def read(filename):
if os.path.exists:
with open(filename,'rb') as p:
try:
newpatient = pickle.load(p)
except:
raise mincError(' -- Problem reading the pickle %s !'
% filename)
return newpatient
else:
raise mincError(' -- Pickle %s does not exists!' % filename)
def build_estimate(description_json, parameters, output_prefix, int_par_count=None):
desc=None
with open(description_json, 'r') as f:
desc=json.load(f)
intensity_parameters=parameters
velocity_parameters=parameters
if int_par_count is not None:
intensity_parameters=parameters[:int_par_count]
velocity_parameters=parameters[int_par_count:]
if len(velocity_parameters)!=len(desc["velocity_model"]["volume"]) or \
len(intensity_parameters)!=len(desc["intensity_model"]["volume"]):
print(desc["intensity_model"]["volume"])
print("intensity_parameters={}".format(repr(intensity_parameters)))
print(desc["velocity_model"]["volume"])
print("velocity_parameters={}".format(repr(velocity_parameters)))
raise mincError("{} inconsisten number of paramters, expected {}".
format(repr(intensity_parameters),
len(desc["velocity_model"]["volume"])))
velocity=MriDatasetRegress(from_dict=desc["velocity_model"])
intensity=MriDatasetRegress(from_dict=desc["intensity_model"])
output_scan=MriDataset(prefix=os.path.dirname(output_prefix),name=os.path.basename(output_prefix))
output_transform=LDDMriTransform(prefix=os.path.dirname(output_prefix),name=os.path.basename(output_prefix))
build_approximation(intensity, velocity,
intensity_parameters, velocity_parameters,
output_scan, output_transform)
def apply_linear_model(lin_model, parameters, output_volume):
"""build a volume, for a given regression model and parameters"""
try:
with mincTools() as m:
if lin_model.N != len(parameters):
raise mincError("Expected: {} parameters, got {}".format(
lin_model.N, len(parameters)))
# create minccalc expression
_exp = []
for i in range(0, lin_model.N):
_exp.append('A[{}]*{}'.format(i, parameters[i]))
exp = '+'.join(_exp)
m.calc(lin_model.volume, exp, output_volume)
return True
except mincError as e:
print("Exception in apply_linear_model:{}".format(str(e)))
traceback.print_exc(file=sys.stdout)
raise
except:
print("Exception in apply_linear_model:{}".format(sys.exc_info()[0]))
traceback.print_exc(file=sys.stdout)
raise
def normalize_intensity(in_scan, out_scan, parameters={}, model=None):
""" Perform global intensity scale normalization
"""
# TODO: make output exp file
with mincTools() as minc:
if not minc.checkfiles(inputs=[in_scan.scan], outputs=[out_scan.scan]):
return
if parameters is not None and not parameters.get('disable', False):
order = parameters.get('order', 1)
_model = None
#
if model is None:
_model = parameters.get('model', None)
else:
_model = model.scan
if _model is None:
raise mincError('Need model ')
scan_mask = None
model_mask = None
if in_scan.mask is not None and model is not None:
scan_mask = in_scan.mask
model_mask = model.mask
elif parameters.get('nuyl', False):
minc.nuyl_normalize(in_scan.scan,
_model,
out_scan.scan,
source_mask=scan_mask,
target_mask=model_mask)
elif parameters.get('nuyl2', False):
hl.nuyl_normalize2(
in_scan.scan,
_model,
out_scan.scan,
#source_mask=input_mask,target_mask=model_mask,
fwhm=parameters.get('nuyl2_fwhm', 2.0),
iterations=parameters.get('nuyl2_iter', 4),
)
else:
minc.volume_pol(in_scan.scan,
_model,
out_scan.scan,
order=order,
source_mask=scan_mask,
target_mask=model_mask)
else: # HACK just by-pass processing if parameters are empty or disabled
shutil.copyfile(in_scan.scan, out_scan.scan)
def main():
options = parse_options()
try:
if os.path.exists(options.output) and not options.clobber:
raise mincError('File exists : {}'.format(options.output))
parameters = None
if options.param is not None:
with open(options.param, 'r') as f:
parameters = json.load(f)
if options.method == 'ants':
ipl.ants_registration.non_linear_register_ants2(
options.source,
options.target,
options.output,
source_mask=options.source_mask,
target_mask=options.target_mask,
init_xfm=options.init_xfm,
parameters=parameters,
downsample=options.downsample,
level=options.level,
start=options.start,
)
elif options.method == 'elastix':
if parameters is None:
levels = ''
for i in range(int(math.log(options.start) / math.log(2)), -1,
-1):
res = 2**i
if res >= options.level:
levels += '{} {} {} '.format(res, res, res)
parameters = {
'pyramid': levels,
'grid_spacing': options.level * 3.0
}
ipl.elastix_registration.register_elastix(
options.source,
options.target,
output_xfm=options.output,
source_mask=options.source_mask,
target_mask=options.target_mask,
init_xfm=options.init_xfm,
parameters=parameters,
downsample=options.downsample,
downsample_grid=options.level,
nl=True)
else:
objective = '-' + options.objective
if parameters is not None:
objective = parameters.get('objective')
ipl.registration.non_linear_register_full(
options.source,
options.target,
options.output,
source_mask=options.source_mask,
target_mask=options.target_mask,
init_xfm=options.init_xfm,
parameters=parameters,
downsample=options.downsample,
level=options.level,
start=options.start)
if options.qc is not None:
# create QC image for deformation field
create_qc(options.source, options.target, options.output,
options.qc)
except mincError as e:
print(str(e), file=sys.stderr)
traceback.print_exc(file=sys.stderr)
except:
print("Exception :{}".format(sys.exc_info()[0]))
traceback.print_exc(file=sys.stderr)
def regress(samples,
initial_model=None,
initial_int_model=None,
initial_def_model=None,
output_int_model=None,
output_def_model=None,
output_residuals_int=None,
output_residuals_def=None,
prefix='.',
options={}):
""" perform iterative model creation"""
try:
# make sure all input scans have parameters
N_int = None
N_def = None
int_design_matrix = []
def_design_matrix = []
nomask = False
for s in samples:
if N_int is None:
N_int = len(s.par_int)
elif N_int != len(s.par_int):
raise mincError(
"Sample {} have inconsisten number of int paramters: {} expected {}"
.format(repr(s), len(s), N_int))
if N_def is None:
N_def = len(s.par_def)
elif N_def != len(s.par_def):
raise mincError(
"Sample {} have inconsisten number of int paramters: {} expected {}"
.format(repr(s), len(s), N_def))
int_design_matrix.append(s.par_int)
def_design_matrix.append(s.par_def)
if s.mask is None:
nomask = True
#print("Intensity design matrix=\n{}".format(repr(int_design_matrix)))
#print("Velocity design matrix=\n{}".format(repr(def_design_matrix)))
ref_model = None
# current estimate of template
if initial_model is not None:
current_int_model = initial_model
current_def_model = None
ref_model = initial_model.scan
else:
current_int_model = initial_int_model
current_def_model = initial_def_model
ref_model = initial_int_model.volume[0]
transforms = []
full_transforms = []
protocol = options.get('protocol', [{
'iter': 4,
'level': 32,
'blur_int': None,
'blur_def': None
}, {
'iter': 4,
'level': 16,
'blur_int': None,
'blur_def': None
}])
cleanup = options.get('cleanup', False)
cleanup_intermediate = options.get('cleanup_intermediate', False)
parameters = options.get('parameters', None)
refine = options.get('refine', False)
qc = options.get('qc', False)
downsample = options.get('downsample', None)
start_level = options.get('start_level', None)
debug = options.get('debug', False)
debias = options.get('debias', True)
nl_mode = options.get('nl_mode', 'animal')
if parameters is None:
pass
#TODO: make sensible parameters?
int_models = []
def_models = []
int_residuals = []
def_residuals = []
int_residual = None
def_residual = None
prev_def_estimate = None
# go through all the iterations
it = 0
residuals = []
for (i, p) in enumerate(protocol):
blur_int_model = p.get('blur_int', None)
blur_def_model = p.get('blur_def', None)
for j in range(1, p['iter'] + 1):
it += 1
_start_level = None
if it == 1:
_start_level = start_level
# this will be a model for next iteration actually
it_prefix = prefix + os.sep + str(it)
if not os.path.exists(it_prefix):
os.makedirs(it_prefix)
next_int_model = MriDatasetRegress(prefix=prefix,
name='model_int',
iter=it,
N=N_int,
nomask=nomask)
next_def_model = MriDatasetRegress(prefix=prefix,
name='model_def',
iter=it,
N=N_def,
nomask=True)
print("next_int_model={}".format(
next_int_model.volume[0].rsplit('_0.mnc', 1)[0] +
'_RMS.mnc'))
int_residual = MriDataset(
prefix=prefix,
scan=next_int_model.volume[0].rsplit('_0.mnc', 1)[0] +
'_RMS.mnc')
#name=next_int_model.name, iter=it )
def_residual = MriDataset(
prefix=prefix,
scan=next_def_model.volume[0].rsplit('_0.mnc', 1)[0] +
'_RMS.mnc')
#name=next_def_model.name, iter=it )
