def test_single_voxel_fit():
"""
Test the implementation of the fitting for a single voxel.
Here, we will use the multi_tensor function to generate a
bi-exponential signal. The multi_tensor generates a multi
tensor signal and expects eigenvalues of each tensor in mevals.
Our basic test requires a scalar signal isotropic signal and
hence we set the same eigenvalue in all three directions to
generate the required signal.
The bvals, f, D_star and D are inspired from the paper by
Federau, Christian, et al. We use the function "generate_bvecs"
to simulate bvectors corresponding to the bvalues.
In the two stage fitting routine, initially we fit the signal
values for bvals less than the specified split_b using the
TensorModel and get an intial guess for f and D. Then, using
these parameters we fit the entire data for all bvalues.
"""
est_signal = ivim_prediction(ivim_fit_single.model_params, gtab)
assert_array_equal(est_signal.shape, data_single.shape)
assert_array_almost_equal(ivim_fit_single.model_params, params_LM)
assert_array_almost_equal(est_signal, data_single)
# Test predict function for single voxel
p = ivim_fit_single.predict(gtab)
assert_array_equal(p.shape, data_single.shape)
assert_array_almost_equal(p, data_single)
def test_single_voxel_fit():
"""
Test the implementation of the fitting for a single voxel.
Here, we will use the multi_tensor function to generate a
bi-exponential signal. The multi_tensor generates a multi
tensor signal and expects eigenvalues of each tensor in mevals.
Our basic test requires a scalar signal isotropic signal and
hence we set the same eigenvalue in all three directions to
generate the required signal.
The bvals, f, D_star and D are inspired from the paper by
Federau, Christian, et al. We use the function "generate_bvecs"
to simulate bvectors corresponding to the bvalues.
In the two stage fitting routine, initially we fit the signal
values for bvals less than the specified split_b using the
TensorModel and get an intial guess for f and D. Then, using
these parameters we fit the entire data for all bvalues.
"""
est_signal = ivim_prediction(ivim_fit_single.model_params, gtab)
assert_array_equal(est_signal.shape, data_single.shape)
assert_array_almost_equal(ivim_fit_single.model_params, params)
assert_array_almost_equal(est_signal, data_single)
# Test predict function for single voxel
p = ivim_fit_single.predict(gtab)
assert_array_equal(p.shape, data_single.shape)
assert_array_almost_equal(p, data_single)
def test_bounds_x0():
"""
Test to check if setting bounds for signal where initial value is
higher than subsequent values works.
These values are from the IVIM dataset which can be obtained by using
the `read_ivim` function from dipy.data.fetcher. These are values from
the voxel [160, 98, 33] which can be obtained by :
.. code-block:: python
from dipy.data.fetcher import read_ivim
img, gtab = read_ivim()
data = img.get_data()
signal = data[160, 98, 33, :]
"""
test_signal = np.array([4574.34814453, 4745.18164062, 4759.51806641,
4618.24951172, 4665.63623047, 4568.046875,
4525.90478516, 4734.54785156, 4526.41357422,
4299.99414062, 4256.61279297, 4254.50292969,
4098.74707031, 3776.10375977, 3614.0769043,
3440.56445312, 3146.52294922, 3006.94287109,
2879.69580078, 2728.44018555, 2600.09472656])
x0_test = np.array([1., 0.13, 0.001, 0.0001])
test_signal = ivim_prediction(x0_test, gtab)
ivim_fit = ivim_model.fit(test_signal)
est_signal = ivim_fit.predict(gtab)
assert_array_equal(est_signal.shape, test_signal.shape)
def test_bounds_x0():
"""
Test to check if setting bounds for signal where initial value is
higher than subsequent values works.
These values are from the IVIM dataset which can be obtained by using
the `read_ivim` function from dipy.data.fetcher. These are values from
the voxel [160, 98, 33] which can be obtained by :
.. code-block:: python
from dipy.data.fetcher import read_ivim
img, gtab = read_ivim()
data = img.get_data()
signal = data[160, 98, 33, :]
"""
test_signal = np.array([4574.34814453, 4745.18164062, 4759.51806641,
4618.24951172, 4665.63623047, 4568.046875,
4525.90478516, 4734.54785156, 4526.41357422,
4299.99414062, 4256.61279297, 4254.50292969,
4098.74707031, 3776.10375977, 3614.0769043,
3440.56445312, 3146.52294922, 3006.94287109,
2879.69580078, 2728.44018555, 2600.09472656])
x0_test = np.array([1., 0.13, 0.001, 0.0001])
test_signal = ivim_prediction(x0_test, gtab)
ivim_fit = ivim_model.fit(test_signal)
est_signal = ivim_fit.predict(gtab)
assert_array_equal(est_signal.shape, test_signal.shape)
def test_bounds_x0():
"""
Test to check if setting bounds for signal where initial value is
higher than subsequent values works.
These values are from the IVIM dataset which can be obtained by using
the `read_ivim` function from dipy.data.fetcher. These are values from
the voxel [160, 98, 33] which can be obtained by :
.. code-block:: python
from dipy.data.fetcher import read_ivim
img, gtab = read_ivim()
data = img.get_data()
signal = data[160, 98, 33, :]
"""
x0_test = np.array([1., 0.13, 0.001, 0.0001])
test_signal = ivim_prediction(x0_test, gtab)
ivim_fit = ivim_model_LM.fit(test_signal)
est_signal = ivim_fit.predict(gtab)
assert_array_equal(est_signal.shape, test_signal.shape)
def test_bounds_x0():
"""
Test to check if setting bounds for signal where initial value is
higher than subsequent values works.
These values are from the IVIM dataset which can be obtained by using
the `read_ivim` function from dipy.data.fetcher. These are values from
the voxel [160, 98, 33] which can be obtained by :
.. code-block:: python
from dipy.data.fetcher import read_ivim
img, gtab = read_ivim()
data = img.get_data()
signal = data[160, 98, 33, :]
"""
x0_test = np.array([1., 0.13, 0.001, 0.0001])
test_signal = ivim_prediction(x0_test, gtab)
ivim_fit = ivim_model.fit(test_signal)
est_signal = ivim_fit.predict(gtab)
assert_array_equal(est_signal.shape, test_signal.shape)
