def log_prop_vals(prop, saved_file, data, bvecs, idx_mask, idx_array, bval_ind_wb0, bvals_wb0, mask):
"""
Tensor model calculations of the given property
Parameters
----------
prop: str
String indicating the property to analyzed
'FA': Fractional anisotropy
'MD': Mean diffusivity
saved_file: 'str'
Indicate whether or not you want the function to create or use saved
parameter files
'no': Function will not create or use saved files
'yes': Function will create or use saved files
data: 4 dimensional array
Diffusion MRI data
bvecs: 3 dimensional array
All the b vectors
idx_array: ndarray
Array with the indices indicating the location of the non-zero values within
the mask
Returns
-------
log_prop: list
List of all the log of the desired property values
"""
prop_dict = {'FA':'FA', 'MD':'MD', 'mean diffusivity':'MD',
'fractional anisotropy':'FA', 'dispersion index':'DI',
'DI':'DI'}
log_prop = list()
for k in idx_array[:len(idx_array)-1]:
if saved_file is 'no':
params_file = 'temp'
elif saved_file is 'yes':
if prop_dict[prop] is 'DI':
params_file = 'SparseDeconvolutionModel{0}.nii.gz'.format(k+1)
else:
params_file = 'TensorModel{0}.nii.gz'.format(k+1)
if prop_dict[prop] is "FA":
tensor_prop = dti.TensorModel(data[:,:,:,bval_ind_wb0[k]], bvecs[:,bval_ind_wb0[k]], bvals_wb0[k], mask = mask, params_file = params_file)
prop_val = tensor_prop.fractional_anisotropy
log_prop.append(np.log(prop_val[idx_mask] + 0.01))
elif prop_dict[prop] is "MD":
tensor_prop = dti.TensorModel(data[:,:,:,bval_ind_wb0[k]], bvecs[:,bval_ind_wb0[k]], bvals_wb0[k], mask = mask, params_file = params_file)
prop_val = tensor_prop.mean_diffusivity
log_prop.append(np.log(prop_val[idx_mask] + 0.01))
elif prop_dict[prop] is 'DI':
sfm_di = sfm.SparseDeconvolutionModel(data[:,:,:,bval_ind_wb0[k]], bvecs[:,bval_ind_wb0[k]], bvals_wb0[k], mask = mask, params_file = params_file)
prop_val = sfm_di.dispersion_index()
log_prop.append(prop_val[idx_mask] + 0.01)
return log_prop
def test_log_prop_vals():
bvals_wb0_t, bval_ind_wb0_t = test_include_b0vals()
tensor_prop1 = dti.TensorModel(data_t[:,:,:,bval_ind_wb0_t[0]], bvecs_t[:,bval_ind_wb0_t[0]], bvals_wb0_t[0], mask = mask_t, params_file = 'temp')
tensor_prop2 = dti.TensorModel(data_t[:,:,:,bval_ind_wb0_t[1]], bvecs_t[:,bval_ind_wb0_t[1]], bvals_wb0_t[1], mask = mask_t, params_file = 'temp')
log_prop_t = [np.log(tensor_prop1.fractional_anisotropy[idx_mask_t]+0.01), np.log(tensor_prop2.fractional_anisotropy[idx_mask_t]+0.01)]
log_prop_a = mf.log_prop_vals('FA', saved_file, data_t, bvecs_t, idx_mask_t, idx_array, bval_ind_wb0_t, bvals_wb0_t, mask_t)
for idx in np.arange(len(log_prop_t)):
npt.assert_equal(abs(log_prop_t[0][idx]-log_prop_a[0][idx])<0.05, 1)
npt.assert_equal(abs(log_prop_t[1][idx]-log_prop_a[1][idx])<0.05, 1)
return log_prop_t
def tensor_reliability(SD_1, SD_2, wm_mask):
"""
given two SD models, calculate the reliability of the PDD of the derived
tensor models
Parameters
----------
SD_1, SD_2 : SparseDeconvolutionModel class instances
wm_mask : ndarray
Returns
-------
pdd_rel : ndarray
The angular difference between the PDD of the two tensor models fit to
the predicted data from each of the SD models.
"""
pred1 = np.concatenate([SD_1.S0[..., np.newaxis], SD_1.fit], -1)
pred2 = np.concatenate([SD_2.S0[..., np.newaxis], SD_2.fit], -1)
new_bvecs1 = np.concatenate(
[np.array([[0, 0, 0]]).T, SD_1.bvecs[:, SD_1.b_idx]], -1)
new_bvecs2 = np.concatenate(
[np.array([[0, 0, 0]]).T, SD_2.bvecs[:, SD_2.b_idx]], -1)
# Freely assume that the bvals are the same
new_bvals = np.hstack([0, SD_1.bvals[:, SD_1.b_idx]])
#
TM1 = dti.TensorModel(pred1,
new_bvecs1,
new_bvals,
mask=wm_mask,
params_file="temp")
TM2 = dti.TensorModel(pred2,
new_bvecs2,
new_bvals,
mask=wm_mask,
params_file="temp")
pdd_rel = oza.pdd_reliability(TM1, TM2)
return pdd_rel
def rm_ventricles(wm_data_file, bvals, bvecs, data, data_path):
"""
Removes the ventricles from the white matter mask and saves a new
white matter mask.
