(BSplineInterpolationOrder 1)
(FinalBSplineInterpolationOrder 1)
(DefaultPixelValue 0)
(WriteResultImage "false")
// The pixel type and format of the resulting deformed moving image
(ResultImagePixelType "float")
(ResultImageFormat "mnc")
"""
if __name__ == '__main__':
with mincTools() as minc:
minc.register_elastix("data/ellipse_0_blur.mnc",
"data/ellipse_1_blur.mnc",
output_par="test_4mm_0_1_par.txt",
output_xfm="test_4mm_0_1.xfm",
parameters=elx_par1)
minc.grid_magnitude("test_4mm_0_1_grid_0.mnc",
"test_4mm_0_1_grid_m.mnc")
minc.register_elastix("data/ellipse_0_blur.mnc",
"data/ellipse_1_blur.mnc",
output_par="test_4mm_0_1_2_par.txt",
output_xfm="test_4mm_0_1_2.xfm",
parameters=elx_par1,
init_xfm="test_4mm_0_1.xfm")
def generate_xfm_direct_minctracc(i,
j,
mri1,
mri2,
mask1,
mask2,
seg1,
seg2,
output_base,
step=2,
baa=False):
with mincTools(verbose=2) as minc:
# normalize xfms
minc.non_linear_register_full(mri1,
mri2,
output_base + '_map.xfm',
level=step,
source_mask=mask1,
target_mask=mask2)
# resample mris
minc.resample_smooth(mri2,
minc.tmp('mri2.mnc'),
transform=output_base + '_map.xfm',
invert_transform=True)
# resample segs
minc.resample_labels(seg2,
minc.tmp('seg2.mnc'),
transform=output_base + '_map.xfm',
invert_transform=True,
baa=baa)
minc.resample_labels(mask2,
minc.tmp('mask2.mnc'),
transform=output_base + '_map.xfm',
invert_transform=True,
baa=baa)
# calculate CC, MI
cc = minc.similarity(mri1,
minc.tmp('mri2.mnc'),
ref_mask=mask1,
sample_mask=minc.tmp('mask2.mnc'),
method='cc')
nmi = minc.similarity(mri1,
minc.tmp('mri2.mnc'),
ref_mask=mask1,
sample_mask=minc.tmp('mask2.mnc'),
method='nmi')
ncc = minc.similarity(mri1,
minc.tmp('mri2.mnc'),
ref_mask=mask1,
sample_mask=minc.tmp('mask2.mnc'),
method='ncc')
msq = minc.similarity(mri1,
minc.tmp('mri2.mnc'),
ref_mask=mask1,
sample_mask=minc.tmp('mask2.mnc'),
method='msq')
# calculate label overlap
gtc = minc.label_similarity(seg1, minc.tmp('seg2.mnc'), method='gtc')
# write out result
with open(output_base + '_similarity.txt', 'w') as f:
f.write("{},{},{},{},{},{},{}\n".format(i, j, cc, ncc, nmi, msq,
gtc))
def generate_xfm_direct_elastix_mi(i,
j,
mri1,
mri2,
mask1,
mask2,
seg1,
seg2,
output_base,
baa=False,
step=2):
with mincTools(verbose=2) as minc:
param_mi = """
(FixedInternalImagePixelType "float")
(MovingInternalImagePixelType "float")
(FixedImageDimension 3)
(MovingImageDimension 3)
(UseDirectionCosines "true")
(ShowExactMetricValue "false")
(Registration "MultiResolutionRegistration")
(Interpolator "BSplineInterpolator" )
(ResampleInterpolator "FinalBSplineInterpolator" )
(Resampler "DefaultResampler" )
(FixedImagePyramid "FixedSmoothingImagePyramid")
(MovingImagePyramid "MovingSmoothingImagePyramid")
(Optimizer "AdaptiveStochasticGradientDescent")
(Transform "BSplineTransform")
(Metric "AdvancedMattesMutualInformation")
(FinalGridSpacingInPhysicalUnits 4)
(HowToCombineTransforms "Compose")
(ErodeMask "false")
(NumberOfResolutions 3)
(ImagePyramidSchedule 8 8 8 4 4 4 2 2 2)
(MaximumNumberOfIterations 2000 2000 2000 )
(MaximumNumberOfSamplingAttempts 3)
(NumberOfSpatialSamples 1024 1024 4096 )
(NewSamplesEveryIteration "true")
(ImageSampler "Random" )
(BSplineInterpolationOrder 1)
(FinalBSplineInterpolationOrder 3)
(DefaultPixelValue 0)
(WriteResultImage "false")
// The pixel type and format of the resulting deformed moving image
(ResultImagePixelType "float")
(ResultImageFormat "mnc")
"""
# normalize xfms
minc.register_elastix(mri1,
mri2,
output_xfm=minc.tmp('transform.xfm'),
source_mask=mask1,
target_mask=mask2,
parameters=param_mi)
minc.xfm_normalize(minc.tmp('transform.xfm'),
mri1,
output_base + '_map.xfm',
step=step)
# resample mris
minc.resample_smooth(mri2,
minc.tmp('mri2.mnc'),
transform=output_base + '_map.xfm',
invert_transform=True)
# resample segs
minc.resample_labels(seg2,
minc.tmp('seg2.mnc'),
transform=output_base + '_map.xfm',
invert_transform=True,
baa=baa)
minc.resample_labels(mask2,
minc.tmp('mask2.mnc'),
transform=output_base + '_map.xfm',
invert_transform=True,
baa=baa)
# calculate CC, MI
cc = minc.similarity(mri1,
