def test_fwdti_errors():
# 1st error - if a unknown fit method is given to the FWTM
assert_raises(ValueError,
fwdti.FreeWaterTensorModel,
gtab_2s,
fit_method="pKT")
# 2nd error - if incorrect mask is given
fwdtiM = fwdti.FreeWaterTensorModel(gtab_2s)
incorrect_mask = np.array([[True, True, False], [True, False, False]])
assert_raises(ValueError, fwdtiM.fit, DWI, mask=incorrect_mask)
# 3rd error - if data with only one non zero b-value is given
assert_raises(ValueError, fwdti.FreeWaterTensorModel, gtab)
# Testing the correct usage
fwdtiM = fwdti.FreeWaterTensorModel(gtab_2s, min_signal=1)
correct_mask = np.zeros((2, 2, 2))
correct_mask[0, :, :] = 1
correct_mask = correct_mask > 0
fwdtiF = fwdtiM.fit(DWI, mask=correct_mask)
assert_array_almost_equal(fwdtiF.fa, FAref)
assert_array_almost_equal(fwdtiF.f, GTF)
# 4th error - if a sigma is selected by no value of sigma is given
fwdm = fwdti.FreeWaterTensorModel(gtab_2s, 'NLS', weighting='sigma')
assert_raises(ValueError, fwdm.fit, DWI)
def test_negative_s0():
# single voxel
gtf = 0.55
mevals = np.array([[0.0017, 0.0003, 0.0003], [0.003, 0.003, 0.003]])
S_conta, peaks = multi_tensor(gtab_2s,
mevals,
S0=100,
angles=[(90, 0), (90, 0)],
fractions=[(1 - gtf) * 100, gtf * 100],
snr=None)
S_conta[gtab_2s.bvals == 0] = -100
fwdm = fwdti.FreeWaterTensorModel(gtab_2s, 'NLS')
fwefit = fwdm.fit(S_conta)
assert_array_almost_equal(fwefit.fa, 0.0)
assert_array_almost_equal(fwefit.md, 0.0)
assert_array_almost_equal(fwefit.f, 0.0)
# multi voxel
DWI[0, 0, 1, gtab_2s.bvals == 0] = -100
GTF[0, 0, 1] = 0
FAref[0, 0, 1] = 0
MDref[0, 0, 1] = 0
fwefit = fwdm.fit(DWI)
assert_array_almost_equal(fwefit.fa, FAref)
assert_array_almost_equal(fwefit.md, MDref)
assert_array_almost_equal(fwefit.f, GTF)
def test_md_regularization():
# single voxel
gtf = 0.97 # for this ground truth value, md is larger than 2.7e-3
mevals = np.array([[0.0017, 0.0003, 0.0003], [0.003, 0.003, 0.003]])
S_conta, peaks = multi_tensor(gtab_2s,
mevals,
S0=100,
angles=[(90, 0), (90, 0)],
fractions=[(1 - gtf) * 100, gtf * 100],
snr=None)
fwdm = fwdti.FreeWaterTensorModel(gtab_2s, 'NLS')
fwefit = fwdm.fit(S_conta)
assert_array_almost_equal(fwefit.fa, 0.0)
assert_array_almost_equal(fwefit.md, 0.0)
assert_array_almost_equal(fwefit.f, 1.0)
# multi voxel
DWI[0, 1, 1] = S_conta
GTF[0, 1, 1] = 1
FAref[0, 1, 1] = 0
MDref[0, 1, 1] = 0
fwefit = fwdm.fit(DWI)
assert_array_almost_equal(fwefit.fa, FAref)
assert_array_almost_equal(fwefit.md, MDref)
assert_array_almost_equal(fwefit.f, GTF)
def fit(self, data, mask):
self.model = fwdti.FreeWaterTensorModel(self.gradient_table)
self.f = self.model.fit(data, mask=mask)
self.FA = self.f.fa
self.MD = self.f.md
return self
def test_fwdti_predictions():
# single voxel case
gtf = 0.50 # ground truth volume fraction
angles = [(90, 0), (90, 0)]
