SHM1 = ozm.SphericalHarmonicsModel(
dwi1,
bvecs1,
bvals1,
mask=brain_mask,
model_coeffs=csd_coeffs[0],
response_file=response_files[0],
#axial_diffusivity=AD1,
#radial_diffusivity=RD1
)
SHM2 = ozm.SphericalHarmonicsModel(
dwi2,
bvecs2,
bvals2,
#mask = mask_array,
mask=brain_mask,
model_coeffs=csd_coeffs[1],
response_file=response_files[1],
#axial_diffusivity=AD2,
#radial_diffusivity=RD2
)
rmse_file_name = '%s%s_relative_rmse_b%s.nii.gz' % (
data_path, 'SphericalHarmonicsModel', bval)
if not os.path.isfile(rmse_file_name):
relative_rmse = ozm.relative_rmse(SHM1, SHM2)
io.nii_from_volume(relative_rmse, rmse_file_name,
ni.load(dwi1).get_affine())
mask=brain_mask)
SphereModel1 = ozm.SphereModel(dwi1, bvecs1, bvals1, mask=brain_mask)
SphereModel2 = ozm.SphereModel(dwi2, bvecs2, bvals2, mask=brain_mask)
ModelFest = zip([
TensorModel1, CanonicalTensorModel1, MultiCanonicalTensorModel1,
SparseDeconvolutionModel1, SphereModel1,
PointyMultiCanonicalTensorModel1, PointyCanonicalTensorModel1
], [
TensorModel2, CanonicalTensorModel2, MultiCanonicalTensorModel2,
SparseDeconvolutionModel2, SphereModel2,
PointyMultiCanonicalTensorModel2, PointyCanonicalTensorModel2
], [
'TensorModel', 'CanonicalTensorModel', 'MultiCanonicalTensorModel',
'SparseDeconvolutionModel', 'SphereModel',
'PointyMultiCanonicalTensorModel', 'PointyCanonicalTensorModel'
])
# Compute the relative RMSE for each model:
relative_rmse = []
for Model1, Model2, model_name in ModelFest:
rmse_file_name = '%s%s_relative_rmse_b%s.nii.gz' % (data_path,
model_name, bval)
# Only do this if you have to:
if not os.path.isfile(rmse_file_name):
relative_rmse = ozm.relative_rmse(Model1, Model2)
io.nii_from_volume(relative_rmse, rmse_file_name,
ni.load(dwi1).get_affine())
csd_coeffs.append(d + "/" + f.split(".")[0] + "_CSD.nii.gz")
response_files.append(d + "/" + f.split(".")[0] + "_ER.mif")
SHM1 = ozm.SphericalHarmonicsModel(
dwi1,
bvecs1,
bvals1,
mask=brain_mask,
model_coeffs=csd_coeffs[0],
response_file=response_files[0],
# axial_diffusivity=AD1,
# radial_diffusivity=RD1
)
SHM2 = ozm.SphericalHarmonicsModel(
dwi2,
bvecs2,
bvals2,
# mask = mask_array,
mask=brain_mask,
model_coeffs=csd_coeffs[1],
response_file=response_files[1],
# axial_diffusivity=AD2,
# radial_diffusivity=RD2
)
rmse_file_name = "%s%s_relative_rmse_b%s.nii.gz" % (data_path, "SphericalHarmonicsModel", bval)
if not os.path.isfile(rmse_file_name):
relative_rmse = ozm.relative_rmse(SHM1, SHM2)
io.nii_from_volume(relative_rmse, rmse_file_name, ni.load(dwi1).get_affine())
ModelFest = zip([TensorModel1,CanonicalTensorModel1,
MultiCanonicalTensorModel1,SparseDeconvolutionModel1,
SphereModel1,PointyMultiCanonicalTensorModel1,
PointyCanonicalTensorModel1],
[TensorModel2,CanonicalTensorModel2,
MultiCanonicalTensorModel2,SparseDeconvolutionModel2,
SphereModel2,PointyMultiCanonicalTensorModel2,
PointyCanonicalTensorModel2],
['TensorModel', 'CanonicalTensorModel',
'MultiCanonicalTensorModel','SparseDeconvolutionModel',
'SphereModel', 'PointyMultiCanonicalTensorModel',
'PointyCanonicalTensorModel'
])
# Compute the relative RMSE for each model:
relative_rmse = []
for Model1,Model2,model_name in ModelFest:
rmse_file_name = '%s%s_relative_rmse_b%s.nii.gz'%(data_path,
model_name,
bval)
# Only do this if you have to:
if not os.path.isfile(rmse_file_name):
relative_rmse = ozm.relative_rmse(Model1, Model2)
io.nii_from_volume(relative_rmse,
rmse_file_name,
ni.load(dwi1).get_affine())
