def calc_fd(mat_file, dataset="fnl"):
"""
Compute Framewise Displacement as in Power et al 2012 and save to file. Uses the default radius of 50mm.
Args:
mat_file (path object): .mat file outputted from FSL
dataset (str): 'fnl' or 'sherlock'
Returns:
file_path
"""
assert isinstance(mat_file, Path), "realigned_series must be a Path object"
out_file = mat_file.parent / "fd_power_2012.csv"
fd = FramewiseDisplacement(
in_file=mat_file, out_file=out_file, parameter_source="FSL", save_plot=False
)
if dataset == "fnl":
fd.inputs.series_tr = 2.0
elif dataset == "sherlock":
fd.inputs.series_tr = 1.5
_ = fd.run()
return out_file
def test_fd():
tempdir = mkdtemp()
ground_truth = np.loadtxt(example_data('fsl_motion_outliers_fd.txt'))
fd = FramewiseDisplacement(in_plots=example_data('fsl_mcflirt_movpar.txt'),
out_file=tempdir + '/fd.txt')
res = fd.run()
yield assert_equal, np.allclose(ground_truth, np.loadtxt(res.outputs.out_file)), True
yield assert_equal, np.abs(ground_truth.mean() - res.outputs.fd_average) < 1e-4, True
rmtree(tempdir)
def test_fd(tmpdir):
tempdir = str(tmpdir)
ground_truth = np.loadtxt(example_data('fsl_motion_outliers_fd.txt'))
fdisplacement = FramewiseDisplacement(in_plots=example_data('fsl_mcflirt_movpar.txt'),
out_file=tempdir + '/fd.txt')
res = fdisplacement.run()
with open(res.outputs.out_file) as all_lines:
for line in all_lines:
assert 'FramewiseDisplacement' in line
break
assert np.allclose(ground_truth, np.loadtxt(res.outputs.out_file, skiprows=1), atol=.16)
assert np.abs(ground_truth.mean() - res.outputs.fd_average) < 1e-2
def test_fd():
tempdir = mkdtemp()
ground_truth = np.loadtxt(example_data('fsl_motion_outliers_fd.txt'))
fdisplacement = FramewiseDisplacement(
in_plots=example_data('fsl_mcflirt_movpar.txt'),
out_file=tempdir + '/fd.txt')
res = fdisplacement.run()
yield assert_true, np.allclose(ground_truth,
np.loadtxt(res.outputs.out_file),
atol=.16)
yield assert_true, np.abs(ground_truth.mean() -
res.outputs.fd_average) < 1e-2
rmtree(tempdir)
def test_fd():
tempdir = mkdtemp()
ground_truth = np.loadtxt(example_data('fsl_motion_outliers_fd.txt'))
fdisplacement = FramewiseDisplacement(in_plots=example_data('fsl_mcflirt_movpar.txt'),
out_file=tempdir + '/fd.txt')
res = fdisplacement.run()
with open(res.outputs.out_file) as all_lines:
for line in all_lines:
yield assert_in, 'framewise_displacement', line
break
yield assert_true, np.allclose(ground_truth, np.loadtxt(res.outputs.out_file), atol=.16)
yield assert_true, np.abs(ground_truth.mean() - res.outputs.fd_average) < 1e-2
rmtree(tempdir)
def test_fd(tmpdir):
tempdir = str(tmpdir)
ground_truth = np.loadtxt(example_data('fsl_motion_outliers_fd.txt'))
fdisplacement = FramewiseDisplacement(
in_plots=example_data('fsl_mcflirt_movpar.txt'),
out_file=tempdir + '/fd.txt')
res = fdisplacement.run()
with open(res.outputs.out_file) as all_lines:
for line in all_lines:
assert 'FramewiseDisplacement' in line
break
assert np.allclose(ground_truth,
np.loadtxt(res.outputs.out_file, skiprows=1),
atol=.16)
assert np.abs(ground_truth.mean() - res.outputs.fd_average) < 1e-2
def test_fd(tmpdir):
tempdir = tmpdir.strpath
ground_truth = np.loadtxt(example_data("fsl_motion_outliers_fd.txt"))
fdisplacement = FramewiseDisplacement(
in_file=example_data("fsl_mcflirt_movpar.txt"),
out_file=tempdir + "/fd.txt",
parameter_source="FSL",
)
res = fdisplacement.run()
with open(res.outputs.out_file) as all_lines:
for line in all_lines:
assert "FramewiseDisplacement" in line
break
