def apply_ants_warp(self, runtime, in_file, out_file, rigid):
out_rigid = op.basename(add_suffix(out_file, "anat"))
continuous_interp = dict(trilinear="trilin",
spline="cubic")[self.interp]
interp = "nearest" if "mask" in in_file else continuous_interp
cmdline_rigid = ["mri_vol2vol",
"--mov", in_file,
"--reg", rigid,
"--fstarg",
"--" + interp,
"--o", out_rigid,
"--no-save-reg"]
runtime = submit_cmdline(runtime, cmdline_rigid)
continuous_interp = dict(trilinear="trilin",
spline="BSpline")[self.interp]
interp = "NN" if "mask" in in_file else continuous_interp
cmdline_warp = ["WarpImageMultiTransform",
"3",
out_rigid,
out_file,
self.inputs.warpfield,
self.inputs.affine,
"-R", self.ref_file]
if interp != "trilin":
cmdline_warp.append("--use-" + interp)
runtime = submit_cmdline(runtime, cmdline_warp)
return runtime
def apply_fsl_rigid(self, runtime, in_file, out_file, rigid):
interp = "nn" if "mask" in in_file else self.interp
cmdline = [
"applywarp", "-i", in_file, "-o", out_file, "-r", self.ref_file,
"--interp={}".format(interp), "--premat={}".format(rigid)
]
runtime = submit_cmdline(runtime, cmdline)
return runtime
def apply_fsl_rigid(self, runtime, in_file, out_file, rigid):
interp = "nn" if "mask" in in_file else self.interp
cmdline = ["applywarp",
"-i", in_file,
"-o", out_file,
"-r", self.ref_file,
"--interp={}".format(interp),
"--premat={}".format(rigid)]
runtime = submit_cmdline(runtime, cmdline)
return runtime
def combine_rigids(self, runtime, first_rigid, second_rigid):
"""Invert the first rigid and combine with the second.
This creates a transformation from run n to run 1 space,
through the anatomical coregistration.
"""
first_inv = op.basename(add_suffix(first_rigid, "inv"))
cmdline_inv = [
"convert_xfm", "-omat", first_inv, "-inverse", first_rigid
]
runtime = submit_cmdline(runtime, cmdline_inv)
full_rigid = op.basename(add_suffix(second_rigid, "to_epi"))
cmdline_full = [
"convert_xfm", "-omat", full_rigid, "-concat", first_inv,
second_rigid
]
runtime = submit_cmdline(runtime, cmdline_full)
return runtime, full_rigid
def combine_rigids(self, runtime, first_rigid, second_rigid):
"""Invert the first rigid and combine with the second.
This creates a transformation from run n to run 1 space,
through the anatomical coregistration.
"""
first_inv = op.basename(add_suffix(first_rigid, "inv"))
cmdline_inv = ["convert_xfm",
"-omat", first_inv,
"-inverse",
first_rigid]
runtime = submit_cmdline(runtime, cmdline_inv)
full_rigid = op.basename(add_suffix(second_rigid, "to_epi"))
cmdline_full = ["convert_xfm",
"-omat", full_rigid,
"-concat", first_inv,
second_rigid]
runtime = submit_cmdline(runtime, cmdline_full)
return runtime, full_rigid
def apply_ants_warp(self, runtime, in_file, out_file, rigid):
out_rigid = op.basename(add_suffix(out_file, "anat"))
continuous_interp = dict(trilinear="trilin",
spline="cubic")[self.interp]
interp = "nearest" if "mask" in in_file else continuous_interp
cmdline_rigid = [
"mri_vol2vol", "--mov", in_file, "--reg", rigid, "--fstarg",
"--" + interp, "--o", out_rigid, "--no-save-reg"
]
runtime = submit_cmdline(runtime, cmdline_rigid)
continuous_interp = dict(trilinear="trilin",
spline="BSpline")[self.interp]
interp = "NN" if "mask" in in_file else continuous_interp
cmdline_warp = [
"WarpImageMultiTransform", "3", out_rigid, out_file,
self.inputs.warpfield, self.inputs.affine, "-R", self.ref_file
]
if interp != "trilin":
cmdline_warp.append("--use-" + interp)
runtime = submit_cmdline(runtime, cmdline_warp)
return runtime
def _run_interface(self, runtime):
target_brain = fsl.Info.standard_image("avg152T1_brain.nii.gz")
cmdline = " ".join([