# skip over existing models here!
if not next_int_model.exists() or \
not next_def_model.exists() or \
not int_residual.exists() or \
not def_residual.exists():
int_estimate = []
def_estimate = []
r = []
# 1 for each sample generate current approximation
# 2. perform non-linear registration between each sample and sample-specific approximation
# 3. update transformation
# 1+2+3 - all together
for (i, s) in enumerate(samples):
sample_def = MriTransform(name=s.name,
prefix=it_prefix,
iter=it)
sample_int = MriDataset(name=s.name,
prefix=it_prefix,
iter=it)
previous_def = None
if refine and it > 1:
previous_def = prev_def_estimate[i]
r.append(
futures.submit(
non_linear_register_step_regress_std,
s,
current_int_model,
current_def_model,
None,
sample_def,
parameters=parameters,
level=p['level'],
start_level=_start_level,
work_dir=prefix,
downsample=downsample,
debug=debug,
previous_def=previous_def,
nl_mode=nl_mode))
def_estimate.append(sample_def)
#int_estimate.append(sample_int)
# wait for jobs to finish
futures.wait(r, return_when=futures.ALL_COMPLETED)
avg_inv_transform = None
if debias:
# here all the transforms should exist
avg_inv_transform = MriTransform(name='avg_inv',
prefix=it_prefix,
iter=it)
# 2 average all transformations
average_transforms(def_estimate,
avg_inv_transform,
symmetric=False,
invert=True,
nl=True)
corr = []
corr_transforms = []
corr_samples = []
# 3 concatenate correction and resample
for (i, s) in enumerate(samples):
c = MriDataset(prefix=it_prefix, iter=it, name=s.name)
x = MriTransform(name=s.name + '_corr',
prefix=it_prefix,
iter=it)
corr.append(
futures.submit(concat_resample_nl,
s,
def_estimate[i],
avg_inv_transform,
c,
x,
current_int_model,
p['level'],
symmetric=False,
qc=qc,
invert_transform=True))
corr_transforms.append(x)
corr_samples.append(c)
futures.wait(corr, return_when=futures.ALL_COMPLETED)
# 4. perform regression and create new estimate
# 5. calculate residulas (?)
# 4+5
result = futures.submit(voxel_regression,
int_design_matrix,
def_design_matrix,
corr_samples,
corr_transforms,
next_int_model,
next_def_model,
int_residual,
def_residual,
blur_int_model=blur_int_model,
blur_def_model=blur_def_model,
qc=qc)
futures.wait([result], return_when=futures.ALL_COMPLETED)
# 6. cleanup
if cleanup:
print("Cleaning up iteration: {}".format(it))
for i in def_estimate:
i.cleanup()
for i in corr_samples:
i.cleanup()
if prev_def_estimate is not None:
for i in prev_def_estimate:
i.cleanup()
avg_inv_transform.cleanup()
else:
# files were there, reuse them
print(
"Iteration {} already performed, skipping".format(it))
corr_transforms = []
# this is a hack right now
for (i, s) in enumerate(samples):
x = MriTransform(name=s.name + '_corr',
prefix=it_prefix,
iter=it)
corr_transforms.append(x)
int_models.append(current_int_model)
def_models.append(current_def_model)
int_residuals.append(int_residual)
def_residuals.append(def_residual)
current_int_model = next_int_model
current_def_model = next_def_model
result = futures.submit(average_stats_regression,
current_int_model, current_def_model,
int_residual, def_residual)
residuals.append(result)
regression_results = {
'int_model': current_int_model,
'def_model': current_def_model,
'int_residuals': int_residual.scan,
'def_residuals': def_residual.scan,
}
with open(prefix + os.sep + 'results_{:03d}.json'.format(it),
'w') as f:
json.dump(regression_results, f, indent=1, cls=MRIEncoder)
