.01)
filt.inputs.polort = 3
filt.inputs.blur = 3
filt.inputs.out_file = 'filtered_func_data.nii.gz'
pipeline.connect_input('tr', filt, 'delta_t')
pipeline.connect(afni_mc, 'out_file', filt, 'in_file')
pipeline.connect_input('brain_mask', filt, 'mask')
meanfunc = pipeline.create_node(ImageMaths(op_string='-Tmean',
suffix='_mean'),
name='meanfunc',
wall_time=5,
requirements=[fsl5_req])
pipeline.connect(afni_mc, 'out_file', meanfunc, 'in_file')
add_mean = pipeline.create_node(ImageMaths(op_string='-add'),
name='add_mean',
wall_time=5,
requirements=[fsl5_req])
pipeline.connect(filt, 'out_file', add_mean, 'in_file')
pipeline.connect(meanfunc, 'out_file', add_mean, 'in_file2')
pipeline.connect_output('filtered_data', add_mean, 'out_file')
pipeline.connect_output('mc_par', afni_mc, 'oned_file')
return pipeline
def single_subject_melodic_pipeline(self, **kwargs):
pipeline = self.create_pipeline(
name='MelodicL1',
inputs=[
DatasetSpec('filtered_data', nifti_gz_format),
FieldSpec('tr', float),
DatasetSpec('brain_mask', nifti_gz_format)
],
outputs=[DatasetSpec('melodic_ica', directory_format)],
desc=("Single subject ICA analysis using FSL MELODIC."),
version=1,
citations=[fsl_cite],
**kwargs)
mel = pipeline.create_node(MELODIC(),
name='melodic_L1',
wall_time=15,
requirements=[fsl5_req])
mel.inputs.no_bet = True
pipeline.connect_input('brain_mask', mel, 'mask')
mel.inputs.bg_threshold = self.parameter('brain_thresh_percent')
mel.inputs.report = True
mel.inputs.out_stats = True
mel.inputs.mm_thresh = 0.5
mel.inputs.out_dir = 'melodic_ica'
pipeline.connect_input('tr', mel, 'tr_sec')
pipeline.connect_input('filtered_data', mel, 'in_files')
pipeline.connect_output('melodic_ica', mel, 'out_dir')
return pipeline
def fix_preparation_pipeline(self, **kwargs):
pipeline = self.create_pipeline(
name='prepare_fix',
inputs=[
DatasetSpec('melodic_ica', directory_format),
DatasetSpec('filtered_data', nifti_gz_format),
DatasetSpec('coreg_to_atlas_mat', text_matrix_format),
DatasetSpec('coreg_matrix', text_matrix_format),
DatasetSpec('preproc', nifti_gz_format),
DatasetSpec('brain', nifti_gz_format),
DatasetSpec('coreg_ref_brain', nifti_gz_format),
DatasetSpec('mc_par', par_format),
DatasetSpec('brain_mask', nifti_gz_format)
],
outputs=[
DatasetSpec('fix_dir', directory_format),
DatasetSpec('hand_label_noise', text_format)
],
desc=("Pipeline to create the right folder structure before "
"running FIX"),
version=1,
citations=[fsl_cite],
**kwargs)
struct_ants2fsl = pipeline.create_node(ANTs2FSLMatrixConversion(),
name='struct_ants2fsl',
requirements=[c3d_req])
struct_ants2fsl.inputs.ras2fsl = True
struct_ants2fsl.inputs.reference_file = self.parameter('MNI_template')
pipeline.connect_input('coreg_to_atlas_mat', struct_ants2fsl,
'itk_file')
pipeline.connect_input('coreg_ref_brain', struct_ants2fsl,
'source_file')
epi_ants2fsl = pipeline.create_node(ANTs2FSLMatrixConversion(),
name='epi_ants2fsl',
requirements=[c3d_req])
epi_ants2fsl.inputs.ras2fsl = True
pipeline.connect_input('brain', epi_ants2fsl, 'source_file')
pipeline.connect_input('coreg_matrix', epi_ants2fsl, 'itk_file')
pipeline.connect_input('coreg_ref_brain', epi_ants2fsl,
'reference_file')
MNI2t1 = pipeline.create_node(ConvertXFM(),
name='MNI2t1',
wall_time=5,
requirements=[fsl509_req])
MNI2t1.inputs.invert_xfm = True
pipeline.connect(struct_ants2fsl, 'fsl_matrix', MNI2t1, 'in_file')
struct2epi = pipeline.create_node(ConvertXFM(),
name='struct2epi',
wall_time=5,
requirements=[fsl509_req])
struct2epi.inputs.invert_xfm = True
pipeline.connect(epi_ants2fsl, 'fsl_matrix', struct2epi, 'in_file')
meanfunc = pipeline.create_node(ImageMaths(op_string='-Tmean',
suffix='_mean'),
name='meanfunc',
wall_time=5,
requirements=[fsl509_req])
pipeline.connect_input('preproc', meanfunc, 'in_file')
prep_fix = pipeline.create_node(PrepareFIX(), name='prep_fix')
pipeline.connect_input('melodic_ica', prep_fix, 'melodic_dir')
pipeline.connect_input('coreg_ref_brain', prep_fix, 't1_brain')
pipeline.connect_input('mc_par', prep_fix, 'mc_par')
pipeline.connect_input('brain_mask', prep_fix, 'epi_brain_mask')
pipeline.connect_input('preproc', prep_fix, 'epi_preproc')
pipeline.connect_input('filtered_data', prep_fix, 'filtered_epi')
pipeline.connect(epi_ants2fsl, 'fsl_matrix', prep_fix, 'epi2t1_mat')
pipeline.connect(struct_ants2fsl, 'fsl_matrix', prep_fix, 't12MNI_mat')
pipeline.connect(MNI2t1, 'out_file', prep_fix, 'MNI2t1_mat')
pipeline.connect(struct2epi, 'out_file', prep_fix, 't12epi_mat')
pipeline.connect(meanfunc, 'out_file', prep_fix, 'epi_mean')
pipeline.connect_output('fix_dir', prep_fix, 'fix_dir')
pipeline.connect_output('hand_label_noise', prep_fix,
'hand_label_file')
return pipeline
def fix_classification_pipeline(self, **kwargs):
pipeline = self.create_pipeline(
name='fix_classification',
inputs=[
DatasetSpec('train_data',
rfile_format,
frequency='per_project'),
DatasetSpec('fix_dir', directory_format)
],
outputs=[DatasetSpec('labelled_components', text_format)],
desc=("Automatic classification of noisy components from the "
"rsfMRI data using fsl FIX."),
version=1,
citations=[fsl_cite],
**kwargs)
fix = pipeline.create_node(FSLFIX(),
name="fix",
wall_time=30,
requirements=[fsl509_req, fix_req])
pipeline.connect_input("fix_dir", fix, "feat_dir")
pipeline.connect_input("train_data", fix, "train_data")
fix.inputs.component_threshold = self.parameter('component_threshold')
fix.inputs.motion_reg = self.parameter('motion_reg')
fix.inputs.classification = True
pipeline.connect_output('labelled_components', fix, 'label_file')
return pipeline
def fix_regression_pipeline(self, **kwargs):
pipeline = self.create_pipeline(
name='signal_regression',
inputs=[
DatasetSpec('fix_dir', directory_format),
DatasetSpec('labelled_components', text_format)
],
outputs=[DatasetSpec('cleaned_file', nifti_gz_format)],
desc=("Regression of the noisy components from the rsfMRI data "
"using a python implementation equivalent to that in FIX."),
version=1,
citations=[fsl_cite],
**kwargs)
signal_reg = pipeline.create_node(SignalRegression(),
name="signal_reg",
wall_time=30,
requirements=[fsl509_req, fix_req])
pipeline.connect_input("fix_dir", signal_reg, "fix_dir")
pipeline.connect_input("labelled_components", signal_reg,
"labelled_components")
signal_reg.inputs.motion_regression = self.parameter('motion_reg')
signal_reg.inputs.highpass = self.parameter('highpass')
pipeline.connect_output('cleaned_file', signal_reg, 'output')
return pipeline
def timeseries_normalization_to_atlas_pipeline(self, **kwargs):
pipeline = self.create_pipeline(
name='timeseries_normalization_to_atlas_pipeline',
inputs=[
DatasetSpec('cleaned_file', nifti_gz_format),
DatasetSpec('coreg_to_atlas_warp', nifti_gz_format),
DatasetSpec('coreg_to_atlas_mat', text_matrix_format),
DatasetSpec('coreg_matrix', text_matrix_format)
],
outputs=[DatasetSpec('normalized_ts', nifti_gz_format)],
desc=("Apply ANTs transformation to the fmri filtered file to "
"normalize it to MNI 2mm."),
version=1,
citations=[fsl_cite],
**kwargs)
merge_trans = pipeline.create_node(NiPypeMerge(3),
name='merge_transforms',
wall_time=1)
pipeline.connect_input('coreg_to_atlas_warp', merge_trans, 'in1')
pipeline.connect_input('coreg_to_atlas_mat', merge_trans, 'in2')
pipeline.connect_input('coreg_matrix', merge_trans, 'in3')
apply_trans = pipeline.create_node(ApplyTransforms(),
name='ApplyTransform',
wall_time=7,
memory=24000,
requirements=[ants2_req])
ref_brain = self.parameter('MNI_template')
apply_trans.inputs.reference_image = ref_brain
apply_trans.inputs.interpolation = 'Linear'
apply_trans.inputs.input_image_type = 3
pipeline.connect(merge_trans, 'out', apply_trans, 'transforms')
pipeline.connect_input('cleaned_file', apply_trans, 'input_image')
pipeline.connect_output('normalized_ts', apply_trans, 'output_image')
return pipeline
def smoothing_pipeline(self, **kwargs):
pipeline = self.create_pipeline(
name='smoothing_pipeline',
inputs=[DatasetSpec('normalized_ts', nifti_gz_format)],
outputs=[DatasetSpec('smoothed_ts', nifti_gz_format)],
desc=("Spatial smoothing of the normalized fmri file"),
version=1,
citations=[fsl_cite],
**kwargs)
smooth = pipeline.create_node(BlurToFWHM(),
name='3dBlurToFWHM',
wall_time=5,
requirements=[afni_req])
smooth.inputs.fwhm = 5
smooth.inputs.out_file = 'smoothed_ts.nii.gz'
smooth.inputs.mask = self.parameter('MNI_template_mask')
pipeline.connect_input('normalized_ts', smooth, 'in_file')
pipeline.connect_output('smoothed_ts', smooth, 'out_file')
