def _get_direction_getter(self, strategy_name, pam, pmf_threshold,
max_angle):
"""Get Tracking Direction Getter object.
Parameters
----------
strategy_name : str
String representing direction getter name.
pam : instance of PeaksAndMetrics
An object with ``gfa``, ``peak_directions``, ``peak_values``,
``peak_indices``, ``odf``, ``shm_coeffs`` as attributes.
pmf_threshold : float
Threshold for ODF functions.
max_angle : float
Maximum angle between streamline segments.
Returns
-------
direction_getter : instance of DirectionGetter
Used to get directions for fiber tracking.
"""
dg, msg = None, ''
if strategy_name.lower() in ["deterministic", "det"]:
msg = "Deterministic"
dg = DeterministicMaximumDirectionGetter.from_shcoeff(
pam.shm_coeff,
sphere=pam.sphere,
max_angle=max_angle,
pmf_threshold=pmf_threshold)
elif strategy_name.lower() in ["probabilistic", "prob"]:
msg = "Probabilistic"
dg = ProbabilisticDirectionGetter.from_shcoeff(
pam.shm_coeff,
sphere=pam.sphere,
max_angle=max_angle,
pmf_threshold=pmf_threshold)
elif strategy_name.lower() in ["closestpeaks", "cp"]:
msg = "ClosestPeaks"
dg = ClosestPeakDirectionGetter.from_shcoeff(
pam.shm_coeff,
sphere=pam.sphere,
max_angle=max_angle,
pmf_threshold=pmf_threshold)
elif strategy_name.lower() in [
"eudx",
]:
msg = "Eudx"
dg = pam
else:
msg = "No direction getter defined. Eudx"
dg = pam
logging.info('{0} direction getter strategy selected'.format(msg))
return dg
def _get_direction_getter(self,
strategy_name,
pam,
pmf_threshold=0.1,
max_angle=30.):
"""Get Tracking Direction Getter object.
Parameters
----------
strategy_name: str
string representing direction getter name
Returns
-------
direction_getter : instance of DirectionGetter
Used to get directions for fiber tracking.
"""
dg, msg = None, ''
if strategy_name.lower() in ["deterministic", "det"]:
msg = "Deterministic"
dg = DeterministicMaximumDirectionGetter.from_shcoeff(
pam.shm_coeff,
sphere=pam.sphere,
max_angle=max_angle,
pmf_threshold=pmf_threshold)
elif strategy_name.lower() in ["probabilistic", "prob"]:
msg = "Probabilistic"
dg = ProbabilisticDirectionGetter.from_shcoeff(
pam.shm_coeff,
sphere=pam.sphere,
max_angle=max_angle,
pmf_threshold=pmf_threshold)
elif strategy_name.lower() in ["closestpeaks", "cp"]:
msg = "ClosestPeaks"
dg = ClosestPeakDirectionGetter.from_shcoeff(
pam.shm_coeff,
sphere=pam.sphere,
max_angle=max_angle,
pmf_threshold=pmf_threshold)
elif strategy_name.lower() in [
"eudx",
]:
msg = "Eudx"
dg = pam
else:
msg = "No direction getter defined. Deterministic"
dg = DeterministicMaximumDirectionGetter.from_shcoeff(
pam.shm_coeff,
sphere=pam.sphere,
max_angle=max_angle,
pmf_threshold=pmf_threshold)
logging.info('{0} direction getter strategy selected'.format(msg))
return dg
def _get_direction_getter(self, strategy_name, pam, pmf_threshold,
max_angle):
"""Get Tracking Direction Getter object.
Parameters
----------
strategy_name: str
String representing direction getter name.
pam: instance of PeaksAndMetrics
An object with ``gfa``, ``peak_directions``, ``peak_values``,
``peak_indices``, ``odf``, ``shm_coeffs`` as attributes.
pmf_threshold : float
Threshold for ODF functions.
max_angle : float
Maximum angle between streamline segments.
Returns
-------
direction_getter : instance of DirectionGetter
Used to get directions for fiber tracking.