assert np.allclose(ground_truth,
np.loadtxt(res.outputs.out_file, skiprows=1),
atol=0.16)
assert np.abs(ground_truth.mean() - res.outputs.fd_average) < 1e-2
def create_motion_confound_workflow(order=2,
fd_cutoff=.2,
name='motion_confound'):
input_node = pe.Node(interface=IdentityInterface(
fields=['par_file', 'output_directory', 'sub_id']),
name='inputspec')
output_node = pe.Node(
interface=IdentityInterface(fields=['out_fd', 'out_ext_moco']),
name='outputspec')
datasink = pe.Node(DataSink(), name='sinker')
datasink.inputs.parameterization = False
extend_motion_parameters = pe.MapNode(Extend_motion_parameters,
iterfield=['par_file'],
name='extend_motion_parameters')
extend_motion_parameters.inputs.order = order
framewise_disp = pe.MapNode(FramewiseDisplacement(parameter_source='FSL'),
iterfield=['in_file'],
name='framewise_disp')
mcf_wf = pe.Workflow(name=name)
mcf_wf.connect(input_node, 'output_directory', datasink, 'base_directory')
mcf_wf.connect(input_node, 'sub_id', datasink, 'container')
mcf_wf.connect(input_node, 'par_file', extend_motion_parameters,
'par_file')
mcf_wf.connect(input_node, 'par_file', framewise_disp, 'in_file')
mcf_wf.connect(extend_motion_parameters, 'out_ext', output_node,
'out_ext_moco')
mcf_wf.connect(framewise_disp, 'out_file', output_node, 'out_fd')
mcf_wf.connect(extend_motion_parameters, 'out_ext', datasink, 'confounds')
mcf_wf.connect(framewise_disp, 'out_file', datasink, 'confounds.@df')
return mcf_wf
def hmc(name='fMRI_HMC'):
"""
Create a :abbr:`HMC (head motion correction)` workflow for fMRI.
.. workflow::
from mriqc.workflows.functional import hmc
from mriqc.testing import mock_config
with mock_config():
wf = hmc()
"""
from nipype.algorithms.confounds import FramewiseDisplacement
from nipype.interfaces.afni import Calc, TShift, Refit, Despike, Volreg
from niworkflows.interfaces.registration import EstimateReferenceImage
mem_gb = config.workflow.biggest_file_gb
workflow = pe.Workflow(name=name)
inputnode = pe.Node(niu.IdentityInterface(
fields=['in_file', 'fd_radius', 'start_idx', 'stop_idx']),
name='inputnode')
outputnode = pe.Node(niu.IdentityInterface(fields=['out_file', 'out_fd']),
name='outputnode')
if any((config.workflow.start_idx is not None, config.workflow.stop_idx
is not None)):
drop_trs = pe.Node(Calc(expr='a', outputtype='NIFTI_GZ'),
name='drop_trs')
workflow.connect([
(inputnode, drop_trs, [('in_file', 'in_file_a'),
('start_idx', 'start_idx'),
('stop_idx', 'stop_idx')]),
])
else:
drop_trs = pe.Node(niu.IdentityInterface(fields=['out_file']),
name='drop_trs')
workflow.connect([
(inputnode, drop_trs, [('in_file', 'out_file')]),
])
gen_ref = pe.Node(EstimateReferenceImage(mc_method="AFNI"), name="gen_ref")
# calculate hmc parameters
hmc = pe.Node(Volreg(args='-Fourier -twopass',
zpad=4,
outputtype='NIFTI_GZ'),
name='motion_correct',
mem_gb=mem_gb * 2.5)
# Compute the frame-wise displacement
fdnode = pe.Node(FramewiseDisplacement(normalize=False,
parameter_source="AFNI"),
name='ComputeFD')
workflow.connect([
(inputnode, fdnode, [('fd_radius', 'radius')]),
(gen_ref, hmc, [('ref_image', 'basefile')]),
(hmc, outputnode, [('out_file', 'out_file')]),
(hmc, fdnode, [('oned_file', 'in_file')]),
(fdnode, outputnode, [('out_file', 'out_fd')]),
])
# Slice timing correction, despiking, and deoblique
st_corr = pe.Node(TShift(outputtype='NIFTI_GZ'), name='TimeShifts')
deoblique_node = pe.Node(Refit(deoblique=True), name='deoblique')
despike_node = pe.Node(Despike(outputtype='NIFTI_GZ'), name='despike')