return pipeline
class NODDIStudy(DiffusionStudy, metaclass=StudyMetaClass):
add_data_specs = [
DatasetSpec('low_b_dw_scan', mrtrix_format),
DatasetSpec('high_b_dw_scan', mrtrix_format),
DatasetSpec('forward_pe', mrtrix_format),
DatasetSpec('reverse_pe', mrtrix_format),
DatasetSpec('dwi_scan', mrtrix_format, 'concatenate_pipeline'),
DatasetSpec('ficvf', nifti_format, 'noddi_fitting_pipeline'),
DatasetSpec('odi', nifti_format, 'noddi_fitting_pipeline'),
DatasetSpec('fiso', nifti_format, 'noddi_fitting_pipeline'),
DatasetSpec('fibredirs_xvec', nifti_format, 'noddi_fitting_pipeline'),
DatasetSpec('fibredirs_yvec', nifti_format, 'noddi_fitting_pipeline'),
DatasetSpec('fibredirs_zvec', nifti_format, 'noddi_fitting_pipeline'),
DatasetSpec('fmin', nifti_format, 'noddi_fitting_pipeline'),
DatasetSpec('kappa', nifti_format, 'noddi_fitting_pipeline'),
DatasetSpec('error_code', nifti_format, 'noddi_fitting_pipeline')
]
add_parameter_specs = [
ParameterSpec('noddi_model', 'WatsonSHStickTortIsoV_B0')
]
add_switch_specs = [SwitchSpec('single_slice', False)]
def concatenate_pipeline(self, **kwargs): # @UnusedVariable
"""
Concatenates two dMRI datasets (with different b-values) along the
DW encoding (4th) axis
"""
pipeline = self.create_pipeline(
name='concatenation',
inputs=[
DatasetSpec('low_b_dw_scan', mrtrix_format),
DatasetSpec('high_b_dw_scan', mrtrix_format)
],
outputs=[DatasetSpec('dwi_scan', mrtrix_format)],
desc=("Concatenate low and high b-value dMRI datasets for NODDI "
"processing"),
version=1,
citations=[mrtrix_cite],
**kwargs)
# Create concatenation node
mrcat = pipeline.create_node(MRCat(),
name='mrcat',
requirements=[mrtrix3_req])
mrcat.inputs.quiet = True
# Connect inputs
pipeline.connect_input('low_b_dw_scan', mrcat, 'first_scan')
pipeline.connect_input('high_b_dw_scan', mrcat, 'second_scan')
# Connect outputs
pipeline.connect_output('dwi_scan', mrcat, 'out_file')
# Check inputs/outputs are connected
return pipeline
def noddi_fitting_pipeline(self, **kwargs): # @UnusedVariable
"""
Creates a ROI in which the NODDI processing will be performed
Parameters
----------
single_slice: Int
If provided the processing is only performed on a single slice
(for testing)
noddi_model: Str
Name of the NODDI model to use for the fitting
nthreads: Int
Number of processes to use
"""
pipeline_name = 'noddi_fitting'
inputs = [
DatasetSpec('bias_correct', nifti_gz_format),
DatasetSpec('grad_dirs', fsl_bvecs_format),
DatasetSpec('bvalues', fsl_bvals_format)
]
if self.switch('single_slice'):
inputs.append(DatasetSpec('eroded_mask', nifti_gz_format))
else:
inputs.append(DatasetSpec('brain_mask', nifti_gz_format))
pipeline = self.create_pipeline(
name=pipeline_name,
inputs=inputs,
outputs=[
DatasetSpec('ficvf', nifti_format),
DatasetSpec('odi', nifti_format),
DatasetSpec('fiso', nifti_format),
DatasetSpec('fibredirs_xvec', nifti_format),
DatasetSpec('fibredirs_yvec', nifti_format),
DatasetSpec('fibredirs_zvec', nifti_format),
DatasetSpec('fmin', nifti_format),
DatasetSpec('kappa', nifti_format),
DatasetSpec('error_code', nifti_format)
],
desc=("Creates a ROI in which the NODDI processing will be "
"performed"),
citations=[noddi_cite],
**kwargs)
# Create node to unzip the nifti files
unzip_bias_correct = pipeline.create_node(MRConvert(),
name="unzip_bias_correct",
requirements=[mrtrix3_req])
unzip_bias_correct.inputs.out_ext = 'nii'
unzip_bias_correct.inputs.quiet = True
unzip_mask = pipeline.create_node(MRConvert(),
name="unzip_mask",
requirements=[mrtrix3_req])
unzip_mask.inputs.out_ext = 'nii'
unzip_mask.inputs.quiet = True
# Create create-roi node
create_roi = pipeline.create_node(
CreateROI(),
name='create_roi',
requirements=[noddi_req, matlab2015_req],
memory=4000)
pipeline.connect(unzip_bias_correct, 'out_file', create_roi, 'in_file')
pipeline.connect(unzip_mask, 'out_file', create_roi, 'brain_mask')
# Create batch-fitting node
batch_fit = pipeline.create_node(
BatchNODDIFitting(),
name="batch_fit",
requirements=[noddi_req, matlab2015_req],
wall_time=180,
memory=8000)
batch_fit.inputs.model = self.parameter('noddi_model')
batch_fit.inputs.nthreads = self.runner.num_processes
pipeline.connect(create_roi, 'out_file', batch_fit, 'roi_file')
# Create output node
save_params = pipeline.create_node(
SaveParamsAsNIfTI(),
name="save_params",
requirements=[noddi_req, matlab2015_req],
memory=4000)
save_params.inputs.output_prefix = 'params'
pipeline.connect(batch_fit, 'out_file', save_params, 'params_file')
pipeline.connect(create_roi, 'out_file', save_params, 'roi_file')
pipeline.connect(unzip_mask, 'out_file', save_params,
'brain_mask_file')
# Connect inputs
pipeline.connect_input('bias_correct', unzip_bias_correct, 'in_file')
if not pipeline.switch('single_slice'):
pipeline.connect_input('eroded_mask', unzip_mask, 'in_file')
else:
pipeline.connect_input('brain_mask', unzip_mask, 'in_file')
pipeline.connect_input('grad_dirs', batch_fit, 'bvecs_file')
pipeline.connect_input('bvalues', batch_fit, 'bvals_file')
# Connect outputs
pipeline.connect_output('ficvf', save_params, 'ficvf')
pipeline.connect_output('odi', save_params, 'odi')
pipeline.connect_output('fiso', save_params, 'fiso')
pipeline.connect_output('fibredirs_xvec', save_params,
'fibredirs_xvec')
pipeline.connect_output('fibredirs_yvec', save_params,
'fibredirs_yvec')
pipeline.connect_output('fibredirs_zvec', save_params,
'fibredirs_zvec')
pipeline.connect_output('fmin', save_params, 'fmin')
pipeline.connect_output('kappa', save_params, 'kappa')
pipeline.connect_output('error_code', save_params, 'error_code')
# Check inputs/outputs are connected
return pipeline
class DiffusionStudy(EPIStudy, metaclass=StudyMetaClass):
add_data_specs = [
DatasetSpec('dwi_reference', nifti_gz_format, optional=True),
DatasetSpec('forward_pe', dicom_format, optional=True),
DatasetSpec('b0',
nifti_gz_format,
'extract_b0_pipeline',
desc="b0 image"),
DatasetSpec('noise_residual', mrtrix_format, 'preproc_pipeline'),
DatasetSpec('tensor', nifti_gz_format, 'tensor_pipeline'),
DatasetSpec('fa', nifti_gz_format, 'tensor_pipeline'),
DatasetSpec('adc', nifti_gz_format, 'tensor_pipeline'),
DatasetSpec('wm_response', text_format, 'response_pipeline'),
DatasetSpec('gm_response', text_format, 'response_pipeline'),
DatasetSpec('csf_response', text_format, 'response_pipeline'),
DatasetSpec('avg_response', text_format, 'average_response_pipeline'),
DatasetSpec('wm_odf', mrtrix_format, 'fod_pipeline'),
DatasetSpec('gm_odf', mrtrix_format, 'fod_pipeline'),
DatasetSpec('csf_odf', mrtrix_format, 'fod_pipeline'),
DatasetSpec('bias_correct', nifti_gz_format, 'bias_correct_pipeline'),
DatasetSpec('grad_dirs', fsl_bvecs_format, 'preproc_pipeline'),
DatasetSpec('bvalues', fsl_bvals_format, 'preproc_pipeline'),
DatasetSpec('eddy_par', eddy_par_format, 'preproc_pipeline'),
DatasetSpec('align_mats', directory_format,
'intrascan_alignment_pipeline'),
DatasetSpec('tbss_mean_fa',
nifti_gz_format,
'tbss_pipeline',
frequency='per_project'),
DatasetSpec('tbss_proj_fa',
nifti_gz_format,
'tbss_pipeline',
frequency='per_project'),
DatasetSpec('tbss_skeleton',
nifti_gz_format,
'tbss_pipeline',
frequency='per_project'),
DatasetSpec('tbss_skeleton_mask',
nifti_gz_format,
'tbss_pipeline',
frequency='per_project'),
DatasetSpec('brain', nifti_gz_format, 'brain_extraction_pipeline'),
DatasetSpec('brain_mask', nifti_gz_format,
'brain_extraction_pipeline'),
DatasetSpec('norm_intensity', mrtrix_format,
'intensity_normalisation_pipeline'),
DatasetSpec('norm_intens_fa_template',
mrtrix_format,
'intensity_normalisation_pipeline',
frequency='per_project'),
DatasetSpec('norm_intens_wm_mask',
mrtrix_format,
'intensity_normalisation_pipeline',
frequency='per_project')
]
add_parameter_specs = [
ParameterSpec('multi_tissue', True),
ParameterSpec('preproc_pe_dir', None, dtype=str),
ParameterSpec('tbss_skel_thresh', 0.2),
ParameterSpec('fsl_mask_f', 0.25),
ParameterSpec('bet_robust', True),
ParameterSpec('bet_f_threshold', 0.2),
ParameterSpec('bet_reduce_bias', False)
]
add_switch_specs = [
SwitchSpec('preproc_denoise', False),
SwitchSpec('response_algorithm', 'tax',
('tax', 'dhollander', 'msmt_5tt')),
SwitchSpec('fod_algorithm', 'csd', ('csd', 'msmt_csd')),
SwitchSpec('brain_extract_method', 'mrtrix', ('mrtrix', 'fsl')),
SwitchSpec('bias_correct_method', 'ants', choices=('ants', 'fsl'))
]
@property
def multi_tissue(self):
return self.branch('response_algorithm', ('msmt_5tt', 'dhollander'))