"""
dg, msg = None, ''
if strategy_name.lower() in ["deterministic", "det"]:
msg = "Deterministic"
dg = DeterministicMaximumDirectionGetter.from_shcoeff(
pam.shm_coeff,
sphere=pam.sphere,
max_angle=max_angle,
pmf_threshold=pmf_threshold)
elif strategy_name.lower() in ["probabilistic", "prob"]:
msg = "Probabilistic"
dg = ProbabilisticDirectionGetter.from_shcoeff(
pam.shm_coeff,
sphere=pam.sphere,
max_angle=max_angle,
pmf_threshold=pmf_threshold)
elif strategy_name.lower() in ["closestpeaks", "cp"]:
msg = "ClosestPeaks"
dg = ClosestPeakDirectionGetter.from_shcoeff(
pam.shm_coeff,
sphere=pam.sphere,
max_angle=max_angle,
pmf_threshold=pmf_threshold)
elif strategy_name.lower() in ["eudx", ]:
msg = "Eudx"
dg = pam
else:
msg = "No direction getter defined. Eudx"
dg = pam
logging.info('{0} direction getter strategy selected'.format(msg))
return dg
def test_closest_peak_tracker():
"""This tests that the Closest Peak Direction Getter plays nice
LocalTracking and produces reasonable streamlines in a simple example.
"""
sphere = HemiSphere.from_sphere(unit_octahedron)
# A simple image with three possible configurations, a vertical tract,
# a horizontal tract and a crossing
pmf_lookup = np.array([[0., 0., 1.],
[1., 0., 0.],
[0., 1., 0.],
[.5, .5, 0.]])
simple_image = np.array([[0, 1, 0, 0, 0, 0],
[0, 1, 0, 0, 0, 0],
[2, 3, 2, 2, 2, 0],
[0, 1, 0, 0, 0, 0],
[0, 1, 0, 0, 0, 0],
])
simple_image = simple_image[..., None]
pmf = pmf_lookup[simple_image]
seeds = [np.array([1., 1., 0.]), np.array([2., 4., 0.])]
mask = (simple_image > 0).astype(float)
tc = BinaryTissueClassifier(mask)
dg = ClosestPeakDirectionGetter.from_pmf(pmf, 90, sphere,
pmf_threshold=0.1)
streamlines = Streamlines(LocalTracking(dg, tc, seeds, np.eye(4), 1.))
expected = [np.array([[0., 1., 0.],
[1., 1., 0.],
[2., 1., 0.],
[3., 1., 0.],
[4., 1., 0.]]),
np.array([[2., 0., 0.],
[2., 1., 0.],
[2., 2., 0.],
[2., 3., 0.],
[2., 4., 0.],
[2., 5., 0.]])]
def allclose(x, y):
return x.shape == y.shape and np.allclose(x, y)
if not allclose(streamlines[0], expected[0]):
raise AssertionError()
if not allclose(streamlines[1], expected[1]):
raise AssertionError()
def _get_direction_getter(self, strategy_name, pam, pmf_threshold=0.1,
max_angle=30.):
"""Get Tracking Direction Getter object.
Parameters
----------
strategy_name: str
string representing direction getter name
Returns
-------
direction_getter : instance of DirectionGetter
Used to get directions for fiber tracking.
"""
dg, msg = None, ''
if strategy_name.lower() in ["deterministic", "det"]:
msg = "Deterministic"
dg = DeterministicMaximumDirectionGetter.from_shcoeff(
pam.shm_coeff,
sphere=pam.sphere,
max_angle=max_angle,
pmf_threshold=pmf_threshold)
elif strategy_name.lower() in ["probabilistic", "prob"]:
msg = "Probabilistic"
dg = ProbabilisticDirectionGetter.from_shcoeff(
pam.shm_coeff,
sphere=pam.sphere,
max_angle=max_angle,
pmf_threshold=pmf_threshold)
elif strategy_name.lower() in ["closestpeaks", "cp"]:
msg = "ClosestPeaks"
dg = ClosestPeakDirectionGetter.from_shcoeff(
pam.shm_coeff,
sphere=pam.sphere,
max_angle=max_angle,
pmf_threshold=pmf_threshold)
elif strategy_name.lower() in ["eudx", ]:
msg = "Eudx"
dg = pam
else:
msg = "No direction getter defined. Deterministic"
dg = DeterministicMaximumDirectionGetter.from_shcoeff(
pam.shm_coeff,
sphere=pam.sphere,
max_angle=max_angle,
pmf_threshold=pmf_threshold)
logging.info('{0} direction getter strategy selected'.format(msg))
return dg
def test_closest_peak_tracker():
"""This tests that the Closest Peak Direction Getter plays nice
LocalTracking and produces reasonable streamlines in a simple example.