mevals = np.array([[0.0017, 0.0003, 0.0003], [0.003, 0.003, 0.003]])
S_conta, peaks = multi_tensor(gtab_2s,
mevals,
S0=100,
angles=angles,
fractions=[(1 - gtf) * 100, gtf * 100],
snr=None)
R = all_tensor_evecs(peaks[0])
R = R.reshape((9))
model_params = np.concatenate(([0.0017, 0.0003, 0.0003], R, [gtf]), axis=0)
S_pred1 = fwdti_prediction(model_params, gtab_2s, S0=100)
assert_array_almost_equal(S_pred1, S_conta)
# Testing in model class
fwdm = fwdti.FreeWaterTensorModel(gtab_2s)
S_pred2 = fwdm.predict(model_params, S0=100)
assert_array_almost_equal(S_pred2, S_conta)
# Testing in fit class
fwefit = fwdm.fit(S_conta)
S_pred3 = fwefit.predict(gtab_2s, S0=100)
assert_array_almost_equal(S_pred3, S_conta, decimal=5)
# Multi voxel simulation
S_pred1 = fwdti_prediction(model_params_mv, gtab_2s, S0=100) # function
assert_array_almost_equal(S_pred1, DWI)
S_pred2 = fwdm.predict(model_params_mv, S0=100) # Model class
assert_array_almost_equal(S_pred2, DWI)
fwefit = fwdm.fit(DWI) # Fit class
S_pred3 = fwefit.predict(gtab_2s, S0=100)
assert_array_almost_equal(S_pred3, DWI)
def test_fwdti_jac_multi_voxel():
fwdm = fwdti.FreeWaterTensorModel(gtab_2s, 'WLS')
fwdm.fit(DWI[0, :, :])
# no f transform
fwdm = fwdti.FreeWaterTensorModel(gtab_2s, 'NLS', f_transform=False,
jac=True)
fwefit = fwdm.fit(DWI[0, :, :])
Ffwe = fwefit.f
assert_array_almost_equal(Ffwe, GTF[0, :])
# with f transform
fwdm = fwdti.FreeWaterTensorModel(gtab_2s, 'NLS', f_transform=True,
jac=True)
fwefit = fwdm.fit(DWI[0, :, :])
Ffwe = fwefit.f
assert_array_almost_equal(Ffwe, GTF[0, :])
def test_fwdti_restore():
# Restore has to work well even in nonproblematic cases
# Simulate a signal corrupted by free water diffusion contamination
gtf = 0.50 # ground truth volume fraction
mevals = np.array([[0.0017, 0.0003, 0.0003], [0.003, 0.003, 0.003]])
S_conta, peaks = multi_tensor(gtab_2s, mevals, S0=100,
angles=[(90, 0), (90, 0)],
fractions=[(1-gtf) * 100, gtf*100], snr=None)
fwdm = fwdti.FreeWaterTensorModel(gtab_2s, 'NLS', weighting='sigma',
sigma=4)
fwdtiF = fwdm.fit(S_conta)
assert_array_almost_equal(fwdtiF.fa, FAdti)
assert_array_almost_equal(fwdtiF.f, gtf)
fwdm2 = fwdti.FreeWaterTensorModel(gtab_2s, 'NLS', weighting='gmm')
fwdtiF2 = fwdm2.fit(S_conta)
assert_array_almost_equal(fwdtiF2.fa, FAdti)
assert_array_almost_equal(fwdtiF2.f, gtf)
def test_fwdti_singlevoxel():
# Simulation when water contamination is added
gtf = 0.44444 # ground truth volume fraction
mevals = np.array([[0.0017, 0.0003, 0.0003], [0.003, 0.003, 0.003]])
S_conta, peaks = multi_tensor(gtab_2s,
mevals,
S0=100,
angles=[(90, 0), (90, 0)],
fractions=[(1 - gtf) * 100, gtf * 100],
snr=None)
fwdm = fwdti.FreeWaterTensorModel(gtab_2s, 'WLS')
fwefit = fwdm.fit(S_conta)
FAfwe = fwefit.fa
Ffwe = fwefit.f
MDfwe = fwefit.md
assert_almost_equal(FAdti, FAfwe, decimal=3)