def preproc_pipeline(self, **kwargs): # @UnusedVariable @IgnorePep8
"""
Performs a series of FSL preprocessing steps, including Eddy and Topup
Parameters
----------
phase_dir : str{AP|LR|IS}
The phase encode direction
"""
outputs = [
DatasetSpec('preproc', nifti_gz_format),
DatasetSpec('grad_dirs', fsl_bvecs_format),
DatasetSpec('bvalues', fsl_bvals_format),
DatasetSpec('eddy_par', eddy_par_format)
]
citations = [fsl_cite, eddy_cite, topup_cite, distort_correct_cite]
if self.switch('preproc_denoise'):
outputs.append(DatasetSpec('noise_residual', mrtrix_format))
citations.extend(dwidenoise_cites)
if 'dwi_reference' in self.input_names or 'reverse_phase' in self.input_names:
inputs = [
DatasetSpec('primary', dicom_format),
FieldSpec('ped', dtype=str),
FieldSpec('pe_angle', dtype=str)
]
if 'dwi_reference' in self.input_names:
inputs.append(DatasetSpec('dwi_reference', nifti_gz_format))
if 'reverse_phase' in self.input_names:
inputs.append(DatasetSpec('reverse_phase', nifti_gz_format))
distortion_correction = True
else:
inputs = [DatasetSpec('primary', dicom_format)]
distortion_correction = False
pipeline = self.create_pipeline(
name='preprocess',
inputs=inputs,
outputs=outputs,
desc=("Preprocess dMRI studies using distortion correction"),
version=1,
citations=citations,
**kwargs)
# Denoise the dwi-scan
if self.switch('preproc_denoise'):
# Run denoising
denoise = pipeline.create_node(DWIDenoise(),
name='denoise',
requirements=[mrtrix3_req])
denoise.inputs.out_file_ext = '.mif'
# Calculate residual noise
subtract_operands = pipeline.create_node(Merge(2),
name='subtract_operands')
subtract = pipeline.create_node(MRCalc(),
name='subtract',
requirements=[mrtrix3_req])
subtract.inputs.out_ext = '.mif'
subtract.inputs.operation = 'subtract'
dwipreproc = pipeline.create_node(
DWIPreproc(),
name='dwipreproc',
requirements=[mrtrix3_req, fsl509_req],
wall_time=60)
dwipreproc.inputs.eddy_parameters = '--data_is_shelled '
dwipreproc.inputs.no_clean_up = True
dwipreproc.inputs.out_file_ext = '.nii.gz'
dwipreproc.inputs.temp_dir = 'dwipreproc_tempdir'
# Create node to reorient preproc out_file
swap = pipeline.create_node(fsl.utils.Reorient2Std(),
name='fslreorient2std',
requirements=[fsl509_req])
if distortion_correction:
# Extract b=0 volumes
dwiextract = pipeline.create_node(ExtractDWIorB0(),
name='dwiextract',
requirements=[mrtrix3_req])
dwiextract.inputs.bzero = True
dwiextract.inputs.out_ext = '.nii.gz'
# Get first b=0 from dwi b=0 volumes
mrconvert = pipeline.create_node(MRConvert(),
name="mrconvert",
requirements=[mrtrix3_req])
mrconvert.inputs.coord = (3, 0)
# Concatenate extracted forward rpe with reverse rpe
mrcat = pipeline.create_node(MRCat(),
name='mrcat',
requirements=[mrtrix3_req])
# Create node to assign the right PED to the diffusion
prep_dwi = pipeline.create_node(PrepareDWI(), name='prepare_dwi')
# Create preprocessing node
dwipreproc.inputs.rpe_pair = True
if self.parameter('preproc_pe_dir') is not None:
dwipreproc.inputs.pe_dir = self.parameter('preproc_pe_dir')
# Create nodes to gradients to FSL format
extract_grad = pipeline.create_node(ExtractFSLGradients(),
name="extract_grad",
requirements=[mrtrix3_req])
# Connect inputs
if 'dwi_reference' in self.input_names:
pipeline.connect_input('dwi_reference', mrcat, 'second_scan')
elif 'reverse_phase' in self.input_names:
pipeline.connect_input('reverse_phase', mrcat, 'second_scan')
else:
assert False
pipeline.connect_input('primary', dwiextract, 'in_file')
# Connect inter-nodes
if self.switch('preproc_denoise'):
pipeline.connect_input('primary', denoise, 'in_file')
pipeline.connect_input('primary', subtract_operands, 'in1')
pipeline.connect(denoise, 'out_file', dwipreproc, 'in_file')
pipeline.connect(denoise, 'noise', subtract_operands, 'in2')
pipeline.connect(subtract_operands, 'out', subtract, 'operands')
else:
pipeline.connect_input('primary', dwipreproc, 'in_file')
if distortion_correction:
pipeline.connect_input('ped', prep_dwi, 'pe_dir')
pipeline.connect_input('pe_angle', prep_dwi, 'ped_polarity')
pipeline.connect(prep_dwi, 'pe', dwipreproc, 'pe_dir')
pipeline.connect(mrcat, 'out_file', dwipreproc, 'se_epi')
pipeline.connect(dwiextract, 'out_file', mrconvert, 'in_file')
pipeline.connect(mrconvert, 'out_file', mrcat, 'first_scan')
pipeline.connect_input('primary', extract_grad, 'in_file')
pipeline.connect(dwipreproc, 'out_file', swap, 'in_file')
# Connect outputs
pipeline.connect_output('preproc', swap, 'out_file')
pipeline.connect_output('grad_dirs', extract_grad, 'bvecs_file')
pipeline.connect_output('bvalues', extract_grad, 'bvals_file')
pipeline.connect_output('eddy_par', dwipreproc, 'eddy_parameters')
if self.switch('preproc_denoise'):
pipeline.connect_output('noise_residual', subtract, 'out_file')
# Check inputs/outputs are connected
return pipeline
def brain_extraction_pipeline(self,
**kwargs): # @UnusedVariable @IgnorePep8
"""
Generates a whole brain mask using MRtrix's 'dwi2mask' command
Parameters
----------
mask_tool: Str
Can be either 'bet' or 'dwi2mask' depending on which mask tool you
want to use
"""
if self.branch('brain_extract_method', 'mrtrix'):
pipeline = self.create_pipeline(
'brain_extraction',
inputs=[
DatasetSpec('preproc', nifti_gz_format),
DatasetSpec('grad_dirs', fsl_bvecs_format),
DatasetSpec('bvalues', fsl_bvals_format)
],
outputs=[DatasetSpec('brain_mask', nifti_gz_format)],
desc="Generate brain mask from b0 images",
version=1,
citations=[mrtrix_cite],
**kwargs)
# Create mask node
dwi2mask = pipeline.create_node(BrainMask(),
name='dwi2mask',
requirements=[mrtrix3_req])
dwi2mask.inputs.out_file = 'brain_mask.nii.gz'
# Gradient merge node
grad_fsl = pipeline.create_node(MergeTuple(2), name="grad_fsl")
# Connect nodes
pipeline.connect(grad_fsl, 'out', dwi2mask, 'grad_fsl')
# Connect inputs
pipeline.connect_input('grad_dirs', grad_fsl, 'in1')
pipeline.connect_input('bvalues', grad_fsl, 'in2')
pipeline.connect_input('preproc', dwi2mask, 'in_file')
# Connect outputs
pipeline.connect_output('brain_mask', dwi2mask, 'out_file')
# Check inputs/outputs are connected
pipeline.assert_connected()
else:
pipeline = super(DiffusionStudy,
self).brain_extraction_pipeline(**kwargs)
return pipeline
def bias_correct_pipeline(self, **kwargs): # @UnusedVariable @IgnorePep8
"""
Corrects B1 field inhomogeneities
"""
bias_method = self.switch('bias_correct_method')
pipeline = self.create_pipeline(
name='bias_correct',
inputs=[
DatasetSpec('preproc', nifti_gz_format),
DatasetSpec('brain_mask', nifti_gz_format),
DatasetSpec('grad_dirs', fsl_bvecs_format),
DatasetSpec('bvalues', fsl_bvals_format)
],
outputs=[DatasetSpec('bias_correct', nifti_gz_format)],
desc="Corrects for B1 field inhomogeneity",
version=1,
citations=[
fast_cite, (n4_cite if bias_method == 'ants' else fsl_cite)
],
**kwargs)
# Create bias correct node
bias_correct = pipeline.create_node(
DWIBiasCorrect(),
name="bias_correct",
requirements=(
[mrtrix3_req] +
[ants2_req if bias_method == 'ants' else fsl509_req]))
bias_correct.inputs.method = bias_method
# Gradient merge node
fsl_grads = pipeline.create_node(MergeTuple(2), name="fsl_grads")
# Connect nodes
pipeline.connect(fsl_grads, 'out', bias_correct, 'grad_fsl')
# Connect to inputs
pipeline.connect_input('grad_dirs', fsl_grads, 'in1')
pipeline.connect_input('bvalues', fsl_grads, 'in2')
pipeline.connect_input('preproc', bias_correct, 'in_file')
pipeline.connect_input('brain_mask', bias_correct, 'mask')
# Connect to outputs
pipeline.connect_output('bias_correct', bias_correct, 'out_file')
# Check inputs/output are connected
return pipeline
def intensity_normalisation_pipeline(self, **kwargs):
pipeline = self.create_pipeline(
name='intensity_normalization',
inputs=[
DatasetSpec('bias_correct', nifti_gz_format),
DatasetSpec('brain_mask', nifti_gz_format),
DatasetSpec('grad_dirs', fsl_bvecs_format),
DatasetSpec('bvalues', fsl_bvals_format)
],
outputs=[
DatasetSpec('norm_intensity', mrtrix_format),
DatasetSpec('norm_intens_fa_template',
mrtrix_format,
frequency='per_project'),
DatasetSpec('norm_intens_wm_mask',
mrtrix_format,
frequency='per_project')
],
desc="Corrects for B1 field inhomogeneity",
version=1,
citations=[mrtrix3_req],
**kwargs)
# Convert from nifti to mrtrix format
grad_merge = pipeline.create_node(MergeTuple(2), name="grad_merge")
mrconvert = pipeline.create_node(MRConvert(), name='mrconvert')
mrconvert.inputs.out_ext = '.mif'
# Set up join nodes
fields = ['dwis', 'masks', 'subject_ids', 'visit_ids']
join_subjects = pipeline.create_join_subjects_node(