"""
sphere = HemiSphere.from_sphere(unit_octahedron)
# A simple image with three possible configurations, a vertical tract,
# a horizontal tract and a crossing
pmf_lookup = np.array([[0., 0., 1.],
[1., 0., 0.],
[0., 1., 0.],
[.5, .5, 0.]])
simple_image = np.array([[0, 1, 0, 0, 0, 0],
[0, 1, 0, 0, 0, 0],
[2, 3, 2, 2, 2, 0],
[0, 1, 0, 0, 0, 0],
[0, 1, 0, 0, 0, 0],
])
simple_image = simple_image[..., None]
pmf = pmf_lookup[simple_image]
seeds = [np.array([1., 1., 0.]), np.array([2., 4., 0.])]
mask = (simple_image > 0).astype(float)
tc = BinaryTissueClassifier(mask)
dg = ClosestPeakDirectionGetter.from_pmf(pmf, 90, sphere,
pmf_threshold=0.1)
streamlines = Streamlines(LocalTracking(dg, tc, seeds, np.eye(4), 1.))
expected = [np.array([[0., 1., 0.],
[1., 1., 0.],
[2., 1., 0.],
[3., 1., 0.],
[4., 1., 0.]]),
np.array([[2., 0., 0.],
[2., 1., 0.],
[2., 2., 0.],
[2., 3., 0.],
[2., 4., 0.]])]
def allclose(x, y):
return x.shape == y.shape and np.allclose(x, y)
if not allclose(streamlines[0], expected[0]):
raise AssertionError()
if not allclose(streamlines[1], expected[1]):
raise AssertionError()
**Corpus Callosum Bootstrap Probabilistic Direction Getter**
We have created a bootstrapped probabilistic set of streamlines. If you repeat
the fiber tracking (keeping all inputs the same) you will NOT get exactly the
same set of streamlines.
"""
"""
Example #2: Closest peak direction getter with CSD Model
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
"""
from dipy.direction import ClosestPeakDirectionGetter
pmf = csd_fit.odf(small_sphere).clip(min=0)
peak_dg = ClosestPeakDirectionGetter.from_pmf(pmf,
max_angle=30.,
sphere=small_sphere)
peak_streamline_generator = LocalTracking(peak_dg,
classifier,
seeds,
affine,
step_size=.5)
streamlines = Streamlines(peak_streamline_generator)
save_trk("closest_peak_dg_CSD.trk", streamlines, affine, labels.shape)
if has_fury:
r = window.Renderer()
r.add(actor.line(streamlines, colormap.line_colors(streamlines)))
window.record(r,
out_path='tractogram_closest_peak_dg.png',
def run_tracking(step_curv_combinations,
recon_shelved,
n_seeds_per_iter,
traversal,
maxcrossing,
max_length,
pft_back_tracking_dist,
pft_front_tracking_dist,
particle_count,
roi_neighborhood_tol,
min_length,
track_type,
min_separation_angle,
sphere,
tiss_class,
tissue_shelved,
verbose=False):
"""
Create a density map of the list of streamlines.