# save for next iteration
# TODO: regularize?
prev_def_estimate = corr_transforms # have to use adjusted def estimate
# copy output to the destination
futures.wait(residuals, return_when=futures.ALL_COMPLETED)
with open(prefix + os.sep + 'stats.txt', 'w') as f:
for s in residuals:
f.write("{}\n".format(s.result()))
with open(prefix + os.sep + 'results_final.json', 'w') as f:
json.dump(regression_results, f, indent=1, cls=MRIEncoder)
if cleanup_intermediate:
for i in range(len(int_models) - 1):
int_models[i].cleanup()
def_models[i].cleanup()
int_residuals[i].cleanup()
def_residuals[i].cleanup()
# delete unneeded models
#shutil.rmtree(prefix+os.sep+'reg')
return regression_results
except mincError as e:
print "Exception in regress:{}".format(str(e))
traceback.print_exc(file=sys.stdout)
raise
except:
print "Exception in regress:{}".format(sys.exc_info()[0])
traceback.print_exc(file=sys.stdout)
raise
like=target,
transform=xfm)
qc(m.tmp('source.mnc'),
qc_file,
mask=target,
image_cmap='red',
mask_cmap='green',
use_max=True) # ,ialpha=0.5,oalpha=0.5
if __name__ == '__main__':
options = parse_options()
try:
if os.path.exists(options.output) and not options.clobber:
raise mincError('File exists : {}'.format(options.output))
parameters = None
if options.param is not None:
with open(options.param, 'r') as f:
parameters = json.load(f)
if options.method == 'ants':
ipl.ants_registration.non_linear_register_ants2(
options.source,
options.target,
options.output,
source_mask=options.source_mask,
target_mask=options.target_mask,
init_xfm=options.init_xfm,
def non_linear_register_full(source,
target,
output_xfm,
source_mask=None,
target_mask=None,
init_xfm=None,
level=4,
start=32,
parameters=None,
work_dir=None,
downsample=None):
"""perform non-linear registration, multiple levels
Args:
source - name of source minc file
target - name of target minc file
output_xfm - name of output transformation file
source_mask - name of source mask file (optional)
target_mask - name of target mask file (optional)
init_xfm - name of initial transformation file (optional)
parameters - configuration for iterative algorithm dict (optional)
work_dir - working directory (optional) , default create one in temp
start - initial step size, default 32mm
level - final step size, default 4mm
downsample - downsample initial files to this step size, default None
Returns:
resulting XFM file
Raises:
mincError when tool fails
"""
with ipl.minc_tools.mincTools() as minc:
if not minc.checkfiles(inputs=[source, target], outputs=[output_xfm]):
return
if parameters is None:
parameters = {
'cost':
'corrcoeff',
'weight':
1,
'stiffness':
1,
'similarity':
0.3,
'sub_lattice':
6,
'conf': [
{
'step': 32.0,
'blur_fwhm': 16.0,
'iterations': 20,
'blur': 'blur',
},
{
'step': 16.0,
'blur_fwhm': 8.0,
'iterations': 20,
'blur': 'blur',
},
{
'step': 12.0,
'blur_fwhm': 6.0,
'iterations': 20,
'blur': 'blur',
},
{
'step': 8.0,
'blur_fwhm': 4.0,
'iterations': 20,
'blur': 'blur',
},
{
'step': 6.0,
'blur_fwhm': 3.0,
'iterations': 20,
'blur': 'blur',
},
{
'step': 4.0,
'blur_fwhm': 2.0,
'iterations': 10,
'blur': 'blur',
},
{
'step': 2.0,
'blur_fwhm': 1.0,
'iterations': 10,
'blur': 'blur',
},
{
'step': 1.0,
'blur_fwhm': 1.0,
'iterations': 10,
'blur': 'blur',
},
{
'step': 1.0,
'blur_fwhm': 0.5,
'iterations': 10,
'blur': 'blur',
},
{
'step': 0.5,
'blur_fwhm': 0.25,
'iterations': 10,
'blur': 'blur',
},
]
}
prev_xfm = None
prev_grid = None
sources = []
targets = []
if isinstance(source, list):
sources.extend(source)
else:
sources.append(source)
if isinstance(target, list):
targets.extend(target)
else:
targets.append(target)
if len(sources) != len(targets):
raise minc_tools.mincError(
' ** Error: Different number of inputs ')
s_base = os.path.basename(sources[0]).rsplit('.gz',
1)[0].rsplit('.mnc', 1)[0]
t_base = os.path.basename(targets[0]).rsplit('.gz',
1)[0].rsplit('.mnc', 1)[0]
# figure out what to do here:
with ipl.minc_tools.cache_files(work_dir=work_dir,
context='reg') as tmp:
# a fitting we shall go...
(sources_lr, targets_lr, source_mask_lr,
target_mask_lr) = minc.downsample_registration_files(
sources, targets, source_mask, target_mask, downsample)
for (i, c) in enumerate(parameters['conf']):
if c['step'] > start:
continue
elif c['step'] < level:
break
# set up intermediate files
tmp_ = tmp.tmp(s_base + '_' + t_base + '_' + str(i))
tmp_xfm = tmp_ + '.xfm'
tmp_grid = tmp_ + '_grid_0.mnc'
tmp_sources = sources_lr
tmp_targets = targets_lr
if c['blur_fwhm'] > 0:
tmp_sources = []
tmp_targets = []
for s_, _ in enumerate(sources_lr):
tmp_source = tmp.cache(s_base + '_' + c['blur'] + '_' +
str(c['blur_fwhm']) + '_' +
str(s_) + '.mnc')
if not os.path.exists(tmp_source):
minc.blur(sources_lr[s_],
tmp_source,
gmag=(c['blur'] == 'dxyz'),
fwhm=c['blur_fwhm'])
tmp_target = tmp.cache(t_base + '_' + c['blur'] + '_' +
str(c['blur_fwhm']) + '_' +
str(s_) + '.mnc')
if not os.path.exists(tmp_target):
minc.blur(targets_lr[s_],
tmp_target,
gmag=(c['blur'] == 'dxyz'),
fwhm=c['blur_fwhm'])
tmp_sources.append(tmp_source)
tmp_targets.append(tmp_target)
# set up registration
args = [
'minctracc',
tmp_sources[0],
tmp_targets[0],
'-clobber',
'-nonlinear',
parameters['cost'],
'-weight',
parameters['weight'],
'-stiffness',
parameters['stiffness'],
'-similarity',
parameters['similarity'],
'-sub_lattice',
parameters['sub_lattice'],
]
args.extend(['-iterations', c['iterations']])
args.extend([
'-lattice_diam', c['step'] * 3.0, c['step'] * 3.0,
c['step'] * 3.0
])
args.extend(['-step', c['step'], c['step'], c['step']])
if c['step'] < 4: #TODO: check if it's 4*minc_step ?
args.append('-no_super')
for s_ in range(len(tmp_targets) - 1):
args.extend([
'-feature_vol', tmp_sources[s_ + 1],
tmp_targets[s_ + 1], parameters['cost'], 1.0
])
# Current transformation at this step
if prev_xfm is not None:
args.extend(['-transformation', prev_xfm])
elif init_xfm is not None:
args.extend(['-transformation', init_xfm])
else:
args.append('-identity')
# masks (even if the blurred image is masked, it's still preferable
# to use the mask in minctracc)
if source_mask is not None:
args.extend(['-source_mask', source_mask_lr])
if target_mask is not None:
args.extend(['-model_mask', target_mask_lr])
# add files and run registration
args.append(tmp_xfm)
minc.command([str(ii) for ii in args],
inputs=[tmp_source, tmp_target],
outputs=[tmp_xfm])
prev_xfm = tmp_xfm
prev_grid = tmp_grid
# done
if prev_xfm is None:
raise ipl.minc_tools.mincError("No iterations were performed!")