IdentityInterface(fields), joinfield=fields, name='join_subjects')
join_visits = pipeline.create_join_visits_node(Chain(fields),
joinfield=fields,
name='join_visits')
# Set up expand nodes
select = pipeline.create_node(SelectSession(), name='expand')
# Intensity normalization
intensity_norm = pipeline.create_node(DWIIntensityNorm(),
name='dwiintensitynorm')
# Connect inputs
pipeline.connect_input('bias_correct', mrconvert, 'in_file')
pipeline.connect_input('grad_dirs', grad_merge, 'in1')
pipeline.connect_input('bvalues', grad_merge, 'in2')
pipeline.connect_subject_id(join_subjects, 'subject_ids')
pipeline.connect_visit_id(join_subjects, 'visit_ids')
pipeline.connect_subject_id(select, 'subject_id')
pipeline.connect_visit_id(select, 'visit_id')
pipeline.connect_input('brain_mask', join_subjects, 'masks')
# Internal connections
pipeline.connect(grad_merge, 'out', mrconvert, 'grad_fsl')
pipeline.connect(mrconvert, 'out_file', join_subjects, 'dwis')
pipeline.connect(join_subjects, 'dwis', join_visits, 'dwis')
pipeline.connect(join_subjects, 'masks', join_visits, 'masks')
pipeline.connect(join_subjects, 'subject_ids', join_visits,
'subject_ids')
pipeline.connect(join_subjects, 'visit_ids', join_visits, 'visit_ids')
pipeline.connect(join_visits, 'dwis', intensity_norm, 'in_files')
pipeline.connect(join_visits, 'masks', intensity_norm, 'masks')
pipeline.connect(join_visits, 'subject_ids', select, 'subject_ids')
pipeline.connect(join_visits, 'visit_ids', select, 'visit_ids')
pipeline.connect(intensity_norm, 'out_files', select, 'items')
# Connect outputs
pipeline.connect_output('norm_intensity', select, 'item')
pipeline.connect_output('norm_intens_fa_template', intensity_norm,
'fa_template')
pipeline.connect_output('norm_intens_wm_mask', intensity_norm,
'wm_mask')
return pipeline
def tensor_pipeline(self, **kwargs): # @UnusedVariable
"""
Fits the apparrent diffusion tensor (DT) to each voxel of the image
"""
pipeline = self.create_pipeline(
name='tensor',
inputs=[
DatasetSpec('bias_correct', nifti_gz_format),
DatasetSpec('grad_dirs', fsl_bvecs_format),
DatasetSpec('bvalues', fsl_bvals_format),
DatasetSpec('brain_mask', nifti_gz_format)
],
outputs=[DatasetSpec('tensor', nifti_gz_format)],
desc=("Estimates the apparent diffusion tensor in each "
"voxel"),
version=1,
citations=[],
**kwargs)
# Create tensor fit node
dwi2tensor = pipeline.create_node(FitTensor(), name='dwi2tensor')
dwi2tensor.inputs.out_file = 'dti.nii.gz'
# Gradient merge node
fsl_grads = pipeline.create_node(MergeTuple(2), name="fsl_grads")
# Connect nodes
pipeline.connect(fsl_grads, 'out', dwi2tensor, 'grad_fsl')
# Connect to inputs
pipeline.connect_input('grad_dirs', fsl_grads, 'in1')
pipeline.connect_input('bvalues', fsl_grads, 'in2')
pipeline.connect_input('bias_correct', dwi2tensor, 'in_file')
pipeline.connect_input('brain_mask', dwi2tensor, 'in_mask')
# Connect to outputs
pipeline.connect_output('tensor', dwi2tensor, 'out_file')
# Check inputs/output are connected
return pipeline
def fa_pipeline(self, **kwargs): # @UnusedVariable
"""
Fits the apparrent diffusion tensor (DT) to each voxel of the image
"""
pipeline = self.create_pipeline(
name='fa',
inputs=[
DatasetSpec('tensor', nifti_gz_format),
DatasetSpec('brain_mask', nifti_gz_format)
],
outputs=[
DatasetSpec('fa', nifti_gz_format),
DatasetSpec('adc', nifti_gz_format)
],
desc=("Calculates the FA and ADC from a tensor image"),
version=1,
citations=[],
**kwargs)
# Create tensor fit node
metrics = pipeline.create_node(TensorMetrics(),
name='metrics',
requirements=[mrtrix3_req])
metrics.inputs.out_fa = 'fa.nii.gz'
metrics.inputs.out_adc = 'adc.nii.gz'
# Connect to inputs
pipeline.connect_input('tensor', metrics, 'in_file')
pipeline.connect_input('brain_mask', metrics, 'in_mask')
# Connect to outputs
pipeline.connect_output('fa', metrics, 'out_fa')
pipeline.connect_output('adc', metrics, 'out_adc')
# Check inputs/output are connected
return pipeline
def response_pipeline(self, **kwargs): # @UnusedVariable
"""
Estimates the fibre orientation distribution (FOD) using constrained
spherical deconvolution
Parameters
----------
response_algorithm : str
Algorithm used to estimate the response
"""
outputs = [DatasetSpec('wm_response', text_format)]
if self.multi_tissue:
outputs.append(DatasetSpec('gm_response', text_format))
outputs.append(DatasetSpec('csf_response', text_format))
pipeline = self.create_pipeline(
name='response',
inputs=[
DatasetSpec('bias_correct', nifti_gz_format),
DatasetSpec('grad_dirs', fsl_bvecs_format),
DatasetSpec('bvalues', fsl_bvals_format),
DatasetSpec('brain_mask', nifti_gz_format)
],
outputs=outputs,
desc=("Estimates the fibre response function"),
version=1,
citations=[mrtrix_cite],
**kwargs)
# Create fod fit node
response = pipeline.create_node(ResponseSD(),
name='response',
requirements=[mrtrix3_req])
response.inputs.algorithm = self.switch('response_algorithm')
# Gradient merge node
fsl_grads = pipeline.create_node(MergeTuple(2), name="fsl_grads")
# Connect nodes
pipeline.connect(fsl_grads, 'out', response, 'grad_fsl')
# Connect to inputs
pipeline.connect_input('grad_dirs', fsl_grads, 'in1')
pipeline.connect_input('bvalues', fsl_grads, 'in2')
pipeline.connect_input('bias_correct', response, 'in_file')
pipeline.connect_input('brain_mask', response, 'in_mask')
# Connect to outputs
pipeline.connect_output('wm_response', response, 'wm_file')
if self.multi_tissue:
response.inputs.gm_file = 'gm.txt'
response.inputs.csf_file = 'csf.txt'
pipeline.connect_output('gm_response', response, 'gm_file')
pipeline.connect_output('csf_response', response, 'csf_file')
# Check inputs/output are connected
return pipeline
def average_response_pipeline(self, **kwargs):
"""
Averages the estimate response function over all subjects in the
project
"""
pipeline = self.create_pipeline(
name='average_response',
inputs=[DatasetSpec('wm_response', text_format)],
outputs=[
DatasetSpec('avg_response',
text_format,
frequency='per_project')
],
desc=("Averages the fibre response function over the project"),
version=1,
citations=[mrtrix_cite],
**kwargs)
join_subjects = pipeline.create_join_subjects_node(
IdentityInterface(['responses']),
name='join_subjects',
joinfield=['responses'])
join_visits = pipeline.create_join_visits_node(Chain(['responses']),
name='join_visits',
joinfield=['responses'])
avg_response = pipeline.create_node(AverageResponse(),
name='avg_response')
# Connect inputs
pipeline.connect_input('wm_response', join_subjects, 'responses')
# Connect inter-nodes
pipeline.connect(join_subjects, 'responses', join_visits, 'responses')
pipeline.connect(join_visits, 'responses', avg_response, 'in_files')
# Connect outputs
pipeline.connect_output('avg_response', avg_response, 'out_file')
# Check inputs/output are connected
return pipeline
#
# algorithm = traits.Enum(
# 'csd',
# 'msmt_csd',
# argstr='%s',
# position=-8,
# mandatory=True,
# desc='FOD algorithm')
# in_file = File(
# exists=True,
# argstr='%s',
# position=-7,
# mandatory=True,
# desc='input DWI image')
# wm_txt = File(
# argstr='%s', position=-6, mandatory=True, desc='WM response text file')
# wm_odf = File(
# 'wm.mif',
# argstr='%s',
# position=-5,
# usedefault=True,
# mandatory=True,
# desc='output WM ODF')
# gm_txt = File(argstr='%s', position=-4, desc='GM response text file')
# gm_odf = File('gm.mif', usedefault=True, argstr='%s',
# position=-3, desc='output GM ODF')
# csf_txt = File(argstr='%s', position=-2, desc='CSF response text file')
# csf_odf = File('csf.mif', usedefault=True, argstr='%s',
# position=-1, desc='output CSF ODF')
# mask_file = File(exists=True, argstr='-mask %s', desc='mask image')
def fod_pipeline(self, **kwargs): # @UnusedVariable
"""
Estimates the fibre orientation distribution (FOD) using constrained
spherical deconvolution
Parameters
----------
"""
inputs = [
DatasetSpec('bias_correct', nifti_gz_format),
DatasetSpec('grad_dirs', fsl_bvecs_format),
DatasetSpec('bvalues', fsl_bvals_format),
DatasetSpec('wm_response', text_format),
DatasetSpec('brain_mask', nifti_gz_format)
]
outputs = [DatasetSpec('wm_odf', mrtrix_format)]
if self.multi_tissue:
inputs.append(DatasetSpec('gm_response', text_format))
inputs.append(DatasetSpec('csf_response', text_format))
outputs.append(DatasetSpec('gm_odf', mrtrix_format))
outputs.append(DatasetSpec('csf_odf', mrtrix_format))
algorithm = 'msmt_csd'
else:
algorithm = 'csd'
pipeline = self.create_pipeline(
name='fod',
inputs=inputs,
outputs=outputs,