Parameters
----------
step_curv_combinations : list
List of tuples representing all pair combinations of step sizes and
curvature thresholds from which to sample streamlines.
recon_path : str
File path to diffusion reconstruction model.
n_seeds_per_iter : int
Number of seeds from which to initiate tracking for each unique
ensemble combination. By default this is set to 250.
directget : str
The statistical approach to tracking. Options are: det (deterministic),
closest (clos), boot (bootstrapped), and prob (probabilistic).
maxcrossing : int
Maximum number if diffusion directions that can be assumed per voxel
while tracking.
max_length : int
Maximum number of steps to restrict tracking.
pft_back_tracking_dist : float
Distance in mm to back track before starting the particle filtering
tractography. The total particle filtering tractography distance is
equal to back_tracking_dist + front_tracking_dist. By default this is
set to 2 mm.
pft_front_tracking_dist : float
Distance in mm to run the particle filtering tractography after the
the back track distance. The total particle filtering tractography
distance is equal to back_tracking_dist + front_tracking_dist. By
default this is set to 1 mm.
particle_count : int
Number of particles to use in the particle filter.
roi_neighborhood_tol : float
Distance (in the units of the streamlines, usually mm). If any
coordinate in the streamline is within this distance from the center
of any voxel in the ROI, the filtering criterion is set to True for
this streamline, otherwise False. Defaults to the distance between
the center of each voxel and the corner of the voxel.
waymask_data : ndarray
Tractography constraint mask array in native diffusion space.
min_length : int
Minimum fiber length threshold in mm to restrict tracking.
track_type : str
Tracking algorithm used (e.g. 'local' or 'particle').
min_separation_angle : float
The minimum angle between directions [0, 90].
sphere : obj
DiPy object for modeling diffusion directions on a sphere.
tiss_class : str
Tissue classification method.
tissue_shelved : str
File path to joblib-shelved 4D T1w tissue segmentations in native
diffusion space.
Returns
-------
streamlines : ArraySequence
DiPy list/array-like object of streamline points from tractography.
"""
import gc
import time
import numpy as np
from dipy.tracking import utils
from dipy.tracking.streamline import select_by_rois
from dipy.tracking.local_tracking import LocalTracking, \
ParticleFilteringTracking
from dipy.direction import (ProbabilisticDirectionGetter,
ClosestPeakDirectionGetter,
DeterministicMaximumDirectionGetter)
from nilearn.image import index_img, math_img
from pynets.dmri.utils import generate_seeds, random_seeds_from_mask
from nibabel.streamlines.array_sequence import ArraySequence
start_time = time.time()
if verbose is True:
print("%s%s%s" % ('Preparing tissue constraints:',
np.round(time.time() - start_time, 1), 's'))
start_time = time.time()
tissue_img = tissue_shelved.get()
# Order:
B0_mask = index_img(tissue_img, 0)
atlas_img = index_img(tissue_img, 1)
t1w2dwi = index_img(tissue_img, 3)
gm_in_dwi = index_img(tissue_img, 4)
vent_csf_in_dwi = index_img(tissue_img, 5)
wm_in_dwi = index_img(tissue_img, 6)
tissue_img.uncache()
tiss_classifier = prep_tissues(t1w2dwi, gm_in_dwi, vent_csf_in_dwi,
wm_in_dwi, tiss_class, B0_mask)
# if verbose is True:
# print("%s%s%s" % (
# 'Fitting tissue classifier:',
# np.round(time.time() - start_time, 1), 's'))
# start_time = time.time()
if verbose is True:
print("%s%s%s" % ('Loading reconstruction:',
np.round(time.time() - start_time, 1), 's'))
start_time = time.time()
print("%s%s" % ("Curvature: ", step_curv_combinations[1]))
# Instantiate DirectionGetter
if traversal.lower() in ["probabilistic", "prob"]:
dg = ProbabilisticDirectionGetter.from_shcoeff(
recon_shelved.get(),
max_angle=float(step_curv_combinations[1]),
sphere=sphere,
min_separation_angle=min_separation_angle,
)
elif traversal.lower() in ["closestpeaks", "cp"]:
dg = ClosestPeakDirectionGetter.from_shcoeff(
recon_shelved.get(),
max_angle=float(step_curv_combinations[1]),
sphere=sphere,
min_separation_angle=min_separation_angle,
)
elif traversal.lower() in ["deterministic", "det"]:
maxcrossing = 1
dg = DeterministicMaximumDirectionGetter.from_shcoeff(
recon_shelved.get(),
max_angle=float(step_curv_combinations[1]),
sphere=sphere,
min_separation_angle=min_separation_angle,
)
else:
raise ValueError("ERROR: No valid direction getter(s) specified.")
