def execute(): #pylint: disable=unused-variable
lmax_option = ''
if app.ARGS.lmax:
lmax_option = ' -lmax ' + app.ARGS.lmax
if app.ARGS.max_iters < 2:
raise MRtrixError('Number of iterations must be at least 2')
progress = app.ProgressBar('Optimising')
iter_voxels = app.ARGS.iter_voxels
if iter_voxels == 0:
iter_voxels = 10 * app.ARGS.number
elif iter_voxels < app.ARGS.number:
raise MRtrixError(
'Number of selected voxels (-iter_voxels) must be greater than number of voxels desired (-number)'
)
iteration = 0
while iteration < app.ARGS.max_iters:
prefix = 'iter' + str(iteration) + '_'
if iteration == 0:
rf_in_path = 'init_RF.txt'
mask_in_path = 'mask.mif'
init_rf = '1 -1 1'
with open(rf_in_path, 'w') as init_rf_file:
init_rf_file.write(init_rf)
iter_lmax_option = ' -lmax 4'
else:
rf_in_path = 'iter' + str(iteration - 1) + '_RF.txt'
mask_in_path = 'iter' + str(iteration - 1) + '_SF_dilated.mif'
iter_lmax_option = lmax_option
# Run CSD
run.command('dwi2fod csd dwi.mif ' + rf_in_path + ' ' + prefix +
'FOD.mif -mask ' + mask_in_path)
# Get amplitudes of two largest peaks, and direction of largest
run.command('fod2fixel ' + prefix + 'FOD.mif ' + prefix +
'fixel -peak peaks.mif -mask ' + mask_in_path +
' -fmls_no_thresholds')
app.cleanup(prefix + 'FOD.mif')
if iteration:
app.cleanup(mask_in_path)
run.command('fixel2voxel ' + prefix + 'fixel/peaks.mif none ' +
prefix + 'amps.mif -number 2')
run.command('mrconvert ' + prefix + 'amps.mif ' + prefix +
'first_peaks.mif -coord 3 0 -axes 0,1,2')
run.command('mrconvert ' + prefix + 'amps.mif ' + prefix +
'second_peaks.mif -coord 3 1 -axes 0,1,2')
app.cleanup(prefix + 'amps.mif')
run.command('fixel2peaks ' + prefix + 'fixel/directions.mif ' +
prefix + 'first_dir.mif -number 1')
app.cleanup(prefix + 'fixel')
# Calculate the 'cost function' Donald derived for selecting single-fibre voxels
# https://github.com/MRtrix3/mrtrix3/pull/426
# sqrt(|peak1|) * (1 - |peak2| / |peak1|)^2
run.command('mrcalc ' + prefix + 'first_peaks.mif -sqrt 1 ' + prefix +
'second_peaks.mif ' + prefix +
'first_peaks.mif -div -sub 2 -pow -mult ' + prefix +
'CF.mif')
app.cleanup(prefix + 'first_peaks.mif')
app.cleanup(prefix + 'second_peaks.mif')
voxel_count = image.statistics(prefix + 'CF.mif').count
# Select the top-ranked voxels
run.command('mrthreshold ' + prefix + 'CF.mif -top ' +
str(min([app.ARGS.number, voxel_count])) + ' ' + prefix +
'SF.mif')
# Generate a new response function based on this selection
run.command('amp2response dwi.mif ' + prefix + 'SF.mif ' + prefix +
'first_dir.mif ' + prefix + 'RF.txt' + iter_lmax_option)
app.cleanup(prefix + 'first_dir.mif')
new_rf = matrix.load_vector(prefix + 'RF.txt')
progress.increment('Optimising (' + str(iteration + 1) +
' iterations, RF: [ ' + ', '.join('{:.3f}'.format(n)
for n in new_rf) +
'] )')
# Should we terminate?
if iteration > 0:
run.command('mrcalc ' + prefix + 'SF.mif iter' +
str(iteration - 1) + '_SF.mif -sub ' + prefix +
'SF_diff.mif')
app.cleanup('iter' + str(iteration - 1) + '_SF.mif')
max_diff = image.statistics(prefix + 'SF_diff.mif').max
app.cleanup(prefix + 'SF_diff.mif')
if not max_diff:
app.cleanup(prefix + 'CF.mif')
run.function(shutil.copyfile, prefix + 'RF.txt',
'response.txt')
run.function(shutil.move, prefix + 'SF.mif', 'voxels.mif')
break
# Select a greater number of top single-fibre voxels, and dilate (within bounds of initial mask);
# these are the voxels that will be re-tested in the next iteration
run.command('mrthreshold ' + prefix + 'CF.mif -top ' +
str(min([iter_voxels, voxel_count])) +
' - | maskfilter - dilate - -npass ' +
str(app.ARGS.dilate) + ' | mrcalc mask.mif - -mult ' +
prefix + 'SF_dilated.mif')
app.cleanup(prefix + 'CF.mif')
iteration += 1
progress.done()
# If terminating due to running out of iterations, still need to put the results in the appropriate location
if os.path.exists('response.txt'):
app.console(
'Convergence of SF voxel selection detected at iteration ' +
str(iteration + 1))
else:
app.console('Exiting after maximum ' + str(app.ARGS.max_iters) +
' iterations')
run.function(shutil.copyfile,
'iter' + str(app.ARGS.max_iters - 1) + '_RF.txt',
'response.txt')
run.function(shutil.move,
'iter' + str(app.ARGS.max_iters - 1) + '_SF.mif',
'voxels.mif')
run.function(shutil.copyfile, 'response.txt',
path.from_user(app.ARGS.output, False))
if app.ARGS.voxels:
run.command('mrconvert voxels.mif ' + path.from_user(app.ARGS.voxels),
mrconvert_keyval=path.from_user(app.ARGS.input, False),
force=app.FORCE_OVERWRITE)
def execute(): #pylint: disable=unused-variable
import math, os, shutil
from mrtrix3 import app, image, matrix, MRtrixError, path, run
lmax_option = ''
if app.ARGS.lmax:
lmax_option = ' -lmax ' + app.ARGS.lmax
convergence_change = 0.01 * app.ARGS.convergence
progress = app.ProgressBar('Optimising')
iteration = 0
while iteration < app.ARGS.max_iters:
prefix = 'iter' + str(iteration) + '_'