"antsIntroduction.sh", "-d", "3", "-r", target_brain, "-i",
self.inputs.in_file, "-o", "ants_"
])
runtime.environ["ITK_NUM_THREADS"] = "1"
runtime = submit_cmdline(runtime, cmdline)
os.rename("ants_Affine.txt", "affine.txt")
os.rename("ants_Warp.nii.gz", "warpfield.nii.gz")
os.rename("ants_deformed.nii.gz", "brain_warp.nii.gz")
os.rename("ants_InverseWarp.nii.gz", "inverse_warpfield.nii.gz")
return runtime
def _run_interface(self, runtime):
target_brain = fsl.Info.standard_image("avg152T1_brain.nii.gz")
cmdline = " ".join(["antsIntroduction.sh",
"-d", "3",
"-r", target_brain,
"-i", self.inputs.in_file,
"-o", "ants_"])
runtime.environ["ITK_NUM_THREADS"] = "1"
runtime = submit_cmdline(runtime, cmdline)
os.rename("ants_Affine.txt", "affine.txt")
os.rename("ants_Warp.nii.gz", "warpfield.nii.gz")
os.rename("ants_deformed.nii.gz", "brain_warp.nii.gz")
os.rename("ants_InverseWarp.nii.gz", "inverse_warpfield.nii.gz")
return runtime
def _run_interface(self, runtime):
n_con = len(self.inputs.contrasts)
# Find the basic geometry of the image
img = nib.load(self.inputs.copes[0])
x, y, z = img.shape
aff, hdr = img.get_affine(), img.get_header()
# Get lists of files for each contrast
copes = self._unpack_files(self.inputs.copes, n_con)
varcopes = self._unpack_files(self.inputs.varcopes, n_con)
# Make an image with the DOF for each run
dofs = np.array([np.loadtxt(f) for f in self.inputs.dofs])
dof_data = np.ones((x, y, z, len(dofs))) * dofs
# Find the intersection of the masks
mask_data = [nib.load(f).get_data() for f in self.inputs.masks]
common_mask = np.all(mask_data, axis=0)
nib.Nifti1Image(common_mask, aff, hdr).to_filename("mask.nii.gz")
# Run the flame models
flame_results = []
zstat_files = []
for i, contrast in enumerate(self.inputs.contrasts):
# Load each run of cope and varcope files into a list
cs = [nib.load(f).get_data()[..., np.newaxis] for f in copes[i]]
vs = [nib.load(f).get_data()[..., np.newaxis] for f in varcopes[i]]
# Find all of the nonzero copes
# This handles cases where there were no events for some of
# the runs for the contrast we're currently dealing with
good_cs = [not np.allclose(d, 0) for d in cs]
good_vs = [not np.allclose(d, 0) for d in vs]
good = np.all([good_cs, good_vs], axis=0)
# Handle the case where no events occured for this contrast
if not good.any():
good = np.ones(len(cs), bool)
# Concatenate the cope and varcope data, save only the good frames
c_data = np.concatenate(cs, axis=-1)[:, :, :, good]
v_data = np.concatenate(vs, axis=-1)[:, :, :, good]
# Write out the concatenated copes and varcopes
nib.Nifti1Image(c_data, aff, hdr).to_filename("cope_4d.nii.gz")
nib.Nifti1Image(v_data, aff, hdr).to_filename("varcope_4d.nii.gz")
# Write out a correctly sized design for this contrast
fsl.L2Model(num_copes=int(good.sum())).run()
# Mask the DOF data and write it out for this run
contrast_dof = dof_data[:, :, :, good]
nib.Nifti1Image(contrast_dof, aff, hdr).to_filename("dof.nii.gz")
# Build the flamo commandline and run
flamecmd = ["flameo",
"--cope=cope_4d.nii.gz",
"--varcope=varcope_4d.nii.gz",
"--mask=mask.nii.gz",
"--dvc=dof.nii.gz",
"--runmode=fe",
"--dm=design.mat",
"--tc=design.con",
"--cs=design.grp",
"--ld=" + contrast,
"--npo"]
runtime = submit_cmdline(runtime, flamecmd)
# Rename the written file and append to the outputs
for kind in ["cope", "varcope"]:
os.rename(kind + "_4d.nii.gz",
"%s/%s_4d.nii.gz" % (contrast, kind))
# Put the zstats and mask on the surface
for hemi in ["lh", "rh"]:
projcmd = ["mri_vol2surf",
"--mov", "%s/zstat1.nii.gz" % contrast,
"--reg", self.inputs.reg_file,
"--hemi", hemi,
"--projfrac-avg", "0", "1", ".1",