if verbose is True:
print("%s%s%s" % ('Extracting directions:',
np.round(time.time() - start_time, 1), 's'))
start_time = time.time()
print("%s%s" % ("Step: ", step_curv_combinations[0]))
# Perform wm-gm interface seeding, using n_seeds at a time
seeds = generate_seeds(
random_seeds_from_mask(np.asarray(
math_img("img > 0.01", img=index_img(
tissue_img, 2)).dataobj).astype("bool").astype("int16") > 0,
seeds_count=n_seeds_per_iter,
random_seed=42))
if verbose is True:
print("%s%s%s" % ('Drawing random seeds:',
np.round(time.time() - start_time, 1), 's'))
start_time = time.time()
# print(seeds)
# Perform tracking
if track_type == "local":
streamline_generator = LocalTracking(dg,
tiss_classifier,
np.stack([i for i in seeds]),
np.eye(4),
max_cross=int(maxcrossing),
maxlen=int(max_length),
step_size=float(
step_curv_combinations[0]),
fixedstep=False,
return_all=True,
random_seed=42)
elif track_type == "particle":
streamline_generator = ParticleFilteringTracking(
dg,
tiss_classifier,
np.stack([i for i in seeds]),
np.eye(4),
max_cross=int(maxcrossing),
step_size=float(step_curv_combinations[0]),
maxlen=int(max_length),
pft_back_tracking_dist=pft_back_tracking_dist,
pft_front_tracking_dist=pft_front_tracking_dist,
pft_max_trial=20,
particle_count=particle_count,
return_all=True,
random_seed=42)
else:
raise ValueError("ERROR: No valid tracking method(s) specified.")
if verbose is True:
print("%s%s%s" % ('Instantiating tracking:',
np.round(time.time() - start_time, 1), 's'))
start_time = time.time()
# print(seeds)
del dg
# Filter resulting streamlines by those that stay entirely
# inside the brain
try:
roi_proximal_streamlines = utils.target(
streamline_generator,
np.eye(4),
np.asarray(B0_mask.dataobj).astype('bool'),
include=True)
except BaseException:
print('No streamlines found inside the brain! ' 'Check registrations.')
#return None
if verbose is True:
print("%s%s%s" % ('Drawing streamlines:',
np.round(time.time() - start_time, 1), 's'))
start_time = time.time()
del seeds, tiss_classifier, streamline_generator
B0_mask.uncache()
atlas_img.uncache()
t1w2dwi.uncache()
gm_in_dwi.uncache()
vent_csf_in_dwi.uncache()
wm_in_dwi.uncache()
gc.collect()
# Filter resulting streamlines by roi-intersection
# characteristics
atlas_data = np.array(atlas_img.dataobj).astype("uint16")
# Build mask vector from atlas for later roi filtering
parcels = [
atlas_data == roi_val
for roi_val in [i for i in np.unique(atlas_data) if i != 0]
]
try:
roi_proximal_streamlines = \
select_by_rois(
roi_proximal_streamlines,
affine=np.eye(4),
rois=parcels,
include=list(np.ones(len(parcels)).astype("bool")),
mode="any",
tol=roi_neighborhood_tol,
)
except BaseException:
print('No streamlines found to connect any parcels! '
'Check registrations.')