desc=("Estimates the fibre orientation distribution in each"
" voxel"),
version=1,
citations=[mrtrix_cite],
**kwargs)
# Create fod fit node
dwi2fod = pipeline.create_node(EstimateFOD(),
name='dwi2fod',
requirements=[mrtrix3_req])
dwi2fod.inputs.algorithm = algorithm
# Gradient merge node
fsl_grads = pipeline.create_node(MergeTuple(2), name="fsl_grads")
# Connect nodes
pipeline.connect(fsl_grads, 'out', dwi2fod, 'grad_fsl')
# Connect to inputs
pipeline.connect_input('grad_dirs', fsl_grads, 'in1')
pipeline.connect_input('bvalues', fsl_grads, 'in2')
pipeline.connect_input('bias_correct', dwi2fod, 'in_file')
pipeline.connect_input('wm_response', dwi2fod, 'wm_txt')
pipeline.connect_input('brain_mask', dwi2fod, 'mask_file')
# Connect to outputs
pipeline.connect_output('wm_odf', dwi2fod, 'wm_odf')
# If multi-tissue
if self.multi_tissue:
pipeline.connect_input('gm_response', dwi2fod, 'gm_txt')
pipeline.connect_input('csf_response', dwi2fod, 'csf_txt')
dwi2fod.inputs.gm_odf = 'gm.mif'
dwi2fod.inputs.csf_odf = 'csf.mif'
pipeline.connect_output('gm_odf', dwi2fod, 'gm_odf')
pipeline.connect_output('csf_odf', dwi2fod, 'csf_odf')
# Check inputs/output are connected
return pipeline
def tbss_pipeline(self, **kwargs): # @UnusedVariable
pipeline = self.create_pipeline(
name='tbss',
inputs=[DatasetSpec('fa', nifti_gz_format)],
outputs=[
DatasetSpec('tbss_mean_fa', nifti_gz_format),
DatasetSpec('tbss_proj_fa',
nifti_gz_format,
frequency='per_project'),
DatasetSpec('tbss_skeleton',
nifti_gz_format,
frequency='per_project'),
DatasetSpec('tbss_skeleton_mask',
nifti_gz_format,
frequency='per_project')
],
version=1,
citations=[tbss_cite, fsl_cite],
**kwargs)
# Create TBSS workflow
tbss = create_tbss_all(name='tbss')
# Connect inputs
pipeline.connect_input('fa', tbss, 'inputnode.fa_list')
# Connect outputs
pipeline.connect_output('tbss_mean_fa', tbss, 'outputnode.meanfa_file')
pipeline.connect_output('tbss_proj_fa', tbss,
'outputnode.projectedfa_file')
pipeline.connect_output('tbss_skeleton', tbss,
'outputnode.skeleton_file')
pipeline.connect_output('tbss_skeleton_mask', tbss,
'outputnode.skeleton_mask')
# Check inputs/output are connected
return pipeline
def extract_b0_pipeline(self, **kwargs): # @UnusedVariable
"""
Extracts the b0 images from a DWI study and takes their mean
"""
pipeline = self.create_pipeline(
name='extract_b0',
inputs=[
DatasetSpec('bias_correct', nifti_gz_format),
DatasetSpec('grad_dirs', fsl_bvecs_format),
DatasetSpec('bvalues', fsl_bvals_format)
],
outputs=[DatasetSpec('b0', nifti_gz_format)],
desc="Extract b0 image from a DWI study",
version=1,
citations=[mrtrix_cite],
**kwargs)
# Gradient merge node
fsl_grads = pipeline.create_node(MergeTuple(2), name="fsl_grads")
# Extraction node
extract_b0s = pipeline.create_node(ExtractDWIorB0(),
name='extract_b0s',
requirements=[mrtrix3_req])
extract_b0s.inputs.bzero = True
extract_b0s.inputs.quiet = True
# FIXME: Need a registration step before the mean
# Mean calculation node
mean = pipeline.create_node(MRMath(),
name="mean",
requirements=[mrtrix3_req])
mean.inputs.axis = 3
mean.inputs.operation = 'mean'
mean.inputs.quiet = True
# Convert to Nifti
mrconvert = pipeline.create_node(MRConvert(),
name="output_conversion",
requirements=[mrtrix3_req])
mrconvert.inputs.out_ext = '.nii.gz'
mrconvert.inputs.quiet = True
# Connect inputs
pipeline.connect_input('bias_correct', extract_b0s, 'in_file')
pipeline.connect_input('grad_dirs', fsl_grads, 'in1')
pipeline.connect_input('bvalues', fsl_grads, 'in2')
# Connect between nodes
pipeline.connect(extract_b0s, 'out_file', mean, 'in_files')
pipeline.connect(fsl_grads, 'out', extract_b0s, 'grad_fsl')
pipeline.connect(mean, 'out_file', mrconvert, 'in_file')
# Connect outputs
pipeline.connect_output('b0', mrconvert, 'out_file')
pipeline.assert_connected()
# Check inputs/outputs are connected
return pipeline
def track_gen_pipeline(self, **kwargs):
pipeline = self.create_pipeline(
name='extract_b0',
inputs=[
DatasetSpec('bias_correct', nifti_gz_format),
DatasetSpec('grad_dirs', fsl_bvecs_format),
DatasetSpec('bvalues', fsl_bvals_format)
],
outputs=[DatasetSpec('b0', nifti_gz_format)],
desc="Extract b0 image from a DWI study",
version=1,
citations=[mrtrix_cite])
return pipeline
def intrascan_alignment_pipeline(self, **kwargs):
pipeline = self.create_pipeline(
name='affine_mat_generation',
inputs=[
DatasetSpec('preproc', nifti_gz_format),
DatasetSpec('eddy_par', eddy_par_format)
],
outputs=[DatasetSpec('align_mats', directory_format)],
desc=("Generation of the affine matrices for the main dwi "
"sequence starting from eddy motion parameters"),
version=1,
citations=[fsl_cite],
**kwargs)
aff_mat = pipeline.create_node(AffineMatrixGeneration(),
name='gen_aff_mats')
pipeline.connect_input('preproc', aff_mat, 'reference_image')
pipeline.connect_input('eddy_par', aff_mat, 'motion_parameters')
pipeline.connect_output('align_mats', aff_mat, 'affine_matrices')
return pipeline
class MRIStudy(Study, metaclass=StudyMetaClass):
add_data_specs = [
DatasetSpec('primary', dicom_format),
DatasetSpec('coreg_ref_brain',
nifti_gz_format,
desc=("A reference scan to coregister the primary "
"scan to. Should be brain extracted"),
optional=True),
DatasetSpec('coreg_matrix', text_matrix_format,
'linear_coregistration_pipeline'),
DatasetSpec('preproc', nifti_gz_format, 'preproc_pipeline'),
DatasetSpec('brain',
nifti_gz_format,
'brain_extraction_pipeline',
desc="The brain masked image"),
DatasetSpec('brain_mask',
nifti_gz_format,
'brain_extraction_pipeline',
desc="Mask of the brain"),
DatasetSpec('coreg_brain',
nifti_gz_format,
'linear_coregistration_pipeline',
desc=""),
DatasetSpec('coreg_to_atlas', nifti_gz_format,
'coregister_to_atlas_pipeline'),
DatasetSpec('coreg_to_atlas_coeff', nifti_gz_format,
'coregister_to_atlas_pipeline'),
DatasetSpec('coreg_to_atlas_mat', text_matrix_format,
'coregister_to_atlas_pipeline'),
DatasetSpec('coreg_to_atlas_warp', nifti_gz_format,
'coregister_to_atlas_pipeline'),
DatasetSpec('coreg_to_atlas_report', gif_format,
'coregister_to_atlas_pipeline'),
DatasetSpec('wm_seg', nifti_gz_format, 'segmentation_pipeline'),
DatasetSpec('dcm_info', text_format,
'header_info_extraction_pipeline'),
DatasetSpec('motion_mats', motion_mats_format, 'motion_mat_pipeline'),
DatasetSpec('qformed', nifti_gz_format, 'qform_transform_pipeline'),
DatasetSpec('qform_mat', text_matrix_format,
'qform_transform_pipeline'),
FieldSpec('tr', float, 'header_info_extraction_pipeline'),
FieldSpec('start_time', str, 'header_info_extraction_pipeline'),
FieldSpec('real_duration', str, 'header_info_extraction_pipeline'),
FieldSpec('tot_duration', str, 'header_info_extraction_pipeline'),
FieldSpec('ped', str, 'header_info_extraction_pipeline'),
FieldSpec('pe_angle', str, 'header_info_extraction_pipeline')
]
add_parameter_specs = [
ParameterSpec('bet_robust', True),
ParameterSpec('bet_f_threshold', 0.5),
ParameterSpec('bet_reduce_bias', False),
ParameterSpec('bet_g_threshold', 0.0),
ParameterSpec('MNI_template',
os.path.join(atlas_path, 'MNI152_T1_2mm.nii.gz')),
ParameterSpec('MNI_template_brain',
os.path.join(atlas_path, 'MNI152_T1_2mm_brain.nii.gz')),
ParameterSpec(
'MNI_template_mask',
os.path.join(atlas_path, 'MNI152_T1_2mm_brain_mask.nii.gz')),
ParameterSpec('optibet_gen_report', False),
ParameterSpec('fnirt_atlas', 'MNI152'),
ParameterSpec('fnirt_resolution', '2mm'),
ParameterSpec('fnirt_intensity_model', 'global_non_linear_with_bias'),
ParameterSpec('fnirt_subsampling', [4, 4, 2, 2, 1, 1]),
ParameterSpec('preproc_new_dims', ('RL', 'AP', 'IS')),
ParameterSpec('preproc_resolution', None, dtype=list),
ParameterSpec(
'flirt_degrees_of_freedom',
6,
desc=("Number of degrees of freedom used in the registration. "
"Default is 6 -> affine transformation.")),
ParameterSpec(
'flirt_cost_func',
'normmi',
desc=("Cost function used for the registration. Can be one of "
"'mutualinfo', 'corratio', 'normcorr', 'normmi', 'leastsq',"
" 'labeldiff', 'bbr'")),
ParameterSpec(
'flirt_qsform',
False,
desc=("Whether to use the QS form supplied in the input image "
"header (the image coordinates of the FOV supplied by the "
"scanner"))
]
add_switch_specs = [
SwitchSpec('linear_reg_method',
'flirt',
choices=('flirt', 'spm', 'ants')),
SwitchSpec('atlas_coreg_tool', 'ants', choices=('fnirt', 'ants')),
SwitchSpec('bet_method', 'fsl_bet', choices=('fsl_bet', 'optibet'))
]
@property
def coreg_brain_spec(self):
"""
The name of the dataset after registration has been applied.