if all((config.workflow.correct_slice_timing, config.workflow.despike,
config.workflow.deoblique)):
workflow.connect([
(drop_trs, st_corr, [('out_file', 'in_file')]),
(st_corr, despike_node, [('out_file', 'in_file')]),
(despike_node, deoblique_node, [('out_file', 'in_file')]),
(deoblique_node, gen_ref, [('out_file', 'in_file')]),
(deoblique_node, hmc, [('out_file', 'in_file')]),
])
elif config.workflow.correct_slice_timing and config.workflow.despike:
workflow.connect([
(drop_trs, st_corr, [('out_file', 'in_file')]),
(st_corr, despike_node, [('out_file', 'in_file')]),
(despike_node, gen_ref, [('out_file', 'in_file')]),
(despike_node, hmc, [('out_file', 'in_file')]),
])
elif config.workflow.correct_slice_timing and config.workflow.deoblique:
workflow.connect([
(drop_trs, st_corr, [('out_file', 'in_file')]),
(st_corr, deoblique_node, [('out_file', 'in_file')]),
(deoblique_node, gen_ref, [('out_file', 'in_file')]),
(deoblique_node, hmc, [('out_file', 'in_file')]),
])
elif config.workflow.correct_slice_timing:
workflow.connect([
(drop_trs, st_corr, [('out_file', 'in_file')]),
(st_corr, gen_ref, [('out_file', 'in_file')]),
(st_corr, hmc, [('out_file', 'in_file')]),
])
elif config.workflow.despike and config.workflow.deoblique:
workflow.connect([
(drop_trs, despike_node, [('out_file', 'in_file')]),
(despike_node, deoblique_node, [('out_file', 'in_file')]),
(deoblique_node, gen_ref, [('out_file', 'in_file')]),
(deoblique_node, hmc, [('out_file', 'in_file')]),
])
elif config.workflow.despike:
workflow.connect([
(drop_trs, despike_node, [('out_file', 'in_file')]),
(despike_node, gen_ref, [('out_file', 'in_file')]),
(despike_node, hmc, [('out_file', 'in_file')]),
])
elif config.workflow.deoblique:
workflow.connect([
(drop_trs, deoblique_node, [('out_file', 'in_file')]),
(deoblique_node, gen_ref, [('out_file', 'in_file')]),
(deoblique_node, hmc, [('out_file', 'in_file')]),
])
else:
workflow.connect([
(drop_trs, gen_ref, [('out_file', 'in_file')]),
(drop_trs, hmc, [('out_file', 'in_file')]),
])
return workflow
def hmc(name="fMRI_HMC"):
"""
Create a :abbr:`HMC (head motion correction)` workflow for fMRI.
.. workflow::
from mriqc.workflows.functional import hmc
from mriqc.testing import mock_config
with mock_config():
wf = hmc()
"""
from nipype.algorithms.confounds import FramewiseDisplacement
from nipype.interfaces.afni import Calc, Despike, Refit, TShift, Volreg
mem_gb = config.workflow.biggest_file_gb
workflow = pe.Workflow(name=name)
inputnode = pe.Node(
niu.IdentityInterface(
fields=["in_file", "fd_radius", "start_idx", "stop_idx"]),
name="inputnode",
)
outputnode = pe.Node(niu.IdentityInterface(fields=["out_file", "out_fd"]),
name="outputnode")
if any((
config.workflow.start_idx is not None,
config.workflow.stop_idx is not None,
)):
drop_trs = pe.Node(Calc(expr="a", outputtype="NIFTI_GZ"),
name="drop_trs")
# fmt: off
workflow.connect([
(inputnode, drop_trs, [("in_file", "in_file_a"),
("start_idx", "start_idx"),
("stop_idx", "stop_idx")]),
])
# fmt: on
else:
drop_trs = pe.Node(niu.IdentityInterface(fields=["out_file"]),
name="drop_trs")
# fmt: off
workflow.connect([
(inputnode, drop_trs, [("in_file", "out_file")]),
])
# fmt: on
# calculate hmc parameters
hmc = pe.Node(
Volreg(args="-Fourier -twopass", zpad=4, outputtype="NIFTI_GZ"),
name="motion_correct",
mem_gb=mem_gb * 2.5,
)
# Compute the frame-wise displacement
fdnode = pe.Node(
FramewiseDisplacement(normalize=False, parameter_source="AFNI"),
name="ComputeFD",
)
# fmt: off
workflow.connect([