minc.tmp('mri2.mnc'),
ref_mask=mask1,
sample_mask=minc.tmp('mask2.mnc'),
method='cc')
nmi = minc.similarity(mri1,
minc.tmp('mri2.mnc'),
ref_mask=mask1,
sample_mask=minc.tmp('mask2.mnc'),
method='nmi')
ncc = minc.similarity(mri1,
minc.tmp('mri2.mnc'),
ref_mask=mask1,
sample_mask=minc.tmp('mask2.mnc'),
method='ncc')
msq = minc.similarity(mri1,
minc.tmp('mri2.mnc'),
ref_mask=mask1,
sample_mask=minc.tmp('mask2.mnc'),
method='msq')
# calculate label overlap
gtc = minc.label_similarity(seg1, minc.tmp('seg2.mnc'), method='gtc')
# write out result
with open(output_base + '_similarity.txt', 'w') as f:
f.write("{},{},{},{},{},{},{}\n".format(i, j, cc, ncc, nmi, msq,
gtc))
def generate_xfm_direct_ANTS_MI(i,
j,
mri1,
mri2,
mask1,
mask2,
seg1,
seg2,
output_base,
baa=False,
step=2):
with mincTools(verbose=2) as minc:
# normalize xfms
param_mi = {
'cost_function': 'MI',
'iter': '40x40x40x00',
'cost_function_par': '1,32'
}
minc.non_linear_register_ants(mri1,
mri2,
minc.tmp('transform.xfm'),
target_mask=mask2,
parameters=param_mi)
minc.xfm_normalize(minc.tmp('transform.xfm'),
mri1,
output_base + '_map.xfm',
step=step)
# resample mris
minc.resample_smooth(mri2,
minc.tmp('mri2.mnc'),
transform=output_base + '_map.xfm',
invert_transform=True)
# resample segs
minc.resample_labels(seg2,
minc.tmp('seg2.mnc'),
transform=output_base + '_map.xfm',
invert_transform=True,
baa=baa)
minc.resample_labels(mask2,
minc.tmp('mask2.mnc'),
transform=output_base + '_map.xfm',
invert_transform=True,
baa=baa)
# calculate CC, MI
cc = minc.similarity(mri1,
minc.tmp('mri2.mnc'),
ref_mask=mask1,
sample_mask=minc.tmp('mask2.mnc'),
method='cc')
nmi = minc.similarity(mri1,
minc.tmp('mri2.mnc'),
ref_mask=mask1,
sample_mask=minc.tmp('mask2.mnc'),
method='nmi')
ncc = minc.similarity(mri1,
minc.tmp('mri2.mnc'),
ref_mask=mask1,
sample_mask=minc.tmp('mask2.mnc'),
method='ncc')
msq = minc.similarity(mri1,
minc.tmp('mri2.mnc'),
ref_mask=mask1,
sample_mask=minc.tmp('mask2.mnc'),
method='msq')
# calculate label overlap
gtc = minc.label_similarity(seg1, minc.tmp('seg2.mnc'), method='gtc')
# write out result
with open(output_base + '_similarity.txt', 'w') as f:
f.write("{},{},{},{},{},{},{}\n".format(i, j, cc, ncc, nmi, msq,
gtc))
def generate_xfm_model(i,
j,
xfm1,
xfm2,
mri1,
mri2,
mask1,
mask2,
seg1,
seg2,
output_base,
step=2,
baa=False):
with mincTools(verbose=2) as minc:
# all xfms are mapping subject to common space, so to map one subject to another it will be xfm1 * xfm2^1
minc.xfminvert(xfm2, minc.tmp('xfm1.xfm'))
# concatenate xfms
minc.xfmconcat([xfm1, minc.tmp('xfm1.xfm')],
minc.tmp('xfm1_dot_xfm2_inv.xfm'))
# normalize xfms
minc.xfm_normalize(minc.tmp('xfm1_dot_xfm2_inv.xfm'),
mri1,
output_base + '_map.xfm',
step=step)
# resample mris
minc.resample_smooth(mri2,
minc.tmp('mri2.mnc'),
transform=output_base + '_map.xfm',
invert_transform=True)
# resample segs
minc.resample_labels(seg2,
minc.tmp('seg2.mnc'),
transform=output_base + '_map.xfm',
invert_transform=True,
baa=baa)
minc.resample_labels(mask2,
minc.tmp('mask2.mnc'),
transform=output_base + '_map.xfm',
invert_transform=True,
baa=baa)
# calculate CC, MI
cc = minc.similarity(mri1,
minc.tmp('mri2.mnc'),
ref_mask=mask1,
sample_mask=minc.tmp('mask2.mnc'),
method='cc')
nmi = minc.similarity(mri1,
minc.tmp('mri2.mnc'),
ref_mask=mask1,
sample_mask=minc.tmp('mask2.mnc'),
method='nmi')
ncc = minc.similarity(mri1,
minc.tmp('mri2.mnc'),
ref_mask=mask1,
sample_mask=minc.tmp('mask2.mnc'),
method='ncc')
msq = minc.similarity(mri1,
minc.tmp('mri2.mnc'),
ref_mask=mask1,
sample_mask=minc.tmp('mask2.mnc'),
method='msq')
# calculate label overlap
gtc = minc.label_similarity(seg1, minc.tmp('seg2.mnc'), method='gtc')
# write out result
with open(output_base + '_similarity.txt', 'w') as f:
f.write("{},{},{},{},{},{},{}\n".format(i, j, cc, ncc, nmi, msq,
gtc))
#! /usr/bin/env python
from iplMincTools import mincTools
if __name__ == '__main__':
m = mincTools()
m.non_linear_register_increment('ellipse_1.mnc',
'ellipse_2.mnc',
'test.xfm',
source_mask='mask.mnc',
target_mask='mask.mnc',
level=8)