# STOP-gap measure to save space for now
# TODO: fix minctracc?
# TODO: fix mincreshape too!
minc.calc([prev_grid],
'A[0]',
tmp.tmp('final_grid_0.mnc'),
datatype='-float')
shutil.move(tmp.tmp('final_grid_0.mnc'), prev_grid)
minc.param2xfm(tmp.tmp('identity.xfm'))
minc.xfmconcat([tmp.tmp('identity.xfm'), prev_xfm], output_xfm)
return output_xfm
def linear_register(source,
target,
output_xfm,
parameters=None,
source_mask=None,
target_mask=None,
init_xfm=None,
objective=None,
conf=None,
debug=False,
close=False,
norot=False,
noshear=False,
noshift=False,
noscale=False,
work_dir=None,
start=None,
downsample=None,
verbose=0):
"""Perform linear registration, replacement for bestlinreg.pl script
Args:
source - name of source minc file
target - name of target minc file
output_xfm - name of output transformation file
parameters - registration parameters (optional), can be
'-lsq6', '-lsq9', '-lsq12'
source_mask - name of source mask file (optional)
target_mask - name of target mask file (optional)
init_xfm - name of initial transformation file (optional)
objective - name of objective function (optional), could be
'-xcorr' (default), '-nmi','-mi'
conf - configuration for iterative algorithm (optional)
array of dict, or a string describing a flawor
bestlinreg (default)
bestlinreg_s
bestlinreg_s2
bestlinreg_new
bestlinreg_20180117 - based on Claude's latest and greatest, circa 2018-01-17
debug - debug flag (optional) , default False
close - closeness flag (optional) , default False
norot - disable rotation flag (optional) , default False
noshear - disable shear flag (optional) , default False
noshift - disable shift flag (optional) , default False
noscale - disable scale flag (optional) , default False
work_dir - working directory (optional) , default create one in temp
start - initial blurring level, default 16mm from configuration
downsample - downsample initial files to this step size, default None
verbose - verbosity level
Returns:
resulting XFM file
Raises:
mincError when tool fails
"""
logger.debug("linear_register s:{} s_m:{} t:{} t_m:{} i:{} ".format(
source, source_mask, target, target_mask, init_xfm))
with ipl.minc_tools.mincTools(verbose=verbose) as minc:
if not minc.checkfiles(inputs=[source, target], outputs=[output_xfm]):
return
sources = []
targets = []
if isinstance(source, list):
sources.extend(source)
else:
sources.append(source)
if isinstance(target, list):
targets.extend(target)
else:
targets.append(target)
if len(sources) != len(targets):
raise minc_tools.mincError(
' ** Error: Different number of inputs ')
# python version
if conf is None:
conf = linear_registration_config[
'bestlinreg_20180117'] # bestlinreg_new ?
elif not isinstance(conf, list): # assume that it is a string
if conf in linear_registration_config:
conf = linear_registration_config[conf]
if parameters is None:
parameters = '-lsq9'
if objective is None:
objective = '-xcorr'
if not isinstance(conf, list): # assume that it is a string
# assume it's external program's name
# else run internally
# TODO: check if we are given multiple sources/targets?
#
with ipl.minc_tools.mincTools() as m:
cmd = [conf, source, target, output_xfm]
if source_mask is not None:
cmd.extend(['-source_mask', source_mask])
if target_mask is not None:
cmd.extend(['-target_mask', target_mask])
if parameters is not None:
cmd.append(parameters)
if objective is not None:
cmd.append(objective)
if init_xfm is not None:
cmd.extend(['-init_xfm', init_xfm])
m.command(cmd,
inputs=[source, target],
outputs=[output_xfm],
verbose=2)
return output_xfm
else:
prev_xfm = None
s_base = os.path.basename(sources[0]).rsplit('.gz', 1)[0].rsplit(
'.mnc', 1)[0]
t_base = os.path.basename(targets[0]).rsplit('.gz', 1)[0].rsplit(
'.mnc', 1)[0]
# figure out what to do here:
with ipl.minc_tools.cache_files(work_dir=work_dir,
context='reg') as tmp:
(sources_lr, targets_lr, source_mask_lr,
target_mask_lr) = minc.downsample_registration_files(
sources, targets, source_mask, target_mask, downsample)