"--o", "%s/%s.zstat1.mgz" % (contrast, hemi)]
submit_cmdline(runtime, projcmd)
# Mask image
projcmd = ["mri_vol2surf",
"--mov", "%s/mask.nii.gz" % contrast,
"--reg", self.inputs.reg_file,
"--hemi", hemi,
"--projfrac-max", "0", "1", ".1",
"--o", "%s/%s.mask.mgz" % (contrast, hemi)]
submit_cmdline(runtime, projcmd)
flame_results.append(op.abspath(contrast))
zstat_files.append(op.abspath("%s/zstat1.nii.gz" % contrast))
self.flame_results = flame_results
self.zstat_files = zstat_files
return runtime
def _run_interface(self, runtime):
n_con = len(self.inputs.contrasts)
# Find the basic geometry of the image
img = nib.load(self.inputs.copes[0])
x, y, z = img.shape
aff, hdr = img.get_affine(), img.get_header()
# Get lists of files for each contrast
copes = self._unpack_files(self.inputs.copes, n_con)
varcopes = self._unpack_files(self.inputs.varcopes, n_con)
# Make an image with the DOF for each run
dofs = np.array([np.loadtxt(f) for f in self.inputs.dofs])
dof_data = np.ones((x, y, z, len(dofs))) * dofs
# Find the intersection of the masks
mask_data = [nib.load(f).get_data() for f in self.inputs.masks]
common_mask = np.all(mask_data, axis=0)
nib.Nifti1Image(common_mask, aff, hdr).to_filename("mask.nii.gz")
# Run the flame models
flame_results = []
zstat_files = []
for i, contrast in enumerate(self.inputs.contrasts):
# Load each run of cope and varcope files into a list
cs = [nib.load(f).get_data()[..., np.newaxis] for f in copes[i]]
vs = [nib.load(f).get_data()[..., np.newaxis] for f in varcopes[i]]
# Find all of the nonzero copes
# This handles cases where there were no events for some of
# the runs for the contrast we're currently dealing with
good_cs = [not np.allclose(d, 0) for d in cs]
good_vs = [not np.allclose(d, 0) for d in vs]
good = np.all([good_cs, good_vs], axis=0)
# Concatenate the cope and varcope data, save only the good frames
c_data = np.concatenate(cs, axis=-1)[:, :, :, good]
v_data = np.concatenate(vs, axis=-1)[:, :, :, good]
# Write out the concatenated copes and varcopes
nib.Nifti1Image(c_data, aff, hdr).to_filename("cope_4d.nii.gz")
nib.Nifti1Image(v_data, aff, hdr).to_filename("varcope_4d.nii.gz")
# Write out a correctly sized design for this contrast
fsl.L2Model(num_copes=int(good.sum())).run()
# Mask the DOF data and write it out for this run
contrast_dof = dof_data[:, :, :, good]
nib.Nifti1Image(contrast_dof, aff, hdr).to_filename("dof.nii.gz")
# Build the flamo commandline and run
flamecmd = ["flameo",
"--cope=cope_4d.nii.gz",
"--varcope=varcope_4d.nii.gz",
"--mask=mask.nii.gz",
"--dvc=dof.nii.gz",
"--runmode=fe",
"--dm=design.mat",
"--tc=design.con",
"--cs=design.grp",
"--ld=" + contrast,
"--npo"]
runtime = submit_cmdline(runtime, flamecmd)
# Rename the written file and append to the outputs
for kind in ["cope", "varcope"]:
os.rename(kind + "_4d.nii.gz",
"%s/%s_4d.nii.gz" % (contrast, kind))
# Put the zstats and mask on the surface
for hemi in ["lh", "rh"]:
projcmd = ["mri_vol2surf",
"--mov", "%s/zstat1.nii.gz" % contrast,
"--reg", self.inputs.reg_file,
"--surf-fwhm", "5",
"--hemi", hemi,
"--projfrac-avg", "0", "1", ".1",
"--o", "%s/%s.zstat1.mgz" % (contrast, hemi)]
submit_cmdline(runtime, projcmd)
# Mask image
projcmd = ["mri_vol2surf",
"--mov", "%s/mask.nii.gz" % contrast,
"--reg", self.inputs.reg_file,
"--hemi", hemi,
"--projfrac-max", "0", "1", ".1",
"--o", "%s/%s.mask.mgz" % (contrast, hemi)]
submit_cmdline(runtime, projcmd)
flame_results.append(op.abspath(contrast))
zstat_files.append(op.abspath("%s/zstat1.nii.gz" % contrast))
self.flame_results = flame_results
self.zstat_files = zstat_files
return runtime
