def test_params_to_affine():
# TODO: test with zooms and shears
eps = np.finfo(np.float).eps
params = np.array([.1, .5, -.6, .1, -4., 5.1,
1., 1., 1., 0., 0., 0.])
affine = np.eye(4)
affine[:3, 3] = params[:3]
pitch = np.array([[1., 0., 0.],
[0., cos(.1), sin(.1)],
[0., -sin(.1), cos(.1)]])
roll = np.array([[cos(-4.), 0., sin(-4.)],
[0., 1., 0.],
[-sin(-4.), 0., cos(-4.)]])
yaw = np.array([[cos(5.1), sin(5.1), 0.],
[-sin(5.1), cos(5.1), 0.],
[0., 0., 1.]])
affine[:3, :3] = pitch.dot(roll).dot(yaw)
np.testing.assert_allclose(params_to_affine(params), affine)
# Test params_to_affine is the inverse of affine_to_params
affine2 = params_to_affine(affine_to_params(affine))
np.testing.assert_allclose(affine2, affine)
# Test same result as spm_matrix function of spm
np.testing.assert_allclose(affine, spm_matrix(params), atol=eps)
def test_params_to_affine():
# TODO: test with zooms and shears
eps = np.finfo(np.float).eps
params = np.array([0.1, 0.5, -0.6, 0.1, -4.0, 5.1, 1.0, 1.0, 1.0, 0.0, 0.0, 0.0])
affine = np.eye(4)
affine[:3, 3] = params[:3]
pitch = np.array([[1.0, 0.0, 0.0], [0.0, cos(0.1), sin(0.1)], [0.0, -sin(0.1), cos(0.1)]])
roll = np.array([[cos(-4.0), 0.0, sin(-4.0)], [0.0, 1.0, 0.0], [-sin(-4.0), 0.0, cos(-4.0)]])
yaw = np.array([[cos(5.1), sin(5.1), 0.0], [-sin(5.1), cos(5.1), 0.0], [0.0, 0.0, 1.0]])
affine[:3, :3] = pitch.dot(roll).dot(yaw)
np.testing.assert_allclose(spm_internals.params_to_affine(params), affine)
# Test params_to_affine is the inverse of affine_to_params
affine2 = spm_internals.params_to_affine(spm_internals.affine_to_params(affine))
np.testing.assert_allclose(affine2, affine)
def test_affine_to_params():
# TODO: test with zooms and shears
eps = np.finfo(np.float).eps
affine = np.eye(4)
affine[:3, 3] = [1.1, 0.55, -.3]
pitch = np.array([[1., 0., 0.],
[0., 0., 1.],
[0., -1., 0.]])
roll = np.array([[cos(-pi / 3.), 0., sin(-pi / 3.)],
[0., 1., 0.],
[-sin(-pi / 3.), 0., cos(-pi / 3.)]])
yaw = np.array([[cos(pi / 4.), sin(pi / 4.), 0.],
[-sin(pi / 4.), cos(pi / 4.), 0.],
[0., 0., 1.]])
affine[:3, :3] = pitch.dot(roll).dot(yaw)
params = np.array([1.1, 0.55, -.3, pi / 2., -pi / 3., pi / 4.,
1., 1., 1., 0., 0., 0.])
np.testing.assert_allclose(affine_to_params(affine), params, atol=eps)
# Test affine_to_params is the inverse of params_to_affine
params2 = affine_to_params(params_to_affine(params))
np.testing.assert_allclose(params2, params, atol=eps)
# Test same result as spm_matrix function of spm
np.testing.assert_allclose(params, spm_imatrix(affine), atol=1e4 * eps)
def _run_interface(self, runtime):
# Set the realignement options
realign = spm.Realign()
realign.inputs.paths = self.inputs.paths
realign.inputs.in_files = self.inputs.in_file
realign.inputs.register_to_mean = self.inputs.register_to_mean
realign.inputs.quality = 0.9
realign.inputs.fwhm = 5.