(inputnode, fdnode, [("fd_radius", "radius")]),
(hmc, outputnode, [("out_file", "out_file")]),
(hmc, fdnode, [("oned_file", "in_file")]),
(fdnode, outputnode, [("out_file", "out_fd")]),
])
# fmt: on
# Slice timing correction, despiking, and deoblique
st_corr = pe.Node(TShift(outputtype="NIFTI_GZ"), name="TimeShifts")
deoblique_node = pe.Node(Refit(deoblique=True), name="deoblique")
despike_node = pe.Node(Despike(outputtype="NIFTI_GZ"), name="despike")
if all((
config.workflow.correct_slice_timing,
config.workflow.despike,
config.workflow.deoblique,
)):
# fmt: off
workflow.connect([
(drop_trs, st_corr, [("out_file", "in_file")]),
(st_corr, despike_node, [("out_file", "in_file")]),
(despike_node, deoblique_node, [("out_file", "in_file")]),
(deoblique_node, hmc, [("out_file", "in_file")]),
])
# fmt: on
elif config.workflow.correct_slice_timing and config.workflow.despike:
# fmt: off
workflow.connect([
(drop_trs, st_corr, [("out_file", "in_file")]),
(st_corr, despike_node, [("out_file", "in_file")]),
(despike_node, hmc, [("out_file", "in_file")]),
])
# fmt: on
elif config.workflow.correct_slice_timing and config.workflow.deoblique:
# fmt: off
workflow.connect([
(drop_trs, st_corr, [("out_file", "in_file")]),
(st_corr, deoblique_node, [("out_file", "in_file")]),
(deoblique_node, hmc, [("out_file", "in_file")]),
])
# fmt: on
elif config.workflow.correct_slice_timing:
# fmt: off
workflow.connect([
(drop_trs, st_corr, [("out_file", "in_file")]),
(st_corr, hmc, [("out_file", "in_file")]),
])
# fmt: on
elif config.workflow.despike and config.workflow.deoblique:
# fmt: off
workflow.connect([
(drop_trs, despike_node, [("out_file", "in_file")]),
(despike_node, deoblique_node, [("out_file", "in_file")]),
(deoblique_node, hmc, [("out_file", "in_file")]),
])
# fmt: on
elif config.workflow.despike:
# fmt: off
workflow.connect([
(drop_trs, despike_node, [("out_file", "in_file")]),
(despike_node, hmc, [("out_file", "in_file")]),
])
# fmt: on
elif config.workflow.deoblique:
# fmt: off
workflow.connect([
(drop_trs, deoblique_node, [("out_file", "in_file")]),
(deoblique_node, hmc, [("out_file", "in_file")]),
])
# fmt: on
else:
# fmt: off
workflow.connect([
(drop_trs, hmc, [("out_file", "in_file")]),
])
# fmt: on
return workflow
def create_rs_qc(subjectlist):
# main workflow for extended qc of diffusion/rsfmri data
# fsl output type
fsl.FSLCommand.set_default_output_type('NIFTI_GZ')
# some hard coded things
fd_thres = 0.2
tr = 2
# Specify the location of the preprocessed data
data_dir = "/data/pt_life/LIFE_fu/wd_preprocessing/hcp_prep_workflow/resting/"
working_dir = "/data/pt_life/LIFE_fu/wd_preprocessing/" #MODIFY
freesurfer_dir = "/data/pt_life_freesurfer/freesurfer_all"
qc = Workflow(name="qc")
qc.base_dir = working_dir + '/'
qc.config['execution']['crashdump_dir'] = qc.base_dir + "/crash_files"
qc.config['execution'] = {'hash_method': 'content'}
#first get all data needed
identitynode = Node(util.IdentityInterface(fields=['subject']),
name='identitynode')
identitynode.iterables = ('subject', subjectlist)
info = dict(func=[[
'transform_timeseries/', '_subject_', 'subj', '/merge/rest2anat.nii.gz'
]],
dvars=[[
'transform_timeseries/', '_subject_', 'subj',
'/dvars/rest2anat_dvars.tsv'
]],
motpars=[[
'/motion_correction/', '_subject_', 'subj',
'/mcflirt/rest_realigned.nii.gz.par'
]],
brainmask=[[
'transform_timeseries/', '_subject_', 'subj',
'/resample_brain/T1_brain_mask_lowres.nii.gz'
]])
ds_rs = Node(interface=nio.DataGrabber(
