def test_standardization():
data_shape = (9, 9, 5)
n_samples = 500
signals = np.random.randn(np.prod(data_shape), n_samples)
means = np.random.randn(np.prod(data_shape), 1) * 50 + 1000
signals += means
img = Nifti1Image(signals.reshape(data_shape + (n_samples, )), np.eye(4))
mask = Nifti1Image(np.ones(data_shape), np.eye(4))
# z-score
masker = NiftiMasker(mask, standardize='zscore')
trans_signals = masker.fit_transform(img)
np.testing.assert_almost_equal(trans_signals.mean(0), 0)
np.testing.assert_almost_equal(trans_signals.std(0), 1)
# psc
masker = NiftiMasker(mask, standardize='psc')
trans_signals = masker.fit_transform(img)
np.testing.assert_almost_equal(trans_signals.mean(0), 0)
np.testing.assert_almost_equal(
trans_signals,
(signals / signals.mean(1)[:, np.newaxis] * 100 - 100).T)
def test_dtype():
data_32 = np.zeros((9, 9, 9), dtype=np.float32)
data_64 = np.zeros((9, 9, 9), dtype=np.float64)
data_32[2:-2, 2:-2, 2:-2] = 10
data_64[2:-2, 2:-2, 2:-2] = 10
affine_32 = np.eye(4, dtype=np.float32)
affine_64 = np.eye(4, dtype=np.float64)
img_32 = Nifti1Image(data_32, affine_32)
img_64 = Nifti1Image(data_64, affine_64)
masker_1 = NiftiMasker(dtype='auto')
assert (masker_1.fit_transform(img_32).dtype == np.float32)
assert (masker_1.fit_transform(img_64).dtype == np.float32)
masker_2 = NiftiMasker(dtype='float64')
assert (masker_2.fit_transform(img_32).dtype == np.float64)
assert (masker_2.fit_transform(img_64).dtype == np.float64)
def test_dtype():
data_32 = np.zeros((9, 9, 9), dtype=np.float32)
data_64 = np.zeros((9, 9, 9), dtype=np.float64)
data_32[2:-2, 2:-2, 2:-2] = 10
data_64[2:-2, 2:-2, 2:-2] = 10
affine_32 = np.eye(4, dtype=np.float32)
affine_64 = np.eye(4, dtype=np.float64)
img_32 = Nifti1Image(data_32, affine_32)
img_64 = Nifti1Image(data_64, affine_64)
masker_1 = NiftiMasker(dtype='auto')
assert(masker_1.fit_transform(img_32).dtype == np.float32)
assert(masker_1.fit_transform(img_64).dtype == np.float32)
masker_2 = NiftiMasker(dtype='float64')
assert(masker_2.fit_transform(img_32).dtype == np.float64)
assert(masker_2.fit_transform(img_64).dtype == np.float64)
def test_resample():
# Check that target_affine triggers the right resampling
data = np.zeros((9, 9, 9))
data[3:-3, 3:-3, 3:-3] = 10
img = Nifti1Image(data, np.eye(4))
mask = data.astype(np.int)
mask_img = Nifti1Image(mask, np.eye(4))
masker = NiftiMasker(mask_img=mask_img, target_affine=2 * np.eye(3))
# Smoke test the fit
X = masker.fit_transform(img)
assert_true(np.any(X != 0))
def test_detrend():
# Check that detrending doesn't do something stupid with 3D images
data = np.zeros((9, 9, 9))
data[3:-3, 3:-3, 3:-3] = 10
img = Nifti1Image(data, np.eye(4))
mask = data.astype(np.int)
mask_img = Nifti1Image(mask, np.eye(4))
masker = NiftiMasker(mask_img=mask_img, detrend=True)
# Smoke test the fit
X = masker.fit_transform(img)
assert_true(np.any(X != 0))
def test_detrend():
# Check that detrending doesn't do something stupid with 3D images
data = np.zeros((9, 9, 9))
data[3:-3, 3:-3, 3:-3] = 10
img = Nifti1Image(data, np.eye(4))
mask = data.astype(np.int)
mask_img = Nifti1Image(mask, np.eye(4))
masker = NiftiMasker(mask_img=mask_img, detrend=True)
# Smoke test the fit
X = masker.fit_transform(img)
assert_true(np.any(X != 0))
def test_resample():
# Check that target_affine triggers the right resampling
data = np.zeros((9, 9, 9))
data[3:-3, 3:-3, 3:-3] = 10
img = Nifti1Image(data, np.eye(4))
mask = data.astype(np.int)
mask_img = Nifti1Image(mask, np.eye(4))
masker = NiftiMasker(mask_img=mask_img, target_affine=2 * np.eye(3))
# Smoke test the fit
X = masker.fit_transform(img)
assert_true(np.any(X != 0))
def test_matrix_orientation():
"""Test if processing is performed along the correct axis."""
# the "step" kind generate heavyside-like signals for each voxel.
# all signals being identical, standardizing along the wrong axis
# would leave a null signal. Along the correct axis, the step remains.
fmri, mask = testing.generate_fake_fmri(shape=(40, 41, 42), kind="step")
masker = NiftiMasker(mask_img=mask, standardize=True, detrend=True)
timeseries = masker.fit_transform(fmri)
assert (timeseries.shape[0] == fmri.shape[3])
assert (timeseries.shape[1] == mask.get_data().sum())
std = timeseries.std(axis=0)
assert (std.shape[0] == timeseries.shape[1]) # paranoid
assert (not np.any(std < 0.1))
# Test inverse transform
masker = NiftiMasker(mask_img=mask, standardize=False, detrend=False)
masker.fit()
timeseries = masker.transform(fmri)
recovered = masker.inverse_transform(timeseries)
np.testing.assert_array_almost_equal(recovered.get_data(), fmri.get_data())
def test_matrix_orientation():
"""Test if processing is performed along the correct axis."""
# the "step" kind generate heavyside-like signals for each voxel.
# all signals being identical, standardizing along the wrong axis
# would leave a null signal. Along the correct axis, the step remains.
fmri, mask = testing.generate_fake_fmri(shape=(40, 41, 42), kind="step")
masker = NiftiMasker(mask_img=mask, standardize=True, detrend=True)
timeseries = masker.fit_transform(fmri)
assert(timeseries.shape[0] == fmri.shape[3])
assert(timeseries.shape[1] == mask.get_data().sum())
std = timeseries.std(axis=0)
assert(std.shape[0] == timeseries.shape[1]) # paranoid
assert(not np.any(std < 0.1))
# Test inverse transform
masker = NiftiMasker(mask_img=mask, standardize=False, detrend=False)
masker.fit()
timeseries = masker.transform(fmri)
recovered = masker.inverse_transform(timeseries)
np.testing.assert_array_almost_equal(recovered.get_data(), fmri.get_data())