realign.inputs.separation = 4 # TODO: understand this parameter
realign.inputs.interp = 2
if self.inputs.correct_tagging:
# Estimate the realignement parameters
realign.inputs.jobtype = 'estimate'
realign.run()
parameters_file = realign.aggregate_outputs().get()[
'realignment_parameters']
rea_parameters = np.loadtxt(parameters_file)
# Correct for tagging: equal means for control and tag scans
n_scans = len(rea_parameters)
if self.inputs.control_scans:
control_scans = self.inputs.control_scans
else:
control_scans = range(0, n_scans, 2)
if self.inputs.tag_scans:
tag_scans = self.inputs.tag_scans
else:
tag_scans = range(1, n_scans, 2)
gap = np.mean(rea_parameters[control_scans], axis=0) -\
np.mean(rea_parameters[tag_scans], axis=0)
rea_parameters[control_scans] -= gap / 2.
rea_parameters[tag_scans] += gap / 2.
# Save the corrected realignement parameters
np.savetxt(parameters_file, rea_parameters, delimiter=' ')
# Save the corrected transforms for each frame in spm compatible
# .mat. This .mat serves as header for spm in case of 4D files
affine = spm_affine(self.inputs.in_file)
rea_affines = np.zeros((4, 4, n_scans))
for n_scan, param in enumerate(rea_parameters):
rea_affines[..., n_scan] = params_to_affine(param).dot(affine)
affines_file = os.path.splitext(self.inputs.in_file)[0] + '.mat'
savemat(affines_file, dict(mat=rea_affines))
else:
realign.inputs.jobtype = 'estimate'
realign.inputs.register_to_mean = self.inputs.register_to_mean
realign.run()
# Reslice and save the aligned volumes
# TODO: update Realign class in nipype.interfaces.spm.preprocess after
# bug is corrected
realign = spm.Realign()
realign.inputs.paths = self.inputs.paths
realign.inputs.in_files = self.inputs.in_file
realign.inputs.jobtype = 'write'
realign.run()
return runtime
def _run_interface(self, runtime):
# Set the realignement options
realign = spm.Realign()
realign.inputs.paths = self.inputs.paths
realign.inputs.in_files = self.inputs.in_file
realign.inputs.register_to_mean = self.inputs.register_to_mean
realign.inputs.quality = 0.9
realign.inputs.fwhm = 5.
realign.inputs.separation = 4 # TODO: understand this parameter
realign.inputs.interp = 2
if self.inputs.correct_tagging:
# Estimate the realignement parameters
realign.inputs.jobtype = 'estimate'
realign.run()
parameters_file = realign.aggregate_outputs().get()[
'realignment_parameters']
rea_parameters = np.loadtxt(parameters_file)
# Correct for tagging: equal means for control and tag scans
n_scans = len(rea_parameters)
if self.inputs.control_scans:
control_scans = self.inputs.control_scans
else:
control_scans = range(0, n_scans, 2)
if self.inputs.tag_scans:
tag_scans = self.inputs.tag_scans
else:
tag_scans = range(1, n_scans, 2)
gap = np.mean(rea_parameters[control_scans], axis=0) -\
np.mean(rea_parameters[tag_scans], axis=0)
rea_parameters[control_scans] -= gap / 2.
rea_parameters[tag_scans] += gap / 2.
# Save the corrected realignement parameters
np.savetxt(parameters_file, rea_parameters, delimiter=' ')
# Save the corrected transforms for each frame in spm compatible
# .mat. This .mat serves as header for spm in case of 4D files
affine = spm_affine(self.inputs.in_file)
rea_affines = np.zeros((4, 4, n_scans))
for n_scan, param in enumerate(rea_parameters):
rea_affines[..., n_scan] = params_to_affine(param).dot(affine)
affines_file = os.path.splitext(self.inputs.in_file)[0] + '.mat'
savemat(affines_file, dict(mat=rea_affines))
else:
realign.inputs.jobtype = 'estimate'
realign.inputs.register_to_mean = self.inputs.register_to_mean
realign.run()
# Reslice and save the aligned volumes
realign = spm.Realign()
realign.inputs.paths = self.inputs.paths
realign.inputs.in_files = self.inputs.in_file
realign.inputs.jobtype = 'write'
realign.run()
return runtime
def test_params_to_affine():
# TODO: test with zooms and shears
eps = np.finfo(np.float).eps
params = np.array([.1, .5, -.6, .1, -4., 5.1,
1., 1., 1., 0., 0., 0.])