# a fitting we shall go...
for (i, c) in enumerate(conf):
_parameters = parameters
if 'parameters' in c and parameters != '-lsq6': # emulate Claude's approach
_parameters = c.get('parameters') #'-lsq7'
_reverse = c.get('reverse',
False) # swap target and source
# set up intermediate files
if start is not None and start > c['blur_fwhm']:
continue
elif close and c['blur_fwhm'] > 8:
continue
tmp_xfm = tmp.tmp(s_base + '_' + t_base + '_' + str(i) +
'.xfm')
tmp_sources = sources_lr
tmp_targets = targets_lr
if c['blur_fwhm'] > 0:
tmp_sources = []
tmp_targets = []
for s_, _ in enumerate(sources_lr):
tmp_source = tmp.cache(s_base + '_' + c['blur'] +
'_' + str(c['blur_fwhm']) +
'_' + str(s_) + '.mnc')
if not os.path.exists(tmp_source):
minc.blur(sources_lr[s_],
tmp_source,
gmag=(c['blur'] == 'dxyz'),
fwhm=c['blur_fwhm'])
tmp_target = tmp.cache(t_base + '_' + c['blur'] +
'_' + str(c['blur_fwhm']) +
'_' + str(s_) + '.mnc')
if not os.path.exists(tmp_target):
minc.blur(targets_lr[s_],
tmp_target,
gmag=(c['blur'] == 'dxyz'),
fwhm=c['blur_fwhm'])
tmp_sources.append(tmp_source)
tmp_targets.append(tmp_target)
objective_ = objective
if isinstance(objective, list):
objective_ = objective[0]
if _reverse:
args = [
'minctracc', tmp_targets[0], tmp_sources[0],
'-clobber', _parameters, objective_, '-simplex',
c['simplex'], '-tol', c['tolerance']
]
# additional modalities
for s_ in range(len(tmp_targets) - 1):
if isinstance(objective, list):
objective_ = objective[s_ + 1]
args.extend([
'-feature_vol', tmp_targets[s_ + 1],
tmp_sources[s_ + 1],
objective_.lstrip('-'), 1.0
])
else:
# set up registration
args = [
'minctracc', tmp_sources[0], tmp_targets[0],
'-clobber', _parameters, objective_, '-simplex',
c['simplex'], '-tol', c['tolerance']
]
for s_ in range(len(tmp_targets) - 1):
if isinstance(objective, list):
objective_ = objective[s_ + 1]
args.extend([
'-feature_vol', tmp_sources[s_ + 1],
tmp_targets[s_ + 1],
objective_.lstrip('-'), 1.0
])
args.append('-step')
args.extend(c['steps'])
# Current transformation at this step
if prev_xfm is not None:
if _reverse:
inv_prev_xfm = tmp.tmp(s_base + '_' + t_base +
'_' + str(i) + '_init.xfm')
minc.xfminvert(prev_xfm, inv_prev_xfm)
args.extend(['-transformation', inv_prev_xfm])
else:
args.extend(['-transformation', prev_xfm])
elif init_xfm is not None:
# _reverse should not be first?
args.extend(
['-transformation', init_xfm, '-est_center'])
elif close:
args.append('-identity')
else:
# _reverse should not be first?
# Initial transformation will be computed from the from Principal axis
# transformation (PAT).
if c['trans'] is not None and c['trans'][
0] != '-est_translations':
args.extend(c['trans'])
else:
# will use manual transformation based on shif of CoM, should be identical to '-est_translations' , but it's not
com_src = minc.stats(source,
['-com', '-world_only'],
single_value=False)
com_trg = minc.stats(target,
['-com', '-world_only'],
single_value=False)
diff = [com_trg[k] - com_src[k] for k in range(3)]
xfm = tmp.cache(s_base + '_init.xfm')
minc.param2xfm(xfm, translation=diff)
args.extend(['-transformation', xfm])
# masks (even if the blurred image is masked, it's still preferable
# to use the mask in minctracc)
if _reverse:
if source_mask is not None:
args.extend(['-model_mask', source_mask_lr])
#disable one mask in this mode
#if target_mask is not None:
#args.extend(['-source_mask', target_mask_lr])
else:
if source_mask is not None:
args.extend(['-source_mask', source_mask_lr])
if target_mask is not None:
args.extend(['-model_mask', target_mask_lr])
if noshear:
args.extend(['-w_shear', 0, 0, 0])
if noscale:
args.extend(['-w_scales', 0, 0, 0])
if noshift:
args.extend(['-w_translations', 0, 0, 0])
if norot:
args.extend(['-w_rotations', 0, 0, 0])
# add files and run registration
args.append(tmp_xfm)
minc.command([str(ii) for ii in args],
inputs=[tmp_source, tmp_target],
outputs=[tmp_xfm])
if _reverse:
inv_tmp_xfm = tmp.tmp(s_base + '_' + t_base + '_' +
str(i) + '_sol.xfm')
minc.xfminvert(tmp_xfm, inv_tmp_xfm)
prev_xfm = inv_tmp_xfm
else:
prev_xfm = tmp_xfm
shutil.copyfile(prev_xfm, output_xfm)
return output_xfm
