def fit(self, data):
dwi_data = data[self._where_dwi]
shm_coeff, _ = csdeconv(dwi_data,
self._X,
self.B_reg,
self.tau,
P=self._P)
return SphHarmFit(self, shm_coeff, None)
def test_SphHarmFit():
coef = np.zeros((3, 4, 5, 45))
mask = np.zeros((3, 4, 5), dtype=bool)
fit = SphHarmFit(None, coef, mask)
item = fit[0, 0, 0]
assert_equal(item.shape, ())
slice = fit[0]
assert_equal(slice.shape, (4, 5))
slice = fit[..., 0]
assert_equal(slice.shape, (3, 4))
def fit(self, data):
s_sh = np.linalg.lstsq(self.B_dwi, data[self._where_dwi])[0]
# initial ODF estimation
odf_sh = np.dot(self.P, s_sh)
qball_odf = np.dot(self.B_reg, odf_sh)
Z = np.linalg.norm(qball_odf)
# normalize ODF
odf_sh /= Z
shm_coeff, num_it = odf_deconv(odf_sh, self.R, self.B_reg,
self.lambda_, self.tau)
# print 'SDT CSD converged after %d iterations' % num_it
return SphHarmFit(self, shm_coeff, None)
def fit(self, data, mask=None):
coeff = np.zeros(data.shape[:-1] + (15, ))
return SphHarmFit(self, coeff, mask=None)
def fit(self, data):
dwi_data = data[self._where_dwi]
X = self.R.diagonal() * self.B_dwi
shm_coeff, num_it = csdeconv(dwi_data, self.sh_order, X, self.B_reg,
self.lambda_, self.tau)
return SphHarmFit(self, shm_coeff, None)
def fit(self, data):
s_sh = np.linalg.lstsq(self.B_dwi, data[self._where_dwi])[0]
shm_coeff, num_it = csdeconv(s_sh, self.sh_order, self.R, self.B_reg,
self.lambda_, self.tau)
return SphHarmFit(self, shm_coeff, None)