# How to initialise response function?
# old dwi2response command used mean & standard deviation of DWI data; however
# this may force the output FODs to lmax=2 at the first iteration
# Chantal used a tensor with low FA, but it'd be preferable to get the scaling right
# Other option is to do as before, but get the ratio between l=0 and l=2, and
# generate l=4,6,... using that amplitude ratio
if iteration == 0:
rf_in_path = 'init_RF.txt'
mask_in_path = 'mask.mif'
# Grab the mean and standard deviation across all volumes in a single mrstats call
# Also scale them to reflect the fact that we're moving to the SH basis
mean = image.statistic('dwi.mif', 'mean',
'-mask mask.mif -allvolumes') * math.sqrt(
4.0 * math.pi)
std = image.statistic('dwi.mif', 'std',
'-mask mask.mif -allvolumes') * math.sqrt(
4.0 * math.pi)
# Now produce the initial response function
# Let's only do it to lmax 4
init_rf = [
str(mean),
str(-0.5 * std),
str(0.25 * std * std / mean)
]
with open('init_RF.txt', 'w') as init_rf_file:
init_rf_file.write(' '.join(init_rf))
else:
rf_in_path = 'iter' + str(iteration - 1) + '_RF.txt'
mask_in_path = 'iter' + str(iteration - 1) + '_SF.mif'
# Run CSD
run.command('dwi2fod csd dwi.mif ' + rf_in_path + ' ' + prefix +
'FOD.mif -mask ' + mask_in_path)
# Get amplitudes of two largest peaks, and directions of largest
run.command('fod2fixel ' + prefix + 'FOD.mif ' + prefix +
'fixel -peak peaks.mif -mask ' + mask_in_path +
' -fmls_no_thresholds')
app.cleanup(prefix + 'FOD.mif')
run.command('fixel2voxel ' + prefix + 'fixel/peaks.mif split_data ' +
prefix + 'amps.mif')
run.command('mrconvert ' + prefix + 'amps.mif ' + prefix +
'first_peaks.mif -coord 3 0 -axes 0,1,2')
run.command('mrconvert ' + prefix + 'amps.mif ' + prefix +
'second_peaks.mif -coord 3 1 -axes 0,1,2')
app.cleanup(prefix + 'amps.mif')
run.command('fixel2voxel ' + prefix +
'fixel/directions.mif split_dir ' + prefix +
'all_dirs.mif')
app.cleanup(prefix + 'fixel')
run.command('mrconvert ' + prefix + 'all_dirs.mif ' + prefix +
'first_dir.mif -coord 3 0:2')
app.cleanup(prefix + 'all_dirs.mif')
# Revise single-fibre voxel selection based on ratio of tallest to second-tallest peak
run.command('mrcalc ' + prefix + 'second_peaks.mif ' + prefix +
'first_peaks.mif -div ' + prefix + 'peak_ratio.mif')
app.cleanup(prefix + 'first_peaks.mif')
app.cleanup(prefix + 'second_peaks.mif')
run.command('mrcalc ' + prefix + 'peak_ratio.mif ' +
str(app.ARGS.peak_ratio) + ' -lt ' + mask_in_path +
' -mult ' + prefix + 'SF.mif -datatype bit')
app.cleanup(prefix + 'peak_ratio.mif')
# Make sure image isn't empty
sf_voxel_count = image.statistic(prefix + 'SF.mif', 'count',
'-mask ' + prefix + 'SF.mif')
if not sf_voxel_count:
raise MRtrixError(
'Aborting: All voxels have been excluded from single-fibre selection'
)
# Generate a new response function
run.command('amp2response dwi.mif ' + prefix + 'SF.mif ' + prefix +
'first_dir.mif ' + prefix + 'RF.txt' + lmax_option)
app.cleanup(prefix + 'first_dir.mif')
new_rf = matrix.load_vector(prefix + 'RF.txt')
progress.increment('Optimising (' + str(iteration + 1) +
' iterations, ' + str(sf_voxel_count) +
' voxels, RF: [ ' + ', '.join('{:.3f}'.format(n)
for n in new_rf) +
'] )')
# Detect convergence
# Look for a change > some percentage - don't bother looking at the masks
if iteration > 0:
old_rf = matrix.load_vector(rf_in_path)
reiterate = False
for old_value, new_value in zip(old_rf, new_rf):
mean = 0.5 * (old_value + new_value)
diff = math.fabs(0.5 * (old_value - new_value))
ratio = diff / mean
if ratio > convergence_change:
reiterate = True
if not reiterate:
run.function(shutil.copyfile, prefix + 'RF.txt',
'response.txt')
run.function(shutil.copyfile, prefix + 'SF.mif', 'voxels.mif')
break
app.cleanup(rf_in_path)
app.cleanup(mask_in_path)
iteration += 1
progress.done()
# If we've terminated due to hitting the iteration limiter, we still need to copy the output file(s) to the correct location
if os.path.exists('response.txt'):
app.console('Exited at iteration ' + str(iteration + 1) + ' with ' +
str(sf_voxel_count) +
' SF voxels due to unchanged RF coefficients')
else:
app.console('Exited after maximum ' + str(app.ARGS.max_iters) +
' iterations with ' + str(sf_voxel_count) + ' SF voxels')
run.function(shutil.copyfile,
'iter' + str(app.ARGS.max_iters - 1) + '_RF.txt',
'response.txt')
run.function(shutil.copyfile,
'iter' + str(app.ARGS.max_iters - 1) + '_SF.mif',
'voxels.mif')
run.function(shutil.copyfile, 'response.txt',
path.from_user(app.ARGS.output, False))
if app.ARGS.voxels:
run.command('mrconvert voxels.mif ' + path.from_user(app.ARGS.voxels),
mrconvert_keyval=path.from_user(app.ARGS.input),
force=app.FORCE_OVERWRITE)
def execute(): #pylint: disable=unused-variable
subject_dir = os.path.abspath(path.from_user(app.ARGS.input, False))
if not os.path.isdir(subject_dir):
raise MRtrixError('Input to hsvs algorithm must be a directory')
surf_dir = os.path.join(subject_dir, 'surf')
mri_dir = os.path.join(subject_dir, 'mri')
check_dir(surf_dir)
check_dir(mri_dir)
#aparc_image = os.path.join(mri_dir, 'aparc+aseg.mgz')
aparc_image = 'aparc.mif'
mask_image = os.path.join(mri_dir, 'brainmask.mgz')
reg_file = os.path.join(mri_dir, 'transforms', 'talairach.xfm')
check_file(aparc_image)
check_file(mask_image)
check_file(reg_file)
template_image = 'template.mif' if app.ARGS.template else aparc_image
have_first = False
have_fast = False
fsl_path = os.environ.get('FSLDIR', '')
if fsl_path:
# Use brain-extracted, bias-corrected image for FSL tools
norm_image = os.path.join(mri_dir, 'norm.mgz')
check_file(norm_image)
run.command('mrconvert ' + norm_image + ' T1.nii -stride -1,+2,+3')
# Verify FAST availability
try:
fast_cmd = fsl.exe_name('fast')
except MRtrixError:
fast_cmd = None
if fast_cmd:
have_fast = True
if fast_cmd == 'fast':
fast_suffix = fsl.suffix()
else:
fast_suffix = '.nii.gz'
else:
app.warn('Could not find FSL program fast; script will not use fast for cerebellar tissue segmentation')
# Verify FIRST availability
try:
first_cmd = fsl.exe_name('run_first_all')
except MRtrixError:
first_cmd = None
first_atlas_path = os.path.join(fsl_path, 'data', 'first', 'models_336_bin')
have_first = first_cmd and os.path.isdir(first_atlas_path)
else:
app.warn('Environment variable FSLDIR is not set; script will run without FSL components')
acpc_string = 'anterior ' + ('& posterior commissures' if ATTEMPT_PC else 'commissure')
have_acpcdetect = bool(find_executable('acpcdetect')) and 'ARTHOME' in os.environ
if have_acpcdetect:
if have_fast:
app.console('ACPCdetect and FSL FAST will be used for explicit segmentation of ' + acpc_string)
else:
app.warn('ACPCdetect is installed, but FSL FAST not found; cannot segment ' + acpc_string)
have_acpcdetect = False
else:
app.warn('ACPCdetect not installed; cannot segment ' + acpc_string)
# Need to perform a better search for hippocampal subfield output: names & version numbers may change
have_hipp_subfields = False
hipp_subfield_has_amyg = False
# Could result in multiple matches
hipp_subfield_regex = re.compile(r'^[lr]h\.hippo[a-zA-Z]*Labels-[a-zA-Z0-9]*\.v[0-9]+\.?[a-zA-Z0-9]*\.mg[hz]$')
hipp_subfield_all_images = sorted(list(filter(hipp_subfield_regex.match, os.listdir(mri_dir))))
# Remove any images that provide segmentations in FreeSurfer voxel space; we want the high-resolution versions
hipp_subfield_all_images = [ item for item in hipp_subfield_all_images if 'FSvoxelSpace' not in item ]
# Arrange the images into lr pairs
hipp_subfield_paired_images = [ ]
for lh_filename in [ item for item in hipp_subfield_all_images if item[0] == 'l' ]:
if 'r' + lh_filename[1:] in hipp_subfield_all_images:
hipp_subfield_paired_images.append(lh_filename[1:])
# Choose which of these image pairs we are going to use
for code in [ '.CA.', '.FS60.' ]:
if any(code in filename for filename in hipp_subfield_paired_images):
hipp_subfield_image_suffix = [ filename for filename in hipp_subfield_paired_images if code in filename ][0]
have_hipp_subfields = True
break
# Choose the pair with the shortest filename string if we have no other criteria
if not have_hipp_subfields and hipp_subfield_paired_images:
hipp_subfield_paired_images = sorted(hipp_subfield_paired_images, key=len)
if hipp_subfield_paired_images:
hipp_subfield_image_suffix = hipp_subfield_paired_images[0]
have_hipp_subfields = True
if have_hipp_subfields:
hipp_subfield_has_amyg = 'Amyg' in hipp_subfield_image_suffix
# Perform a similar search for thalamic nuclei submodule output
thal_nuclei_image = None
thal_nuclei_regex = re.compile(r'^ThalamicNuclei\.v[0-9]+\.?[a-zA-Z0-9]*.mg[hz]$')
thal_nuclei_all_images = sorted(list(filter(thal_nuclei_regex.match, os.listdir(mri_dir))))
thal_nuclei_all_images = [ item for item in thal_nuclei_all_images if 'FSvoxelSpace' not in item ]
if thal_nuclei_all_images:
if len(thal_nuclei_all_images) == 1:
thal_nuclei_image = thal_nuclei_all_images[0]
else:
# How to choose which version to use?
# Start with software version
thal_nuclei_versions = [ int(item.split('.')[1].lstrip('v')) for item in thal_nuclei_all_images ]
thal_nuclei_all_images = [ filepath for filepath, version_number in zip(thal_nuclei_all_images, thal_nuclei_versions) if version_number == max(thal_nuclei_versions) ]
if len(thal_nuclei_all_images) == 1:
thal_nuclei_image = thal_nuclei_all_images[0]
else:
# Revert to filename length
thal_nuclei_all_images = sorted(thal_nuclei_all_images, key=len)
thal_nuclei_image = thal_nuclei_all_images[0]
# If particular hippocampal segmentation method is requested, make sure we can perform such;
# if not, decide how to segment hippocampus based on what's available
hippocampi_method = app.ARGS.hippocampi
if hippocampi_method:
if hippocampi_method == 'subfields':
if not have_hipp_subfields:
raise MRtrixError('Could not isolate hippocampal subfields module output (candidate images: ' + str(hipp_subfield_all_images) + ')')
elif hippocampi_method == 'first':
if not have_first:
raise MRtrixError('Cannot use "first" method for hippocampi segmentation; check FSL installation')
else:
if have_hipp_subfields:
hippocampi_method = 'subfields'
app.console('Hippocampal subfields module output detected; will utilise for hippocampi '
+ ('and amygdalae ' if hipp_subfield_has_amyg else '')
+ 'segmentation')
elif have_first:
hippocampi_method = 'first'
app.console('No hippocampal subfields module output detected, but FSL FIRST is installed; '
'will utilise latter for hippocampi segmentation')
else:
hippocampi_method = 'aseg'
app.console('Neither hippocampal subfields module output nor FSL FIRST detected; '
'FreeSurfer aseg will be used for hippocampi segmentation')
if hippocampi_method == 'subfields':
if 'FREESURFER_HOME' not in os.environ:
raise MRtrixError('FREESURFER_HOME environment variable not set; required for use of hippocampal subfields module')
freesurfer_lut_file = os.path.join(os.environ['FREESURFER_HOME'], 'FreeSurferColorLUT.txt')
check_file(freesurfer_lut_file)
hipp_lut_file = os.path.join(path.shared_data_path(), path.script_subdir_name(), 'hsvs', 'HippSubfields.txt')
check_file(hipp_lut_file)
if hipp_subfield_has_amyg:
amyg_lut_file = os.path.join(path.shared_data_path(), path.script_subdir_name(), 'hsvs', 'AmygSubfields.txt')
check_file(amyg_lut_file)
if app.ARGS.sgm_amyg_hipp:
app.warn('Option -sgm_amyg_hipp ignored '
'(hsvs algorithm always assigns hippocampi & ampygdalae as sub-cortical grey matter)')
# Similar logic for thalami
thalami_method = app.ARGS.thalami
if thalami_method:
if thalami_method == 'nuclei':
if not thal_nuclei_image:
raise MRtrixError('Could not find thalamic nuclei module output')
elif thalami_method == 'first':
if not have_first:
raise MRtrixError('Cannot use "first" method for thalami segmentation; check FSL installation')
else:
# Not happy with outputs of thalamic nuclei submodule; default to FIRST
if have_first:
thalami_method = 'first'
if thal_nuclei_image:
app.console('Thalamic nuclei submodule output ignored in favour of FSL FIRST '
'(can override using -thalami option)')
else:
app.console('Will utilise FSL FIRST for thalami segmentation')
elif thal_nuclei_image:
thalami_method = 'nuclei'
app.console('Will utilise detected thalamic nuclei submodule output')
else:
thalami_method = 'aseg'
app.console('Neither thalamic nuclei module output nor FSL FIRST detected; '
'FreeSurfer aseg will be used for thalami segmentation')
###########################
# Commencing segmentation #
###########################
tissue_images = [ [ 'lh.pial.mif', 'rh.pial.mif' ],
[],
[ 'lh.white.mif', 'rh.white.mif' ],
[],
[] ]
# Get the main cerebrum segments; these are already smooth
progress = app.ProgressBar('Mapping FreeSurfer cortical reconstruction to partial volume images', 8)
for hemi in [ 'lh', 'rh' ]:
for basename in [ hemi+'.white', hemi+'.pial' ]:
filepath = os.path.join(surf_dir, basename)
check_file(filepath)
transformed_path = basename + '_realspace.obj'
run.command('meshconvert ' + filepath + ' ' + transformed_path + ' -binary -transform fs2real ' + aparc_image)
progress.increment()
run.command('mesh2voxel ' + transformed_path + ' ' + template_image + ' ' + basename + '.mif')
app.cleanup(transformed_path)
progress.increment()
progress.done()
# Get other structures that need to be converted from the aseg voxel image
from_aseg = list(ASEG_STRUCTURES)
if hippocampi_method == 'subfields':
if not hipp_subfield_has_amyg and not have_first:
from_aseg.extend(AMYG_ASEG)
elif hippocampi_method == 'aseg':
from_aseg.extend(HIPP_ASEG)
from_aseg.extend(AMYG_ASEG)
if thalami_method == 'aseg':
from_aseg.extend(THAL_ASEG)
if not have_first:
from_aseg.extend(OTHER_SGM_ASEG)
progress = app.ProgressBar('Smoothing non-cortical structures segmented by FreeSurfer', len(from_aseg) + 2)
for (index, tissue, name) in from_aseg:
init_mesh_path = name + '_init.vtk'
smoothed_mesh_path = name + '.vtk'
run.command('mrcalc ' + aparc_image + ' ' + str(index) + ' -eq - | voxel2mesh - -threshold 0.5 ' + init_mesh_path)
run.command('meshfilter ' + init_mesh_path + ' smooth ' + smoothed_mesh_path)
app.cleanup(init_mesh_path)
run.command('mesh2voxel ' + smoothed_mesh_path + ' ' + template_image + ' ' + name + '.mif')
app.cleanup(smoothed_mesh_path)
tissue_images[tissue-1].append(name + '.mif')
progress.increment()
# Lateral ventricles are separate as we want to combine with choroid plexus prior to mesh conversion
for hemi_index, hemi_name in enumerate(['Left', 'Right']):
name = hemi_name + '_LatVent_ChorPlex'
init_mesh_path = name + '_init.vtk'
smoothed_mesh_path = name + '.vtk'
run.command('mrcalc ' + ' '.join(aparc_image + ' ' + str(index) + ' -eq' for index, tissue, name in VENTRICLE_CP_ASEG[hemi_index]) + ' -add - | '
+ 'voxel2mesh - -threshold 0.5 ' + init_mesh_path)
run.command('meshfilter ' + init_mesh_path + ' smooth ' + smoothed_mesh_path)
app.cleanup(init_mesh_path)
run.command('mesh2voxel ' + smoothed_mesh_path + ' ' + template_image + ' ' + name + '.mif')
app.cleanup(smoothed_mesh_path)
tissue_images[3].append(name + '.mif')
progress.increment()
progress.done()
# Combine corpus callosum segments before smoothing
progress = app.ProgressBar('Combining and smoothing corpus callosum segmentation', len(CORPUS_CALLOSUM_ASEG) + 3)
for (index, name) in CORPUS_CALLOSUM_ASEG:
run.command('mrcalc ' + aparc_image + ' ' + str(index) + ' -eq ' + name + '.mif -datatype bit')
progress.increment()
cc_init_mesh_path = 'combined_corpus_callosum_init.vtk'
cc_smoothed_mesh_path = 'combined_corpus_callosum.vtk'
run.command('mrmath ' + ' '.join([ name + '.mif' for (index, name) in CORPUS_CALLOSUM_ASEG ]) + ' sum - | voxel2mesh - -threshold 0.5 ' + cc_init_mesh_path)
for name in [ n for _, n in CORPUS_CALLOSUM_ASEG ]:
app.cleanup(name + '.mif')
progress.increment()
run.command('meshfilter ' + cc_init_mesh_path + ' smooth ' + cc_smoothed_mesh_path)
app.cleanup(cc_init_mesh_path)
progress.increment()
run.command('mesh2voxel ' + cc_smoothed_mesh_path + ' ' + template_image + ' combined_corpus_callosum.mif')
app.cleanup(cc_smoothed_mesh_path)
progress.done()
tissue_images[2].append('combined_corpus_callosum.mif')
# Deal with brain stem, including determining those voxels that should
# be erased from the 5TT image in order for streamlines traversing down
# the spinal column to be terminated & accepted
bs_fullmask_path = 'brain_stem_init.mif'
bs_cropmask_path = ''
progress = app.ProgressBar('Segmenting and cropping brain stem', 5)
run.command('mrcalc ' + aparc_image + ' ' + str(BRAIN_STEM_ASEG[0][0]) + ' -eq '
+ ' -add '.join([ aparc_image + ' ' + str(index) + ' -eq' for index, name in BRAIN_STEM_ASEG[1:] ]) + ' -add '
+ bs_fullmask_path + ' -datatype bit')
progress.increment()
bs_init_mesh_path = 'brain_stem_init.vtk'
run.command('voxel2mesh ' + bs_fullmask_path + ' ' + bs_init_mesh_path)
progress.increment()
bs_smoothed_mesh_path = 'brain_stem.vtk'
run.command('meshfilter ' + bs_init_mesh_path + ' smooth ' + bs_smoothed_mesh_path)
app.cleanup(bs_init_mesh_path)
progress.increment()
run.command('mesh2voxel ' + bs_smoothed_mesh_path + ' ' + template_image + ' brain_stem.mif')
app.cleanup(bs_smoothed_mesh_path)
progress.increment()
fourthventricle_zmin = min([ int(line.split()[2]) for line in run.command('maskdump 4th-Ventricle.mif')[0].splitlines() ])
if fourthventricle_zmin:
bs_cropmask_path = 'brain_stem_crop.mif'
run.command('mredit brain_stem.mif - ' + ' '.join([ '-plane 2 ' + str(index) + ' 0' for index in range(0, fourthventricle_zmin) ]) + ' | '
'mrcalc brain_stem.mif - -sub 1e-6 -gt ' + bs_cropmask_path + ' -datatype bit')
app.cleanup(bs_fullmask_path)
progress.done()
if hippocampi_method == 'subfields':