Parameters
----------
wm_data_file: obj
File handle for the white matter mask
bvals: 1 dimensional array
All b values
bvecs: 2 dimensional array
All the b vectors
data: 4 dimensional array
The diffusion data
data_path: str
Path to the data directory with all the sub files.
Returns
-------
new_wm_mask: 3 dimensional array
A new white matter mask with the ventricles removed.
"""
# Separate b-values and find the indices corresponding to the b0 and
# bNk measurements where N = the lowest b-value other than 0
wm_data = wm_data_file.get_data()
bval_list, b_inds, unique_b, bvals_scaled = ozu.separate_bvals(bvals)
all_b_idx = np.where(bvals_scaled != 0)
bNk_b0_inds = np.concatenate((b_inds[0], b_inds[1]))
# Fit a tensor model
tm = dti.TensorModel(data[..., bNk_b0_inds],
bvecs[:, bNk_b0_inds],
bvals[bNk_b0_inds],
mask=wm_data,
params_file="temp")
# Find the median, and 25th and 75th percentiles of mean diffusivities
md_median = np.median(tm.mean_diffusivity[np.where(wm_data)])
q1 = stats.scoreatpercentile(tm.mean_diffusivity[np.where(wm_data)], 25)
q3 = stats.scoreatpercentile(tm.mean_diffusivity[np.where(wm_data)], 75)
# Exclude voxels with MDs above median + 2*interquartile range
md_exclude = md_median + 2 * (q3 - q1)
md_include = np.where(tm.mean_diffusivity[np.where(wm_data)] < md_exclude)
new_wm_mask = np.zeros(wm_data.shape)
new_wm_mask[np.where(wm_data)[0][md_include],
np.where(wm_data)[1][md_include],
np.where(wm_data)[2][md_include]] = 1
wm = ni.Nifti1Image(new_wm_mask, wm_data_file.get_affine())
ni.save(wm, os.path.join(data_path, 'wm_mask_no_vent.nii.gz'))
return new_wm_mask
def make_wm_mask(seg_path, dwi, out_path, cutoff=2):
"""
Function to make a conservative WM mask, excluding partial volumed
voxels
Parameters
----------
seg_path : the full path to a white matter segmentation generated with the
itkGray conventions (in particular, 3 and 4 are the classification for WM
in left and right hemispheres).
dwi : the full path to a diffusion imaging data set, which will be used
to calculate the mean diffusivity for the exclusion of outliers (presumably
partial-volumed with the CSF) OR a TensorModel object, from which the MD
will be calculated (which can be a way to save time on the calculation of
the tensor parameters
out_path : the full path to where you want to put the wm mask once it's
calculated
cutoff : how many IQR away from median should we cut off on.
Note
----
Don't forget to look at your data, to make sure that things are sane.
"""
# Load the classification information:
seg_ni = ni.load(seg_path)
seg_vimg = as_volume_img(seg_ni)
if isinstance(dwi, str):
dwi_ni = ni.load(dwi + '.nii.gz')
else:
dwi_ni = ni.load(dwi.data_file)
vimg = as_volume_img(seg_ni)
# This does the magic - resamples using nearest neighbor interpolation:
seg_resamp_vimg = vimg.as_volume_img(affine=dwi_ni.get_affine(),
shape=dwi_ni.shape[:-1],
interpolation='nearest')
seg_resamp = seg_resamp_vimg.get_data()
# We know that WM is classified as 3 and 4 (for the two different
# hemispheres):
wm_idx = np.where(np.logical_or(seg_resamp == 3, seg_resamp == 4))
vol = np.zeros(seg_resamp.shape)
vol[wm_idx] = 1
if cutoff:
if isinstance(dwi, str):
# OK - now we need to find and exclude MD outliers:
TensorModel = ozm.TensorModel(dwi + '.nii.gz',
dwi + '.bvecs',
dwi + '.bvals',
mask=vol,
params_file='temp')
else:
TensorModel = dwi
MD = TensorModel.mean_diffusivity
IQR = (stats.scoreatpercentile(MD[np.isfinite(MD)], 75) -
stats.scoreatpercentile(MD[np.isfinite(MD)], 25))
co = np.median(MD[np.isfinite(MD)]) + cutoff * IQR
co_idx = np.where(MD > co)
# Null 'em out:
vol[co_idx] = 0
# Now, let's save some output:
ni.Nifti1Image(vol, dwi_ni.get_affine()).to_filename(out_path)
metavar='File',
help='Output file name (.nii.gz)')
parser.add_argument(
'--mask_file',
action='store',
metavar='File',
help=
'Mask file (only the voxels within the binary mask will be analyzed (.nii.gz; default: analyze all) ',
default=None)
params = parser.parse_args()