infields=['subj'], outfields=['func', 'dvars', 'motpars',
'brainmask']),
name='ds_rs')
ds_rs.inputs.base_directory = data_dir
ds_rs.inputs.template = '%s%s%s%s'
ds_rs.inputs.template_args = info
ds_rs.inputs.sort_filelist = True
def juggle_subj(input_id):
import pandas as pd
from datetime import datetime as dt
import os
import random, string
sic_pseudo = pd.read_csv(
"/data/gh_gr_agingandobesity_share/life_shared/Data/Preprocessed/derivatives/pseudo_mrt_20201214.csv"
)
tmp = sic_pseudo.loc[sic_pseudo.sic == input_id, 'pseudonym']
pseudo = tmp.get_values()[0] + "_fu"
return pseudo
rename = Node(util.Function(input_names=['input_id'],
output_names=['output_id'],
function=juggle_subj),
name="rename")
get_fs = Node(nio.FreeSurferSource(), name="get_fs")
get_fs.inputs.subjects_dir = freesurfer_dir
get_correct_aseg = Node(util.Function(input_names=['in_list'],
output_names=['out_aseg'],
function=get_aseg),
name="get_correct_aseg")
convert = Node(fs.MRIConvert(), name="convert")
convert.inputs.out_type = "niigz"
downsample = Node(afni.Resample(resample_mode='NN',
outputtype='NIFTI_GZ',
out_file='aparcaseg_lowres.nii.gz'),
name='downsample')
calc_fd_official = Node(FramewiseDisplacement(parameter_source='FSL'),
name='calc_fd_official')
calc_fd = Node(util.Function(
input_names=['realignment_parameters_file', 'parameter_source'],
output_names=['FD_power', 'fn'],
function=calc_frame_displacement),
name="calc_fd")
calc_fd.inputs.parameter_source = 'FSL'
outliers = Node(afni.OutlierCount(fraction=True, out_file='outliers.out'),
name='outliers',
mem_gb=1 * 2.5)
bigplot = Node(util.Function(input_names=[
'func', 'seg', 'tr', 'fd_thres', 'outliers', 'dvars', 'fd', 'subj',
'outfile'
],
output_names=['fn', 'dataframe'],
function=make_the_plot),
name="bigplot")
bigplot.inputs.tr = tr
bigplot.inputs.fd_thres = fd_thres
bigplot.inputs.outfile = "summary_fmriplot.png"
fftplot = Node(util.Function(input_names=['fn_pd', 'tr'],
output_names=['fn'],
function=plot_fft),
name="fftplot")
fftplot.inputs.tr = tr
datasink = Node(name="datasink", interface=nio.DataSink())
datasink.inputs.base_directory = "/data/pt_life_restingstate_followup/Results/QA"
datasink.inputs.substitutions = [('_subject_', '')]
qc.connect([
(identitynode, rename, [('subject', 'input_id')]),
(rename, get_fs, [('output_id', 'subject_id')]),
(identitynode, ds_rs, [('subject', 'subj')]),
(identitynode, bigplot, [('subject', 'subj')]),
(get_fs, get_correct_aseg, [('aparc_aseg', 'in_list')]),
(get_correct_aseg, convert, [('out_aseg', 'in_file')]),
(convert, downsample, [('out_file', 'in_file')]),
(ds_rs, downsample, [('func', 'master')]),
(downsample, bigplot, [('out_file', 'seg')]),
(ds_rs, calc_fd, [('motpars', 'realignment_parameters_file')]),
(ds_rs, calc_fd_official, [('motpars', 'in_file')]),
(ds_rs, bigplot, [('func', 'func')]),
(ds_rs, bigplot, [('dvars', 'dvars')]),
(calc_fd, bigplot, [('fn', 'fd')]), #FD_power
(ds_rs, outliers, [('func', 'in_file')]),
(ds_rs, outliers, [('brainmask', 'mask')]),
(outliers, bigplot, [('out_file', 'outliers')]),
(bigplot, datasink, [('fn', 'detailedQA.@bigplot')]),
(bigplot, fftplot, [('dataframe', 'fn_pd')]),
(bigplot, datasink, [('dataframe', 'detailedQA.metrics.@dataframe')]),
(fftplot, datasink, [('fn', 'detailedQA.@fftplot')]),
(calc_fd, datasink, [('fn', 'detailedQA.metrics.@fd')]),
(calc_fd_official, datasink, [('out_file',
'detailedQA.metrics.@fd_official')])
])
qc.run(plugin="MultiProc", plugin_args={"n_procs": 16, "non_daemon": True})
return qc