assert_almost_equal(Ffwe, gtf, decimal=3)
assert_almost_equal(MDfwe, MDdti, decimal=3)
# Test non-linear fit
fwdm = fwdti.FreeWaterTensorModel(gtab_2s, 'NLS', cholesky=False)
fwefit = fwdm.fit(S_conta)
FAfwe = fwefit.fa
Ffwe = fwefit.f
MDfwe = fwefit.md
assert_almost_equal(FAdti, FAfwe)
assert_almost_equal(Ffwe, gtf)
assert_almost_equal(MDfwe, MDdti)
# Test cholesky
fwdm = fwdti.FreeWaterTensorModel(gtab_2s, 'NLS', cholesky=True)
fwefit = fwdm.fit(S_conta)
FAfwe = fwefit.fa
Ffwe = fwefit.f
MDfwe = fwefit.md
assert_almost_equal(FAdti, FAfwe)
assert_almost_equal(Ffwe, gtf)
assert_almost_equal(MDfwe, MDfwe)
def test_fwdti_multi_voxel():
fwdm = fwdti.FreeWaterTensorModel(gtab_2s, 'NLS', cholesky=False)
fwefit = fwdm.fit(DWI)
FAfwe = fwefit.fa
Ffwe = fwefit.f
MDfwe = fwefit.md
assert_almost_equal(FAfwe, FAref)
assert_almost_equal(Ffwe, GTF)
assert_almost_equal(MDfwe, MDref)
# Test cholesky
fwdm = fwdti.FreeWaterTensorModel(gtab_2s, 'NLS', cholesky=True)
fwefit = fwdm.fit(DWI)
FAfwe = fwefit.fa
Ffwe = fwefit.f
MDfwe = fwefit.md
assert_almost_equal(FAfwe, FAref)
assert_almost_equal(Ffwe, GTF)
assert_almost_equal(MDfwe, MDref)
def test_fwdti_precision():
# Simulation when water contamination is added
gtf = 0.63416 # ground truth volume fraction
mevals = np.array([[0.0017, 0.0003, 0.0003], [0.003, 0.003, 0.003]])
S_conta, peaks = multi_tensor(gtab_2s, mevals, S0=100,
angles=[(90, 0), (90, 0)],
fractions=[(1-gtf) * 100, gtf*100], snr=None)
fwdm = fwdti.FreeWaterTensorModel(gtab_2s, 'WLS', piterations=5)
fwefit = fwdm.fit(S_conta)
FAfwe = fwefit.fa
Ffwe = fwefit.f
MDfwe = fwefit.md
assert_almost_equal(FAdti, FAfwe, decimal=5)
assert_almost_equal(Ffwe, gtf, decimal=5)
assert_almost_equal(MDfwe, MDdti, decimal=5)
def fit_fwdti(data_files,
bval_files,
bvec_files,
mask=None,
out_dir=None,
b0_threshold=50):
"""
Fit the free water DTI model [1]_, save files with derived maps
Parameters
----------
data_files : str or list
Files containing DWI data. If this is a str, that's the full path to a
single file. If it's a list, each entry is a full path.
bval_files : str or list
Equivalent to `data_files`.
bvec_files : str or list
Equivalent to `data_files`.
mask : ndarray, optional
Binary mask, set to True or 1 in voxels to be processed.
Default: Process all voxels.
out_dir : str, optional
A full path to a directory to store the maps that get computed.
Default: maps get stored in the same directory as the last DWI file
in `data_files`.
b0_threshold : float
Returns
-------
file_paths : a dict with the derived maps that were computed and full-paths
to the files containing these maps.
Note
----
..[1] R. Neto Henriques, A. Rokem, E. Garyfallidis, S. St-Jean, E.T.
Peterson, M. Correia (2017). [Re] Optimization of a free water
elimination two-compartment model for diffusion tensor imaging.