progress = app.ProgressBar('Using detected FreeSurfer hippocampal subfields module output',
64 if hipp_subfield_has_amyg else 32)
subfields = [ ( hipp_lut_file, 'hipp' ) ]
if hipp_subfield_has_amyg:
subfields.append(( amyg_lut_file, 'amyg' ))
for subfields_lut_file, structure_name in subfields:
for hemi, filename in zip([ 'Left', 'Right'], [ prefix + hipp_subfield_image_suffix for prefix in [ 'l', 'r' ] ]):
# Extract individual components from image and assign to different tissues
subfields_all_tissues_image = hemi + '_' + structure_name + '_subfields.mif'
run.command('labelconvert ' + os.path.join(mri_dir, filename) + ' ' + freesurfer_lut_file + ' ' + subfields_lut_file + ' ' + subfields_all_tissues_image)
progress.increment()
for tissue in range(0, 5):
init_mesh_path = hemi + '_' + structure_name + '_subfield_' + str(tissue) + '_init.vtk'
smooth_mesh_path = hemi + '_' + structure_name + '_subfield_' + str(tissue) + '.vtk'
subfield_tissue_image = hemi + '_' + structure_name + '_subfield_' + str(tissue) + '.mif'
run.command('mrcalc ' + subfields_all_tissues_image + ' ' + str(tissue+1) + ' -eq - | ' + \
'voxel2mesh - ' + init_mesh_path)
progress.increment()
# Since the hippocampal subfields segmentation can include some fine structures, reduce the extent of smoothing
run.command('meshfilter ' + init_mesh_path + ' smooth ' + smooth_mesh_path + ' -smooth_spatial 2 -smooth_influence 2')
app.cleanup(init_mesh_path)
progress.increment()
run.command('mesh2voxel ' + smooth_mesh_path + ' ' + template_image + ' ' + subfield_tissue_image)
app.cleanup(smooth_mesh_path)
progress.increment()
tissue_images[tissue].append(subfield_tissue_image)
app.cleanup(subfields_all_tissues_image)
progress.done()
if thalami_method == 'nuclei':
progress = app.ProgressBar('Using detected FreeSurfer thalamic nuclei module output', 6)
for hemi in ['Left', 'Right']:
thal_mask_path = hemi + '_Thalamus_mask.mif'
init_mesh_path = hemi + '_Thalamus_init.vtk'
smooth_mesh_path = hemi + '_Thalamus.vtk'
thalamus_image = hemi + '_Thalamus.mif'
if hemi == 'Right':
run.command('mrthreshold ' + os.path.join(mri_dir, thal_nuclei_image) + ' -abs 8200 ' + thal_mask_path)
else:
run.command('mrcalc ' + os.path.join(mri_dir, thal_nuclei_image) + ' 0 -gt '
+ os.path.join(mri_dir, thal_nuclei_image) + ' 8200 -lt '
+ '-mult ' + thal_mask_path)
run.command('voxel2mesh ' + thal_mask_path + ' ' + init_mesh_path)
app.cleanup(thal_mask_path)
progress.increment()
run.command('meshfilter ' + init_mesh_path + ' smooth ' + smooth_mesh_path + ' -smooth_spatial 2 -smooth_influence 2')
app.cleanup(init_mesh_path)
progress.increment()
run.command('mesh2voxel ' + smooth_mesh_path + ' ' + template_image + ' ' + thalamus_image)
app.cleanup(smooth_mesh_path)
progress.increment()
tissue_images[1].append(thalamus_image)
progress.done()
if have_first:
app.console('Running FSL FIRST to segment sub-cortical grey matter structures')
from_first = SGM_FIRST_MAP.copy()
if hippocampi_method == 'subfields':
from_first = { key: value for key, value in from_first.items() if 'Hippocampus' not in value }
if hipp_subfield_has_amyg:
from_first = { key: value for key, value in from_first.items() if 'Amygdala' not in value }
elif hippocampi_method == 'aseg':
from_first = { key: value for key, value in from_first.items() if 'Hippocampus' not in value and 'Amygdala' not in value }
if thalami_method != 'first':
from_first = { key: value for key, value in from_first.items() if 'Thalamus' not in value }
run.command(first_cmd + ' -s ' + ','.join(from_first.keys()) + ' -i T1.nii -b -o first')
fsl.check_first('first', from_first.keys())
app.cleanup(glob.glob('T1_to_std_sub.*'))
progress = app.ProgressBar('Mapping FIRST segmentations to image', 2*len(from_first))
for key, value in from_first.items():
vtk_in_path = 'first-' + key + '_first.vtk'
vtk_converted_path = 'first-' + key + '_transformed.vtk'
run.command('meshconvert ' + vtk_in_path + ' ' + vtk_converted_path + ' -transform first2real T1.nii')
app.cleanup(vtk_in_path)
progress.increment()
run.command('mesh2voxel ' + vtk_converted_path + ' ' + template_image + ' ' + value + '.mif')
app.cleanup(vtk_converted_path)
tissue_images[1].append(value + '.mif')
progress.increment()
if not have_fast:
app.cleanup('T1.nii')
app.cleanup(glob.glob('first*'))
progress.done()
# Run ACPCdetect, use results to draw spherical ROIs on T1 that will be fed to FSL FAST,
# the WM components of which will then be added to the 5TT
if have_acpcdetect:
progress = app.ProgressBar('Using ACPCdetect and FAST to segment ' + acpc_string, 5)
# ACPCdetect requires input image to be 16-bit
# We also want to realign to RAS beforehand so that we can interpret the output voxel locations properly
acpcdetect_input_image = 'T1RAS_16b.nii'
run.command('mrconvert ' + norm_image + ' -datatype uint16 -stride +1,+2,+3 ' + acpcdetect_input_image)
progress.increment()
run.command('acpcdetect -i ' + acpcdetect_input_image)
progress.increment()
# We need the header in order to go from voxel coordinates to scanner coordinates
acpcdetect_input_header = image.Header(acpcdetect_input_image)
acpcdetect_output_path = os.path.splitext(acpcdetect_input_image)[0] + '_ACPC.txt'
app.cleanup(acpcdetect_input_image)
with open(acpcdetect_output_path, 'r') as acpc_file:
acpcdetect_output_data = acpc_file.read().splitlines()
app.cleanup(glob.glob(os.path.splitext(acpcdetect_input_image)[0] + "*"))
# Need to scan through the contents of this file,
# isolating the AC and PC locations
ac_voxel = pc_voxel = None
for index, line in enumerate(acpcdetect_output_data):
if 'AC' in line and 'voxel location' in line:
ac_voxel = [float(item) for item in acpcdetect_output_data[index+1].strip().split()]
elif 'PC' in line and 'voxel location' in line:
pc_voxel = [float(item) for item in acpcdetect_output_data[index+1].strip().split()]
if not ac_voxel or not pc_voxel:
raise MRtrixError('Error parsing text file from "acpcdetect"')