If registration is not required, i.e. a reg_ref is not supplied
then it is simply the 'brain' dataset.
"""
if 'coreg_ref_brain' in self.input_names:
name = 'coreg_brain'
else:
name = 'brain'
return DatasetSpec(name, nifti_gz_format)
def linear_coregistration_pipeline(self, **kwargs):
if self.branch('linear_reg_method', 'flirt'):
pipeline = self._flirt_factory('linear_coreg', 'brain',
'coreg_ref_brain', 'coreg_brain',
'coreg_matrix', **kwargs)
elif self.branch('linear_reg_method', 'ants'):
pipeline = self._ants_linear_coreg_pipeline(
'linear_coreg', 'brain', 'coreg_ref_brain', 'coreg_brain',
'coreg_matrix', **kwargs)
elif self.branch('linear_reg_method', 'spm'):
raise NotImplementedError
else:
self.unhandled_branch('linear_reg_method')
return pipeline
def qform_transform_pipeline(self, **kwargs):
return self._qform_transform_factory('qform_transform', 'brain',
'coreg_ref_brain', 'qformed',
'qform_mat', **kwargs)
def _flirt_factory(self, name, to_reg, ref, reg, matrix, **kwargs):
"""
Registers a MR scan to a refernce MR scan using FSL's FLIRT command
Parameters
----------
name : str
Name for the generated pipeline
to_reg : str
Name of the DatasetSpec to register
ref : str
Name of the DatasetSpec to use as a reference
reg : str
Name of the DatasetSpec to output as registered image
matrix : str
Name of the DatasetSpec to output as registration matrix
"""
pipeline = self.create_pipeline(
name=name,
inputs=[
DatasetSpec(to_reg, nifti_gz_format),
DatasetSpec(ref, nifti_gz_format)
],
outputs=[
DatasetSpec(reg, nifti_gz_format),
DatasetSpec(matrix, text_matrix_format)
],
desc="Registers a MR scan against a reference image using FLIRT",
version=1,
citations=[fsl_cite],
**kwargs)
flirt = pipeline.create_node(interface=FLIRT(),
name='flirt',
requirements=[fsl5_req],
wall_time=5)
# Set registration parameters
flirt.inputs.dof = self.parameter('flirt_degrees_of_freedom')
flirt.inputs.cost = self.parameter('flirt_cost_func')
flirt.inputs.cost_func = self.parameter('flirt_cost_func')
flirt.inputs.output_type = 'NIFTI_GZ'
# Connect inputs
pipeline.connect_input(to_reg, flirt, 'in_file')
pipeline.connect_input(ref, flirt, 'reference')
# Connect outputs
pipeline.connect_output(reg, flirt, 'out_file')
pipeline.connect_output(matrix, flirt, 'out_matrix_file')
return pipeline
def _qform_transform_factory(self, name, to_reg, ref, qformed, qformed_mat,
**kwargs):
pipeline = self.create_pipeline(
name=name,
inputs=[
DatasetSpec(to_reg, nifti_gz_format),
DatasetSpec(ref, nifti_gz_format)
],
outputs=[
DatasetSpec(qformed, nifti_gz_format),
DatasetSpec(qformed_mat, text_matrix_format)
],
desc="Registers a MR scan against a reference image",
version=1,
citations=[fsl_cite],
**kwargs)
flirt = pipeline.create_node(interface=FLIRT(),
name='flirt',
requirements=[fsl5_req],
wall_time=5)
flirt.inputs.uses_qform = True
flirt.inputs.apply_xfm = True
# Connect inputs
pipeline.connect_input(to_reg, flirt, 'in_file')
pipeline.connect_input(ref, flirt, 'reference')
# Connect outputs
pipeline.connect_output(qformed, flirt, 'out_file')
pipeline.connect_output(qformed_mat, flirt, 'out_matrix_file')
return pipeline
def _spm_coreg_pipeline(self, **kwargs): # @UnusedVariable
"""
Coregisters T2 image to T1 image using SPM's
"Register" method.
NB: Default values come from the W2MHS toolbox
"""
pipeline = self.create_pipeline(
name='registration',
inputs=[
DatasetSpec('t1', nifti_format),
DatasetSpec('t2', nifti_format)
],
outputs=[DatasetSpec('t2_coreg_t1', nifti_format)],
desc="Coregister T2-weighted images to T1",
version=1,
citations=[spm_cite],
**kwargs)
coreg = pipeline.create_node(Coregister(),
name='coreg',
requirements=[spm12_req],
wall_time=30)
coreg.inputs.jobtype = 'estwrite'
coreg.inputs.cost_function = 'nmi'
coreg.inputs.separation = [4, 2]
coreg.inputs.tolerance = [
0.02, 0.02, 0.02, 0.001, 0.001, 0.001, 0.01, 0.01, 0.01, 0.001,
0.001, 0.001
]
coreg.inputs.fwhm = [7, 7]
coreg.inputs.write_interp = 4
coreg.inputs.write_wrap = [0, 0, 0]
coreg.inputs.write_mask = False
coreg.inputs.out_prefix = 'r'
# Connect inputs
pipeline.connect_input('t1', coreg, 'target')
pipeline.connect_input('t2', coreg, 'source')
# Connect outputs
pipeline.connect_output('t2_coreg_t1', coreg, 'coregistered_source')
return pipeline
def _ants_linear_coreg_pipeline(self, name, to_reg, ref, reg, matrix,
**kwargs):
pipeline = self.create_pipeline(
name=name,
inputs=[
DatasetSpec(to_reg, nifti_gz_format),
DatasetSpec(ref, nifti_gz_format)
],
outputs=[
DatasetSpec(reg, nifti_gz_format),
DatasetSpec(matrix, text_matrix_format)
],
desc="Registers a MR scan against a reference image using ANTs",
version=1,
citations=[],
**kwargs)
ants_linear = pipeline.create_node(AntsRegSyn(num_dimensions=3,
transformation='r',
out_prefix='reg2hires'),
name='ANTs_linear_Reg',
wall_time=10,
requirements=[ants2_req])
pipeline.connect_input(ref, ants_linear, 'ref_file')
pipeline.connect_input(to_reg, ants_linear, 'input_file')
pipeline.connect_output(reg, ants_linear, 'reg_file')
pipeline.connect_output(matrix, ants_linear, 'regmat')
return pipeline
def brain_extraction_pipeline(self, in_file='preproc', **kwargs):
if self.branch('bet_method', 'fsl_bet'):
pipeline = self._fsl_bet_brain_extraction_pipeline(
in_file, **kwargs)
elif self.branch('bet_method', 'optibet'):
pipeline = self._optiBET_brain_extraction_pipeline(
in_file, **kwargs)
else:
self.unhandled_branch('bet_method')
return pipeline
def _fsl_bet_brain_extraction_pipeline(self, in_file, **kwargs):
"""
Generates a whole brain mask using FSL's BET command.