*ReScience*
"""
img, data, gtab, mask = ut.prepare_data(data_files,
bval_files,
bvec_files,
mask=mask,
b0_threshold=b0_threshold)
fwmodel = fwdti.FreeWaterTensorModel(gtab)
fwfit = fwmodel.fit(data, mask=mask)
FA = fwfit.fa
MD = fwfit.md
AD = fwfit.ad
RD = fwfit.rd
fwvf = fwfit.f
params = fwfit.model_params
maps = [FA, MD, AD, RD, fwvf, params]
names = ['FA', 'MD', 'AD', 'RD', 'FWVF', 'params']
if out_dir is None:
if isinstance(data_files, list):
out_dir = op.join(op.split(data_files[0])[0], 'fwdti')
else:
out_dir = op.join(op.split(data_files)[0], 'fwdti')
if not op.exists(out_dir):
os.makedirs(out_dir)
aff = img.affine
file_paths = {}
for m, n in zip(maps, names):
file_paths[n] = op.join(out_dir, 'fwdti_%s.nii.gz' % n)
nib.save(nib.Nifti1Image(m, aff), file_paths[n])
return file_paths
def _fit(gtab, data, mask=None):
fwmodel = fwdti.FreeWaterTensorModel(gtab)
return fwmodel.fit(data, mask=mask)
dilate=1) """ Moreover, for illustration purposes we process only an axial slice of the data. """ axial_slice = 40 mask_roi = np.zeros(data.shape[:-1], dtype=bool) mask_roi[:, :, axial_slice] = mask[:, :, axial_slice] """ The free water elimination model fit can then be initialized by instantiating a FreeWaterTensorModel class object: """ fwdtimodel = fwdti.FreeWaterTensorModel(gtab) """ The data can then be fitted using the ``fit`` function of the defined model object: """ fwdtifit = fwdtimodel.fit(data, mask=mask_roi) """ This 2-steps procedure will create a FreeWaterTensorFit object which contains all the diffusion tensor statistics free for free water contaminations. Below we extract the fractional anisotropy (FA) and the mean diffusivity (MD) of the free water diffusion tensor.""" FA = fwdtifit.fa MD = fwdtifit.md """
def fit_fwdti_dipy(input_dwi,
input_bval,
input_bvec,
output_dir,
fit_method='',
mask=''):
import dipy.reconst.fwdti as fwdti
if not os.path.exists(output_dir):
os.makedirs(output_dir)
if fit_method == '':
fit_method = 'WLS'
img = nib.load(input_dwi)
data = img.get_data()
bvals, bvecs = read_bvals_bvecs(
input_bval,
input_bvec,
)
gtab = gradient_table(bvals, bvecs)
if mask != '':
mask_data = nib.load(mask).get_data()
values = np.array(bvals)
ii = np.where(values == bvals.min())[0]
b0_average = np.mean(data[:, :, :, ii], axis=3)
fwidtimodel = fwdti.FreeWaterTensorModel(gtab, fit_method)
if mask != '':
fwidti_fit = fwidtimodel.fit(data, mask_data)
else:
fwidti_fit = fwidtimodel.fit(data)
if not os.path.exists(output_dir):
os.makedirs(output_dir)
output_evecs = output_dir + '/fwe_dti_eigenvectors.nii.gz'
output_evals = output_dir + '/fwe_dti_eigenvalues.nii.gz'
output_fa = output_dir + '/fwe_dti_FA.nii.gz'
output_md = output_dir + '/fwe_dti_MD.nii.gz'
output_rd = output_dir + '/fwe_dti_RD.nii.gz'
output_ad = output_dir + '/fwe_dti_AD.nii.gz'
output_f = output_dir + '/fwe_dti_F.nii.gz'
#Calculate Parameters for FWDTI Model
evals_img = nib.Nifti1Image(fwidti_fit.evals.astype(np.float32),
img.get_affine(), img.header)
nib.save(evals_img, output_evals)
evecs_img = nib.Nifti1Image(fwidti_fit.evecs.astype(np.float32),
img.get_affine(), img.header)
nib.save(evecs_img, output_evecs)
fwidti_fa = fwidti_fit.fa
fwidti_fa_img = nib.Nifti1Image(fwidti_fa.astype(np.float32),
img.get_affine(), img.header)
nib.save(fwidti_fa_img, output_fa)
fwidti_md = fwidti_fit.md
fwidti_md_img = nib.Nifti1Image(fwidti_md.astype(np.float32),
img.get_affine(), img.header)
nib.save(fwidti_md_img, output_md)
fwidti_ad = fwidti_fit.ad
fwidti_ad_img = nib.Nifti1Image(fwidti_ad.astype(np.float32),
img.get_affine(), img.header)
nib.save(fwidti_ad_img, output_ad)
fwidti_rd = fwidti_fit.rd
fwidti_rd_img = nib.Nifti1Image(fwidti_rd.astype(np.float32),
img.get_affine(), img.header)
nib.save(fwidti_rd_img, output_rd)
fwidti_f = fwidti_fit.f
fwidti_f_img = nib.Nifti1Image(fwidti_f.astype(np.float32),
img.get_affine(), img.header)
nib.save(fwidti_f_img, output_f)