def voxel2scanner(voxel, header):
return [ voxel[0]*header.spacing()[0]*header.transform()[axis][0]
+ voxel[1]*header.spacing()[1]*header.transform()[axis][1]
+ voxel[2]*header.spacing()[2]*header.transform()[axis][2]
+ header.transform()[axis][3]
for axis in range(0,3) ]
ac_scanner = voxel2scanner(ac_voxel, acpcdetect_input_header)
pc_scanner = voxel2scanner(pc_voxel, acpcdetect_input_header)
# Generate the mask image within which FAST will be run
acpc_prefix = 'ACPC' if ATTEMPT_PC else 'AC'
acpc_mask_image = acpc_prefix + '_FAST_mask.mif'
run.command('mrcalc ' + template_image + ' nan -eq - | '
'mredit - ' + acpc_mask_image + ' -scanner '
'-sphere ' + ','.join(str(value) for value in ac_scanner) + ' 8 1 '
+ ('-sphere ' + ','.join(str(value) for value in pc_scanner) + ' 5 1' if ATTEMPT_PC else ''))
progress.increment()
acpc_t1_masked_image = acpc_prefix + '_T1.nii'
run.command('mrtransform ' + norm_image + ' -template ' + template_image + ' - | '
'mrcalc - ' + acpc_mask_image + ' -mult ' + acpc_t1_masked_image)
app.cleanup(acpc_mask_image)
progress.increment()
run.command(fast_cmd + ' -N ' + acpc_t1_masked_image)
app.cleanup(acpc_t1_masked_image)
progress.increment()
# Ideally don't want to have to add these manually; instead add all outputs from FAST
# to the 5TT (both cerebellum and AC / PC) in a single go
# This should involve grabbing just the WM component of these images
# Actually, in retrospect, it may be preferable to do the AC PC segmentation
# earlier on, and simply add them to the list of WM structures
acpc_wm_image = acpc_prefix + '.mif'
run.command('mrconvert ' + fsl.find_image(acpc_prefix + '_T1_pve_2') + ' ' + acpc_wm_image)
tissue_images[2].append(acpc_wm_image)
app.cleanup(glob.glob(os.path.splitext(acpc_t1_masked_image)[0] + '*'))
progress.done()
# If we don't have FAST, do cerebellar segmentation in a comparable way to the cortical GM / WM:
# Generate one 'pial-like' surface containing the GM and WM of the cerebellum,
# and another with just the WM
if not have_fast:
progress = app.ProgressBar('Adding FreeSurfer cerebellar segmentations directly', 6)
for hemi in [ 'Left-', 'Right-' ]:
wm_index = [ index for index, tissue, name in CEREBELLUM_ASEG if name.startswith(hemi) and 'White' in name ][0]
gm_index = [ index for index, tissue, name in CEREBELLUM_ASEG if name.startswith(hemi) and 'Cortex' in name ][0]
run.command('mrcalc ' + aparc_image + ' ' + str(wm_index) + ' -eq ' + aparc_image + ' ' + str(gm_index) + ' -eq -add - | ' + \
'voxel2mesh - ' + hemi + 'cerebellum_all_init.vtk')
progress.increment()
run.command('mrcalc ' + aparc_image + ' ' + str(gm_index) + ' -eq - | ' + \
'voxel2mesh - ' + hemi + 'cerebellum_grey_init.vtk')
progress.increment()
for name, tissue in { 'all':2, 'grey':1 }.items():
run.command('meshfilter ' + hemi + 'cerebellum_' + name + '_init.vtk smooth ' + hemi + 'cerebellum_' + name + '.vtk')
app.cleanup(hemi + 'cerebellum_' + name + '_init.vtk')
progress.increment()
run.command('mesh2voxel ' + hemi + 'cerebellum_' + name + '.vtk ' + template_image + ' ' + hemi + 'cerebellum_' + name + '.mif')
app.cleanup(hemi + 'cerebellum_' + name + '.vtk')
progress.increment()
tissue_images[tissue].append(hemi + 'cerebellum_' + name + '.mif')
progress.done()
# Construct images with the partial volume of each tissue
progress = app.ProgressBar('Combining segmentations of all structures corresponding to each tissue type', 5)
for tissue in range(0,5):
run.command('mrmath ' + ' '.join(tissue_images[tissue]) + (' brain_stem.mif' if tissue == 2 else '') + ' sum - | mrcalc - 1.0 -min tissue' + str(tissue) + '_init.mif')
app.cleanup(tissue_images[tissue])
progress.increment()
progress.done()
# This can hopefully be done with a connected-component analysis: Take just the WM image, and
# fill in any gaps (i.e. select the inverse, select the largest connected component, invert again)
# Make sure that floating-point values are handled appropriately
# Combine these images together using the appropriate logic in order to form the 5TT image
progress = app.ProgressBar('Modulating segmentation images based on other tissues', 9)
tissue_images = [ 'tissue0.mif', 'tissue1.mif', 'tissue2.mif', 'tissue3.mif', 'tissue4.mif' ]
run.function(os.rename, 'tissue4_init.mif', 'tissue4.mif')
progress.increment()
run.command('mrcalc tissue3_init.mif tissue3_init.mif ' + tissue_images[4] + ' -add 1.0 -sub 0.0 -max -sub 0.0 -max ' + tissue_images[3])
app.cleanup('tissue3_init.mif')
progress.increment()
run.command('mrmath ' + ' '.join(tissue_images[3:5]) + ' sum tissuesum_34.mif')
progress.increment()
run.command('mrcalc tissue1_init.mif tissue1_init.mif tissuesum_34.mif -add 1.0 -sub 0.0 -max -sub 0.0 -max ' + tissue_images[1])
app.cleanup('tissue1_init.mif')
app.cleanup('tissuesum_34.mif')
progress.increment()
run.command('mrmath ' + tissue_images[1] + ' ' + ' '.join(tissue_images[3:5]) + ' sum tissuesum_134.mif')
progress.increment()
run.command('mrcalc tissue2_init.mif tissue2_init.mif tissuesum_134.mif -add 1.0 -sub 0.0 -max -sub 0.0 -max ' + tissue_images[2])
app.cleanup('tissue2_init.mif')
app.cleanup('tissuesum_134.mif')
progress.increment()
run.command('mrmath ' + ' '.join(tissue_images[1:5]) + ' sum tissuesum_1234.mif')
progress.increment()
run.command('mrcalc tissue0_init.mif tissue0_init.mif tissuesum_1234.mif -add 1.0 -sub 0.0 -max -sub 0.0 -max ' + tissue_images[0])
app.cleanup('tissue0_init.mif')
app.cleanup('tissuesum_1234.mif')
progress.increment()
tissue_sum_image = 'tissuesum_01234.mif'
run.command('mrmath ' + ' '.join(tissue_images) + ' sum ' + tissue_sum_image)
progress.done()
if app.ARGS.template:
run.command('mrtransform ' + mask_image + ' -template template.mif - | mrthreshold - brainmask.mif -abs 0.5')
mask_image = 'brainmask.mif'
# Branch depending on whether or not FSL fast will be used to re-segment the cerebellum
if have_fast:
# How to support -template option?
# - Re-grid norm.mgz to template image before running FAST
# - Re-grid FAST output to template image
# Consider splitting, including initial mapping of cerebellar regions:
# - If we're not using a separate template image, just map cerebellar regions to voxels to
# produce a mask, and run FAST within that mask
# - If we have a template, combine cerebellar regions, convert to surfaces (one per hemisphere),
# map these to the template image, run FIRST on a binary mask from this, then
# re-combine this with the tissue maps from other sources based on the estimated PVF of
# cerebellum meshes
cerebellum_volume_image = 'Cerebellum_volume.mif'
cerebellum_mask_image = 'Cerebellum_mask.mif'
t1_cerebellum_masked = 'T1_cerebellum_precrop.mif'
if app.ARGS.template:
# If this is the case, then we haven't yet performed any cerebellar segmentation / meshing
# What we want to do is: for each hemisphere, combine all three "cerebellar" segments from FreeSurfer,
# convert to a surface, map that surface to the template image
progress = app.ProgressBar('Preparing images of cerebellum for intensity-based segmentation', 9)
cerebellar_hemi_pvf_images = [ ]
for hemi in [ 'Left', 'Right' ]:
init_mesh_path = hemi + '-Cerebellum-All-Init.vtk'
smooth_mesh_path = hemi + '-Cerebellum-All-Smooth.vtk'
pvf_image_path = hemi + '-Cerebellum-PVF-Template.mif'
cerebellum_aseg_hemi = [ entry for entry in CEREBELLUM_ASEG if hemi in entry[2] ]
run.command('mrcalc ' + aparc_image + ' ' + str(cerebellum_aseg_hemi[0][0]) + ' -eq ' + \
' -add '.join([ aparc_image + ' ' + str(index) + ' -eq' for index, tissue, name in cerebellum_aseg_hemi[1:] ]) + ' -add - | ' + \
'voxel2mesh - ' + init_mesh_path)
progress.increment()
run.command('meshfilter ' + init_mesh_path + ' smooth ' + smooth_mesh_path)
app.cleanup(init_mesh_path)
progress.increment()
run.command('mesh2voxel ' + smooth_mesh_path + ' ' + template_image + ' ' + pvf_image_path)
app.cleanup(smooth_mesh_path)
cerebellar_hemi_pvf_images.append(pvf_image_path)
progress.increment()
# Combine the two hemispheres together into:
# - An image in preparation for running FAST
# - A combined total partial volume fraction image that will be later used for tissue recombination
run.command('mrcalc ' + ' '.join(cerebellar_hemi_pvf_images) + ' -add 1.0 -min ' + cerebellum_volume_image)
app.cleanup(cerebellar_hemi_pvf_images)
progress.increment()
run.command('mrthreshold ' + cerebellum_volume_image + ' ' + cerebellum_mask_image + ' -abs 1e-6')
progress.increment()
run.command('mrtransform ' + norm_image + ' -template ' + template_image + ' - | ' + \
'mrcalc - ' + cerebellum_mask_image + ' -mult ' + t1_cerebellum_masked)
progress.done()
else:
app.console('Preparing images of cerebellum for intensity-based segmentation')
run.command('mrcalc ' + aparc_image + ' ' + str(CEREBELLUM_ASEG[0][0]) + ' -eq ' + \
' -add '.join([ aparc_image + ' ' + str(index) + ' -eq' for index, tissue, name in CEREBELLUM_ASEG[1:] ]) + ' -add ' + \
cerebellum_volume_image)
cerebellum_mask_image = cerebellum_volume_image
run.command('mrcalc T1.nii ' + cerebellum_mask_image + ' -mult ' + t1_cerebellum_masked)
app.cleanup('T1.nii')
# Any code below here should be compatible with cerebellum_volume_image.mif containing partial volume fractions
# (in the case of no explicit template image, it's a mask, but the logic still applies)
app.console('Running FSL fast to segment the cerebellum based on intensity information')
# Run FSL FAST just within the cerebellum
# FAST memory usage can also be huge when using a high-resolution template image:
# Crop T1 image around the cerebellum before feeding to FAST, then re-sample to full template image FoV
fast_input_image = 'T1_cerebellum.nii'
run.command('mrgrid ' + t1_cerebellum_masked + ' crop -mask ' + cerebellum_mask_image + ' ' + fast_input_image)
app.cleanup(t1_cerebellum_masked)
# Cleanup of cerebellum_mask_image:
# May be same image as cerebellum_volume_image, which is required later
if cerebellum_mask_image != cerebellum_volume_image:
app.cleanup(cerebellum_mask_image)
run.command(fast_cmd + ' -N ' + fast_input_image)
app.cleanup(fast_input_image)
# Use glob to clean up unwanted FAST outputs
fast_output_prefix = os.path.splitext(fast_input_image)[0]
fast_pve_output_prefix = fast_output_prefix + '_pve_'
app.cleanup([ entry for entry in glob.glob(fast_output_prefix + '*') if not fast_pve_output_prefix in entry ])
progress = app.ProgressBar('Introducing intensity-based cerebellar segmentation into the 5TT image', 10)
fast_outputs_cropped = [ fast_pve_output_prefix + str(n) + fast_suffix for n in range(0,3) ]
fast_outputs_template = [ 'FAST_' + str(n) + '.mif' for n in range(0,3) ]
for inpath, outpath in zip(fast_outputs_cropped, fast_outputs_template):
run.command('mrtransform ' + inpath + ' -interp nearest -template ' + template_image + ' ' + outpath)
app.cleanup(inpath)
progress.increment()
if app.ARGS.template:
app.cleanup(template_image)