"""
pipeline = self.create_pipeline(
name='brain_extraction',
inputs=[DatasetSpec(in_file, nifti_gz_format)],
outputs=[
DatasetSpec('brain', nifti_gz_format),
DatasetSpec('brain_mask', nifti_gz_format)
],
desc="Generate brain mask from mr_scan",
version=1,
citations=[fsl_cite, bet_cite, bet2_cite],
**kwargs)
# Create mask node
bet = pipeline.create_node(interface=fsl.BET(),
name="bet",
requirements=[fsl509_req])
bet.inputs.mask = True
bet.inputs.output_type = 'NIFTI_GZ'
if self.parameter('bet_robust'):
bet.inputs.robust = True
if self.parameter('bet_reduce_bias'):
bet.inputs.reduce_bias = True
bet.inputs.frac = self.parameter('bet_f_threshold')
bet.inputs.vertical_gradient = self.parameter('bet_g_threshold')
# Connect inputs/outputs
pipeline.connect_input(in_file, bet, 'in_file')
pipeline.connect_output('brain', bet, 'out_file')
pipeline.connect_output('brain_mask', bet, 'mask_file')
return pipeline
def _optiBET_brain_extraction_pipeline(self, in_file, **kwargs):
"""
Generates a whole brain mask using a modified optiBET approach.
"""
outputs = [
DatasetSpec('brain', nifti_gz_format),
DatasetSpec('brain_mask', nifti_gz_format)
]
if self.parameter('optibet_gen_report'):
outputs.append(DatasetSpec('optiBET_report', gif_format))
pipeline = self.create_pipeline(
name='brain_extraction',
inputs=[DatasetSpec(in_file, nifti_gz_format)],
outputs=outputs,
desc=("Modified implementation of optiBET.sh"),
version=1,
citations=[fsl_cite],
**kwargs)
mni_reg = pipeline.create_node(AntsRegSyn(num_dimensions=3,
transformation='s',
out_prefix='T12MNI',
num_threads=4),
name='T1_reg',
wall_time=25,
requirements=[ants2_req])
mni_reg.inputs.ref_file = self.parameter('MNI_template')
pipeline.connect_input(in_file, mni_reg, 'input_file')
merge_trans = pipeline.create_node(Merge(2),
name='merge_transforms',
wall_time=1)
pipeline.connect(mni_reg, 'inv_warp', merge_trans, 'in1')
pipeline.connect(mni_reg, 'regmat', merge_trans, 'in2')
trans_flags = pipeline.create_node(Merge(2),
name='trans_flags',
wall_time=1)
trans_flags.inputs.in1 = False
trans_flags.inputs.in2 = True
apply_trans = pipeline.create_node(ApplyTransforms(),
name='ApplyTransform',
wall_time=7,
memory=24000,
requirements=[ants2_req])
apply_trans.inputs.input_image = self.parameter('MNI_template_mask')
apply_trans.inputs.interpolation = 'NearestNeighbor'
apply_trans.inputs.input_image_type = 3
pipeline.connect(merge_trans, 'out', apply_trans, 'transforms')
pipeline.connect(trans_flags, 'out', apply_trans,
'invert_transform_flags')
pipeline.connect_input(in_file, apply_trans, 'reference_image')
maths1 = pipeline.create_node(fsl.ImageMaths(
suffix='_optiBET_brain_mask', op_string='-bin'),
name='binarize',
wall_time=5,
requirements=[fsl5_req])
pipeline.connect(apply_trans, 'output_image', maths1, 'in_file')
maths2 = pipeline.create_node(fsl.ImageMaths(suffix='_optiBET_brain',
op_string='-mas'),
name='mask',
wall_time=5,
requirements=[fsl5_req])
pipeline.connect_input(in_file, maths2, 'in_file')
pipeline.connect(maths1, 'out_file', maths2, 'in_file2')
if self.parameter('optibet_gen_report'):
slices = pipeline.create_node(FSLSlices(),
name='slices',
wall_time=5,
requirements=[fsl5_req])
slices.inputs.outname = 'optiBET_report'
pipeline.connect_input(in_file, slices, 'im1')
pipeline.connect(maths2, 'out_file', slices, 'im2')
pipeline.connect_output('optiBET_report', slices, 'report')
pipeline.connect_output('brain_mask', maths1, 'out_file')
pipeline.connect_output('brain', maths2, 'out_file')
return pipeline
def coregister_to_atlas_pipeline(self, **kwargs):
if self.branch('atlas_coreg_tool', 'fnirt'):
pipeline = self._fsl_fnirt_to_atlas_pipeline(**kwargs)
elif self.branch('atlas_coreg_tool', 'ants'):
pipeline = self._ants_to_atlas_pipeline(**kwargs)
else:
self.unhandled_branch('atlas_coreg_tool')
return pipeline
# @UnusedVariable @IgnorePep8
def _fsl_fnirt_to_atlas_pipeline(self, **kwargs):
"""
Registers a MR scan to a refernce MR scan using FSL's nonlinear FNIRT
command
Parameters
----------
atlas : Which atlas to use, can be one of 'mni_nl6'
"""
pipeline = self.create_pipeline(
name='coregister_to_atlas',
inputs=[
DatasetSpec('preproc', nifti_gz_format),
DatasetSpec('brain_mask', nifti_gz_format),
DatasetSpec('brain', nifti_gz_format)
],
outputs=[
DatasetSpec('coreg_to_atlas', nifti_gz_format),
DatasetSpec('coreg_to_atlas_coeff', nifti_gz_format)
],
desc=("Nonlinearly registers a MR scan to a standard space,"
"e.g. MNI-space"),
version=1,
citations=[fsl_cite],
**kwargs)
# Get the reference atlas from FSL directory
ref_atlas = get_atlas_path(self.parameter('fnirt_atlas'),
'image',
resolution=self.parameter('resolution'))
ref_mask = get_atlas_path(self.parameter('fnirt_atlas'),
'mask_dilated',
resolution=self.parameter('resolution'))
ref_brain = get_atlas_path(self.parameter('fnirt_atlas'),
'brain',
resolution=self.parameter('resolution'))
# Basic reorientation to standard MNI space
reorient = pipeline.create_node(Reorient2Std(),
name='reorient',
requirements=[fsl5_req])
reorient.inputs.output_type = 'NIFTI_GZ'
reorient_mask = pipeline.create_node(Reorient2Std(),
name='reorient_mask',
requirements=[fsl5_req])
reorient_mask.inputs.output_type = 'NIFTI_GZ'
reorient_brain = pipeline.create_node(Reorient2Std(),
name='reorient_brain',
requirements=[fsl5_req])
reorient_brain.inputs.output_type = 'NIFTI_GZ'
# Affine transformation to MNI space
flirt = pipeline.create_node(interface=FLIRT(),
name='flirt',
requirements=[fsl5_req],
wall_time=5)
flirt.inputs.reference = ref_brain
flirt.inputs.dof = 12
flirt.inputs.output_type = 'NIFTI_GZ'
# Nonlinear transformation to MNI space
fnirt = pipeline.create_node(interface=FNIRT(),
name='fnirt',
requirements=[fsl5_req],
wall_time=60)
fnirt.inputs.ref_file = ref_atlas
fnirt.inputs.refmask_file = ref_mask
fnirt.inputs.output_type = 'NIFTI_GZ'
intensity_model = self.parameter('fnirt_intensity_model')
if intensity_model is None:
intensity_model = 'none'
fnirt.inputs.intensity_mapping_model = intensity_model
fnirt.inputs.subsampling_scheme = self.parameter('fnirt_subsampling')
fnirt.inputs.fieldcoeff_file = True
fnirt.inputs.in_fwhm = [8, 6, 5, 4.5, 3, 2]
fnirt.inputs.ref_fwhm = [8, 6, 5, 4, 2, 0]
fnirt.inputs.regularization_lambda = [300, 150, 100, 50, 40, 30]
fnirt.inputs.apply_intensity_mapping = [1, 1, 1, 1, 1, 0]
fnirt.inputs.max_nonlin_iter = [5, 5, 5, 5, 5, 10]