affine = np.eye(4)
affine[:3, 3] = params[:3]
pitch = np.array([[1., 0., 0.],
[0., cos(.1), sin(.1)],
[0., -sin(.1), cos(.1)]])
roll = np.array([[cos(-4.), 0., sin(-4.)],
[0., 1., 0.],
[-sin(-4.), 0., cos(-4.)]])
yaw = np.array([[cos(5.1), sin(5.1), 0.],
[-sin(5.1), cos(5.1), 0.],
[0., 0., 1.]])
affine[:3, :3] = pitch.dot(roll).dot(yaw)
np.testing.assert_allclose(spm_internals.params_to_affine(params), affine)
# Test params_to_affine is the inverse of affine_to_params
affine2 = spm_internals.params_to_affine(
spm_internals.affine_to_params(affine))
np.testing.assert_allclose(affine2, affine)
def test_affine_to_params():
# TODO: test with zooms and shears
eps = np.finfo(np.float).eps
affine = np.eye(4)
affine[:3, 3] = [1.1, 0.55, -0.3]
pitch = np.array([[1.0, 0.0, 0.0], [0.0, 0.0, 1.0], [0.0, -1.0, 0.0]])
roll = np.array([[cos(-pi / 3.0), 0.0, sin(-pi / 3.0)], [0.0, 1.0, 0.0], [-sin(-pi / 3.0), 0.0, cos(-pi / 3.0)]])
yaw = np.array([[cos(pi / 4.0), sin(pi / 4.0), 0.0], [-sin(pi / 4.0), cos(pi / 4.0), 0.0], [0.0, 0.0, 1.0]])
affine[:3, :3] = pitch.dot(roll).dot(yaw)
params = np.array([1.1, 0.55, -0.3, pi / 2.0, -pi / 3.0, pi / 4.0, 1.0, 1.0, 1.0, 0.0, 0.0, 0.0])
np.testing.assert_allclose(spm_internals.affine_to_params(affine), params, atol=eps)
# Test affine_to_params is the inverse of params_to_affine
params2 = spm_internals.affine_to_params(spm_internals.params_to_affine(params))
np.testing.assert_allclose(params2, params, atol=eps)
def test_affine_to_params():
# TODO: test with zooms and shears
eps = np.finfo(np.float).eps
affine = np.eye(4)
affine[:3, 3] = [1.1, 0.55, -.3]
pitch = np.array([[1., 0., 0.],
[0., 0., 1.],
[0., -1., 0.]])
roll = np.array([[cos(-pi / 3.), 0., sin(-pi / 3.)],
[0., 1., 0.],
[-sin(-pi / 3.), 0., cos(-pi / 3.)]])
yaw = np.array([[cos(pi / 4.), sin(pi / 4.), 0.],
[-sin(pi / 4.), cos(pi / 4.), 0.],
[0., 0., 1.]])
affine[:3, :3] = pitch.dot(roll).dot(yaw)
params = np.array([1.1, 0.55, -.3, pi / 2., -pi / 3., pi / 4.,
1., 1., 1., 0., 0., 0.])
np.testing.assert_allclose(spm_internals.affine_to_params(affine), params,
atol=eps)
# Test affine_to_params is the inverse of params_to_affine
params2 = spm_internals.affine_to_params(
spm_internals.params_to_affine(params))
np.testing.assert_allclose(params2, params, atol=eps)