if __name__ == "__main__":
Model1 = dti.TensorModel(params.dwi_file1,
params.bvecs_file1,
params.bvals_file1,
mask=params.mask_file,
params_file='temp')
Model2 = dti.TensorModel(params.dwi_file2,
params.bvecs_file2,
params.bvals_file2,
mask=params.mask_file,
params_file='temp')
# Do it and save:
nib.Nifti1Image(ana.rsquared(Model1, Model2),
Model1.affine).to_filename(params.out_file)
data_path = oio.data_path
rrmse_dti = {}
rrmse_ssd = {}
for subject in ['FP', 'HT']:
subject_path = os.path.join(data_path, subject)
wm_mask_file = os.path.join(subject_path, '%s_wm_mask.nii.gz'%subject)
wm_nifti = ni.load(wm_mask_file).get_data()
wm_mask = np.zeros(wm_nifti.shape)
wm_mask[np.where(wm_nifti==1)] = 1
wm_idx = np.where(wm_nifti>0)
rrmse_dti[subject] = {}
rrmse_ssd[subject] = {}
for b in [1000, 2000, 4000]:
data_1, data_2 = oio.get_dwi_data(b, subject=subject)
TM1 = dti.TensorModel(*data_1, mask=wm_mask)
TM2 = dti.TensorModel(*data_2, mask=wm_mask)
rrmse_dti[subject][b] = ozm.cross_predict(TM1, TM2)
print subject
print b
rmse_mask = rrmse_dti[subject][b][wm_idx]
print "DTI: %s voxels above 1"%(len(np.where(rmse_mask>1)[0])/float(len(rmse_mask)))
print "Median rRMSE: %s"%np.median(rmse_mask)
ad_rd = oio.get_ad_rd(subject, b)
SD1 = ssd.SparseDeconvolutionModel(*data_1, mask=wm_mask,
axial_diffusivity=ad_rd[0]['AD'],
radial_diffusivity=ad_rd[0]['RD'])
SD2 = ssd.SparseDeconvolutionModel(*data_2, mask=wm_mask,
axial_diffusivity=ad_rd[1]['AD'],
radial_diffusivity=ad_rd[1]['RD'])
rrmse_ssd[subject][b] = ozm.cross_predict(SD1, SD2)
def initial_params(data, bvecs, bvals, model, mask=None, params_file='temp'):
"""
Determine the initial values for fitting the isotropic diffusion model.
This only works on the models that fit to the relative diffusion signal.
Parameters
----------
data: 4 dimensional array
Diffusion MRI data
bvecs: 2 dimensional array
All the b vectors
bvals: 1 dimensional array
All b values
model: str
Isotropic model
mask: 3 dimensional array
Mask of the data
params_file: obj or str
File handle of the param_files containing the tensor parameters.
Returns
-------
bounds: list
A list containing the bounds for each parameter for least squares
fitting.
initial: list
A list containing the initial values for each parameter for least
squares fitting.
"""
dti_mod = dti.TensorModel(data,
bvecs,
bvals,
mask=mask,
params_file=params_file)
d = dti_mod.mean_diffusivity[np.where(mask)]
# Find initial noise floor
_, b_inds, _, _ = ozu.separate_bvals(bvals)
b0_data = data[np.where(mask)][:, b_inds[0]]
#nf = np.std(b0_data, -1)/np.mean(b0_data, -1)
nf = np.min(data[np.where(mask)], -1)
if model == single_exp_rs:
bounds = [(0, 4)]
initial = d
elif model == single_exp_nf_rs:
bounds = [(0, 10000), (0, 4)]
initial = np.concatenate(
[nf[..., None], np.ones(d[..., None].shape)], -1)
elif model == bi_exp_rs:
bounds = [(0, 1), (0, 4), (0, 4)]
initial = np.concatenate(
[0.5 * np.ones((len(d), 1)), d[..., None], d[..., None]], -1)
elif model == bi_exp_nf_rs:
bounds = [(0, 10000), (0, 1), (0, 4), (0, 4)]
initial = np.concatenate([
nf[..., None], 0.5 * np.ones(
(len(d), 1)), d[..., None], d[..., None]
], -1)
return bounds, initial
bounds_max = maxs - diff
CC_box[bounds_min[0]:bounds_max[0], bounds_min[1]:bounds_max[1],
bounds_min[2]:bounds_max[2]] = 1
mask_corpus_callosum, cfa = segment_from_cfa(tensorfit,
CC_box,
threshold,
return_cfa=True)
# Clean up the cc isolation
new_mask = isolate_cc(mask_corpus_callosum)
tm = dti.TensorModel(bnk_data,
bvecs[:, bnk_b0_inds],
bvals[bnk_b0_inds],
mask=new_mask,
params_file="temp")
# Extract the median axial and radial diffusivities for a tensor fit to
# the voxels in the corpus callosum.
ad_arr[b_idx - 1] = np.median(tm.axial_diffusivity[np.where(new_mask)])
rd_arr[b_idx - 1] = np.median(
tm.radial_diffusivity[np.where(new_mask)])
os.chdir(data_path)
# Save a mask of the corpus callsoum.
aff = data_file.get_affine()
nib.Nifti1Image(new_mask).to_filename(
os.path.join(data_path, "cc_mask.nii.gz"))