# Apply mask if corresponding subsampling scheme is 1
# (i.e. 1-to-1 resolution) otherwise don't.
apply_mask = [int(s == 1) for s in self.parameter('fnirt_subsampling')]
fnirt.inputs.apply_inmask = apply_mask
fnirt.inputs.apply_refmask = apply_mask
# Connect nodes
pipeline.connect(reorient_brain, 'out_file', flirt, 'in_file')
pipeline.connect(reorient, 'out_file', fnirt, 'in_file')
pipeline.connect(reorient_mask, 'out_file', fnirt, 'inmask_file')
pipeline.connect(flirt, 'out_matrix_file', fnirt, 'affine_file')
# Set registration parameters
# TOD: Need to work out which parameters to use
# Connect inputs
pipeline.connect_input('preproc', reorient, 'in_file')
pipeline.connect_input('brain_mask', reorient_mask, 'in_file')
pipeline.connect_input('brain', reorient_brain, 'in_file')
# Connect outputs
pipeline.connect_output('coreg_to_atlas', fnirt, 'warped_file')
pipeline.connect_output('coreg_to_atlas_coeff', fnirt,
'fieldcoeff_file')
return pipeline
def _ants_to_atlas_pipeline(self, **kwargs):
pipeline = self.create_pipeline(
name='coregister_to_atlas',
inputs=[DatasetSpec('coreg_ref_brain', nifti_gz_format)],
outputs=[
DatasetSpec('coreg_to_atlas', nifti_gz_format),
DatasetSpec('coreg_to_atlas_mat', text_matrix_format),
DatasetSpec('coreg_to_atlas_warp', nifti_gz_format),
DatasetSpec('coreg_to_atlas_report', gif_format)
],
desc=("Nonlinearly registers a MR scan to a standard space,"
"e.g. MNI-space"),
version=1,
citations=[fsl_cite],
**kwargs)
ants_reg = pipeline.create_node(AntsRegSyn(num_dimensions=3,
transformation='s',
out_prefix='Struct2MNI',
num_threads=4),
name='Struct2MNI_reg',
wall_time=25,
requirements=[ants2_req])
ref_brain = self.parameter('MNI_template_brain')
ants_reg.inputs.ref_file = ref_brain
pipeline.connect_input('coreg_ref_brain', ants_reg, 'input_file')
slices = pipeline.create_node(FSLSlices(),
name='slices',
wall_time=1,
requirements=[fsl5_req])
slices.inputs.outname = 'coreg_to_atlas_report'
slices.inputs.im1 = self.parameter('MNI_template')
pipeline.connect(ants_reg, 'reg_file', slices, 'im2')
pipeline.connect_output('coreg_to_atlas', ants_reg, 'reg_file')
pipeline.connect_output('coreg_to_atlas_mat', ants_reg, 'regmat')
pipeline.connect_output('coreg_to_atlas_warp', ants_reg, 'warp_file')
pipeline.connect_output('coreg_to_atlas_report', slices, 'report')
return pipeline
def segmentation_pipeline(self, img_type=2, **kwargs):
pipeline = self.create_pipeline(
name='FAST_segmentation',
inputs=[DatasetSpec('brain', nifti_gz_format)],
outputs=[DatasetSpec('wm_seg', nifti_gz_format)],
desc="White matter segmentation of the reference image",
version=1,
citations=[fsl_cite],
**kwargs)
fast = pipeline.create_node(fsl.FAST(),
name='fast',
requirements=[fsl509_req])
fast.inputs.img_type = img_type
fast.inputs.segments = True
fast.inputs.out_basename = 'Reference_segmentation'
pipeline.connect_input('brain', fast, 'in_files')
split = pipeline.create_node(Split(), name='split')
split.inputs.splits = [1, 1, 1]
split.inputs.squeeze = True
pipeline.connect(fast, 'tissue_class_files', split, 'inlist')
if img_type == 1:
pipeline.connect_output('wm_seg', split, 'out3')
elif img_type == 2:
pipeline.connect_output('wm_seg', split, 'out2')
else:
raise ArcanaUsageError(
"'img_type' parameter can either be 1 or 2 (not {})".format(
img_type))
return pipeline
def preproc_pipeline(self, in_file_name='primary', **kwargs):
"""
Performs basic preprocessing, such as swapping dimensions into
standard orientation and resampling (if required)
Parameters
-------
new_dims : tuple(str)[3]
A 3-tuple with the new orientation of the image (see FSL
swap dim)
resolution : list(float)[3] | None
New resolution of the image. If None no resampling is
performed
"""
pipeline = self.create_pipeline(
name='preproc_pipeline',
inputs=[DatasetSpec(in_file_name, nifti_gz_format)],
outputs=[DatasetSpec('preproc', nifti_gz_format)],
desc=("Dimensions swapping to ensure that all the images "
"have the same orientations."),
version=1,
citations=[fsl_cite],
**kwargs)
swap = pipeline.create_node(fsl.utils.Reorient2Std(),
name='fslreorient2std',
requirements=[fsl509_req])
# swap.inputs.new_dims = self.parameter('preproc_new_dims')
pipeline.connect_input(in_file_name, swap, 'in_file')
if self.parameter('preproc_resolution') is not None:
resample = pipeline.create_node(MRResize(),
name="resample",
requirements=[mrtrix3_req])
resample.inputs.voxel = self.parameter('preproc_resolution')
pipeline.connect(swap, 'out_file', resample, 'in_file')
pipeline.connect_output('preproc', resample, 'out_file')
else:
pipeline.connect_output('preproc', swap, 'out_file')
return pipeline
def header_info_extraction_pipeline(self, **kwargs):
if self.input('primary').format != dicom_format:
raise ArcanaUsageError(
"Can only extract header info if 'primary' dataset "
"is provided in DICOM format ({})".format(
self.input('primary').format))
return self.header_info_extraction_pipeline_factory(
'header_info_extraction', 'primary', **kwargs)
def header_info_extraction_pipeline_factory(self,
name,
dcm_in_name,
multivol=False,
output_prefix='',
**kwargs):
tr = output_prefix + 'tr'
start_time = output_prefix + 'start_time'
tot_duration = output_prefix + 'tot_duration'
real_duration = output_prefix + 'real_duration'
ped = output_prefix + 'ped'
pe_angle = output_prefix + 'pe_angle'
dcm_info = output_prefix + 'dcm_info'
outputs = [
FieldSpec(tr, dtype=float),
FieldSpec(start_time, dtype=str),
FieldSpec(tot_duration, dtype=str),
FieldSpec(real_duration, dtype=str),
FieldSpec(ped, dtype=str),
FieldSpec(pe_angle, dtype=str),
DatasetSpec(dcm_info, text_format)
]
pipeline = self.create_pipeline(
name=name,
inputs=[DatasetSpec(dcm_in_name, dicom_format)],
outputs=outputs,
desc=("Pipeline to extract the most important scan "
"information from the image header"),
version=1,
citations=[],
**kwargs)
hd_extraction = pipeline.create_node(DicomHeaderInfoExtraction(),
name='hd_info_extraction')
hd_extraction.inputs.multivol = multivol
pipeline.connect_input(dcm_in_name, hd_extraction, 'dicom_folder')
pipeline.connect_output(tr, hd_extraction, 'tr')
pipeline.connect_output(start_time, hd_extraction, 'start_time')
pipeline.connect_output(tot_duration, hd_extraction, 'tot_duration')
pipeline.connect_output(real_duration, hd_extraction, 'real_duration')
pipeline.connect_output(ped, hd_extraction, 'ped')
pipeline.connect_output(pe_angle, hd_extraction, 'pe_angle')
pipeline.connect_output(dcm_info, hd_extraction, 'dcm_info')
return pipeline
def motion_mat_pipeline(self, **kwargs):
if not self.spec('coreg_matrix').derivable:
logger.info("Cannot derive 'coreg_matrix' for {} required for "
"motion matrix calculation, assuming that it "
"is the reference study".format(self))
inputs = [DatasetSpec('primary', dicom_format)]
ref = True
else:
inputs = [
DatasetSpec('coreg_matrix', text_matrix_format),
DatasetSpec('qform_mat', text_matrix_format)
]
if 'align_mats' in self.data_spec_names():
inputs.append(DatasetSpec('align_mats', directory_format))
ref = False
pipeline = self.create_pipeline(
name='motion_mat_calculation',
inputs=inputs,
outputs=[DatasetSpec('motion_mats', motion_mats_format)],
desc=("Motion matrices calculation"),
version=1,
citations=[fsl_cite],
**kwargs)
mm = pipeline.create_node(MotionMatCalculation(), name='motion_mats')
if ref:
mm.inputs.reference = True
pipeline.connect_input('primary', mm, 'dummy_input')
else:
pipeline.connect_input('coreg_matrix', mm, 'reg_mat')
pipeline.connect_input('qform_mat', mm, 'qform_mat')
if 'align_mats' in self.data_spec_names():
pipeline.connect_input('align_mats', mm, 'align_mats')
pipeline.connect_output('motion_mats', mm, 'motion_mats')
return pipeline
class TestDerivableStudy(with_metaclass(StudyMetaClass, Study)):
add_data_specs = [
AcquiredFilesetSpec('required', text_format),
AcquiredFilesetSpec('optional', text_format, optional=True),
FilesetSpec('derivable', text_format, 'pipeline1'),
FilesetSpec('missing_input', text_format, 'pipeline2'),
FilesetSpec('another_derivable', text_format, 'pipeline3'),
FilesetSpec('requires_switch', text_format, 'pipeline3'),
FilesetSpec('requires_switch2', text_format, 'pipeline4'),
FilesetSpec('requires_foo', text_format, 'pipeline5'),
FilesetSpec('requires_bar', text_format, 'pipeline5')]
add_param_specs = [
SwitchSpec('switch', False),
SwitchSpec('branch', 'foo', ('foo', 'bar', 'wee'))]
def pipeline1(self, **name_maps):
pipeline = self.pipeline(
'pipeline1',
desc="",
references=[],
name_maps=name_maps)
identity = pipeline.add('identity', IdentityInterface(['a']))
pipeline.connect_input('required', identity, 'a')
pipeline.connect_output('derivable', identity, 'a')
return pipeline
def pipeline2(self, **name_maps):
pipeline = self.pipeline(
'pipeline2',
desc="",
references=[],
name_maps=name_maps)
identity = pipeline.add('identity', IdentityInterface(['a', 'b']))
pipeline.connect_input('required', identity, 'a')
pipeline.connect_input('optional', identity, 'b')
pipeline.connect_output('missing_input', identity, 'a')
return pipeline
def pipeline3(self, **name_maps):
pipeline = self.pipeline(
'pipeline3',
desc="",
references=[],
name_maps=name_maps)
identity = pipeline.add('identity', IdentityInterface(['a', 'b']))
pipeline.connect_input('required', identity, 'a')
pipeline.connect_input('required', identity, 'b')
pipeline.connect_output('another_derivable', identity, 'a')
if self.branch('switch'):
pipeline.connect_output('requires_switch', identity, 'b')
return pipeline
def pipeline4(self, **name_maps):
pipeline = self.pipeline(
'pipeline4',
desc="",
references=[],
name_maps=name_maps)
identity = pipeline.add('identity', IdentityInterface(['a']))
pipeline.connect_input('requires_switch', identity, 'a')
pipeline.connect_output('requires_switch2', identity, 'a')
return pipeline
def pipeline5(self, **name_maps):
pipeline = self.pipeline(
'pipeline5',
desc="",
references=[],
name_maps=name_maps)
identity = pipeline.add('identity', IdentityInterface(['a']))
pipeline.connect_input('required', identity, 'a')
if self.branch('branch', 'foo'):
pipeline.connect_output('requires_foo', identity, 'a')
elif self.branch('branch', 'bar'):
pipeline.connect_output('requires_bar', identity, 'a')
else:
self.unhandled_branch('branch')
return pipeline