# Generate the revised tissue images, using output from FAST inside the cerebellum and
# output from previous processing everywhere else
# Note that the middle intensity (grey matter) in the FAST output here gets assigned
# to the sub-cortical grey matter component
# Some of these voxels may have existing non-zero tissue components.
# In that case, let's find a multiplier to apply to cerebellum tissues such that the
# sum does not exceed 1.0
new_tissue_images = [ 'tissue0_fast.mif', 'tissue1_fast.mif', 'tissue2_fast.mif', 'tissue3_fast.mif', 'tissue4_fast.mif' ]
new_tissue_sum_image = 'tissuesum_01234_fast.mif'
cerebellum_multiplier_image = 'Cerebellar_multiplier.mif'
run.command('mrcalc ' + cerebellum_volume_image + ' ' + tissue_sum_image + ' -add 0.5 -gt 1.0 ' + tissue_sum_image + ' -sub 0.0 -if ' + cerebellum_multiplier_image)
app.cleanup(cerebellum_volume_image)
progress.increment()
run.command('mrconvert ' + tissue_images[0] + ' ' + new_tissue_images[0])
app.cleanup(tissue_images[0])
progress.increment()
run.command('mrcalc ' + tissue_images[1] + ' ' + cerebellum_multiplier_image + ' ' + fast_outputs_template[1] + ' -mult -add ' + new_tissue_images[1])
app.cleanup(tissue_images[1])
app.cleanup(fast_outputs_template[1])
progress.increment()
run.command('mrcalc ' + tissue_images[2] + ' ' + cerebellum_multiplier_image + ' ' + fast_outputs_template[2] + ' -mult -add ' + new_tissue_images[2])
app.cleanup(tissue_images[2])
app.cleanup(fast_outputs_template[2])
progress.increment()
run.command('mrcalc ' + tissue_images[3] + ' ' + cerebellum_multiplier_image + ' ' + fast_outputs_template[0] + ' -mult -add ' + new_tissue_images[3])
app.cleanup(tissue_images[3])
app.cleanup(fast_outputs_template[0])
app.cleanup(cerebellum_multiplier_image)
progress.increment()
run.command('mrconvert ' + tissue_images[4] + ' ' + new_tissue_images[4])
app.cleanup(tissue_images[4])
progress.increment()
run.command('mrmath ' + ' '.join(new_tissue_images) + ' sum ' + new_tissue_sum_image)
app.cleanup(tissue_sum_image)
progress.done()
tissue_images = new_tissue_images
tissue_sum_image = new_tissue_sum_image
# For all voxels within FreeSurfer's brain mask, add to the CSF image in order to make the sum 1.0
progress = app.ProgressBar('Performing fill operations to preserve unity tissue volume', 2)
# Some voxels may get a non-zero cortical GM fraction due to native use of the surface representation, yet
# these voxels are actually outside FreeSurfer's own provided brain mask. So what we need to do here is
# get the union of the tissue sum nonzero image and the mask image, and use that at the -mult step of the
# mrcalc call.
# Required image: (tissue_sum_image > 0.0) || mask_image
# tissue_sum_image 0.0 -gt mask_image -add 1.0 -min
new_tissue_images = [ tissue_images[0],
tissue_images[1],
tissue_images[2],
os.path.splitext(tissue_images[3])[0] + '_filled.mif',
tissue_images[4] ]
csf_fill_image = 'csf_fill.mif'
run.command('mrcalc 1.0 ' + tissue_sum_image + ' -sub ' + tissue_sum_image + ' 0.0 -gt ' + mask_image + ' -add 1.0 -min -mult 0.0 -max ' + csf_fill_image)
app.cleanup(tissue_sum_image)
# If no template is specified, this file is part of the FreeSurfer output; hence don't modify
if app.ARGS.template:
app.cleanup(mask_image)
progress.increment()
run.command('mrcalc ' + tissue_images[3] + ' ' + csf_fill_image + ' -add ' + new_tissue_images[3])
app.cleanup(csf_fill_image)
app.cleanup(tissue_images[3])
progress.done()
tissue_images = new_tissue_images
# Move brain stem from white matter to pathology at final step:
# this prevents the brain stem segmentation from overwriting other
# structures that it otherwise wouldn't if it were written to WM
if not app.ARGS.white_stem:
progress = app.ProgressBar('Moving brain stem to volume index 4', 3)
new_tissue_images = [ tissue_images[0],
tissue_images[1],
os.path.splitext(tissue_images[2])[0] + '_no_brainstem.mif',
tissue_images[3],
os.path.splitext(tissue_images[4])[0] + '_with_brainstem.mif' ]
run.command('mrcalc ' + tissue_images[2] + ' brain_stem.mif -min brain_stem_white_overlap.mif')
app.cleanup('brain_stem.mif')
progress.increment()
run.command('mrcalc ' + tissue_images[2] + ' brain_stem_white_overlap.mif -sub ' + new_tissue_images[2])
app.cleanup(tissue_images[2])
progress.increment()
run.command('mrcalc ' + tissue_images[4] + ' brain_stem_white_overlap.mif -add ' + new_tissue_images[4])
app.cleanup(tissue_images[4])
app.cleanup('brain_stem_white_overlap.mif')
progress.done()
tissue_images = new_tissue_images
# Finally, concatenate the volumes to produce the 5TT image
app.console('Concatenating tissue volumes into 5TT format')
precrop_result_image = '5TT.mif'
if bs_cropmask_path:
run.command('mrcat ' + ' '.join(tissue_images) + ' - -axis 3 | ' + \
'5ttedit - ' + precrop_result_image + ' -none ' + bs_cropmask_path)
app.cleanup(bs_cropmask_path)
else:
run.command('mrcat ' + ' '.join(tissue_images) + ' ' + precrop_result_image + ' -axis 3')
app.cleanup(tissue_images)
# Maybe don't go off all tissues here, since FreeSurfer's mask can be fairly liberal;
# instead get just a voxel clearance from all other tissue types (maybe two)
if app.ARGS.nocrop:
run.function(os.rename, precrop_result_image, 'result.mif')
else:
app.console('Cropping final 5TT image')
crop_mask_image = 'crop_mask.mif'
run.command('mrconvert ' + precrop_result_image + ' -coord 3 0,1,2,4 - | mrmath - sum - -axis 3 | mrthreshold - - -abs 0.001 | maskfilter - dilate ' + crop_mask_image)
run.command('mrgrid ' + precrop_result_image + ' crop result.mif -mask ' + crop_mask_image)
app.cleanup(crop_mask_image)
app.cleanup(precrop_result_image)
run.command('mrconvert result.mif ' + path.from_user(app.ARGS.output),
mrconvert_keyval=path.from_user(os.path.join(app.ARGS.input, 'mri', 'aparc+aseg.mgz'), True),
force=app.FORCE_OVERWRITE)