#return None
del atlas_data
if verbose is True:
print("%s%s%s" % ('Selecting by parcellation:',
np.round(time.time() - start_time, 1), 's'))
start_time = time.time()
del parcels
gc.collect()
if verbose is True:
print("%s%s%s" % ('Selecting by minimum length criterion:',
np.round(time.time() - start_time, 1), 's'))
gc.collect()
return ArraySequence([
s.astype("float32") for s in roi_proximal_streamlines
if len(s) > float(min_length)
])
def run_tracking(step_curv_combinations,
recon_path,
n_seeds_per_iter,
directget,
maxcrossing,
max_length,
pft_back_tracking_dist,
pft_front_tracking_dist,
particle_count,
roi_neighborhood_tol,
waymask,
min_length,
track_type,
min_separation_angle,
sphere,
tiss_class,
tissues4d,
cache_dir,
min_seeds=100):
import gc
import os
import h5py
from dipy.tracking import utils
from dipy.tracking.streamline import select_by_rois
from dipy.tracking.local_tracking import LocalTracking, \
ParticleFilteringTracking
from dipy.direction import (ProbabilisticDirectionGetter,
ClosestPeakDirectionGetter,
DeterministicMaximumDirectionGetter)
from nilearn.image import index_img
from pynets.dmri.track import prep_tissues
from nibabel.streamlines.array_sequence import ArraySequence
from nipype.utils.filemanip import copyfile, fname_presuffix
import uuid
from time import strftime
run_uuid = f"{strftime('%Y%m%d_%H%M%S')}_{uuid.uuid4()}"
recon_path_tmp_path = fname_presuffix(
recon_path,
suffix=f"_{'_'.join([str(i) for i in step_curv_combinations])}_"
f"{run_uuid}",
newpath=cache_dir)
copyfile(recon_path, recon_path_tmp_path, copy=True, use_hardlink=False)
tissues4d_tmp_path = fname_presuffix(
tissues4d,
suffix=f"_{'_'.join([str(i) for i in step_curv_combinations])}_"
f"{run_uuid}",
newpath=cache_dir)
copyfile(tissues4d, tissues4d_tmp_path, copy=True, use_hardlink=False)
if waymask is not None:
waymask_tmp_path = fname_presuffix(
waymask,
suffix=f"_{'_'.join([str(i) for i in step_curv_combinations])}_"
f"{run_uuid}",
newpath=cache_dir)
copyfile(waymask, waymask_tmp_path, copy=True, use_hardlink=False)
else:
waymask_tmp_path = None
tissue_img = nib.load(tissues4d_tmp_path)
# Order:
B0_mask = index_img(tissue_img, 0)
atlas_img = index_img(tissue_img, 1)
seeding_mask = index_img(tissue_img, 2)
t1w2dwi = index_img(tissue_img, 3)
gm_in_dwi = index_img(tissue_img, 4)
vent_csf_in_dwi = index_img(tissue_img, 5)
wm_in_dwi = index_img(tissue_img, 6)
tiss_classifier = prep_tissues(t1w2dwi, gm_in_dwi, vent_csf_in_dwi,
wm_in_dwi, tiss_class, B0_mask)
B0_mask_data = np.asarray(B0_mask.dataobj).astype("bool")
seeding_mask = np.asarray(
seeding_mask.dataobj).astype("bool").astype("int16")
with h5py.File(recon_path_tmp_path, 'r+') as hf:
mod_fit = hf['reconstruction'][:].astype('float32')
print("%s%s" % ("Curvature: ", step_curv_combinations[1]))
# Instantiate DirectionGetter
if directget.lower() in ["probabilistic", "prob"]:
dg = ProbabilisticDirectionGetter.from_shcoeff(
mod_fit,
max_angle=float(step_curv_combinations[1]),
sphere=sphere,
min_separation_angle=min_separation_angle,
)
elif directget.lower() in ["closestpeaks", "cp"]:
dg = ClosestPeakDirectionGetter.from_shcoeff(
mod_fit,
max_angle=float(step_curv_combinations[1]),
sphere=sphere,
min_separation_angle=min_separation_angle,
)
elif directget.lower() in ["deterministic", "det"]:
maxcrossing = 1
dg = DeterministicMaximumDirectionGetter.from_shcoeff(
mod_fit,
max_angle=float(step_curv_combinations[1]),
sphere=sphere,
min_separation_angle=min_separation_angle,
)
else:
raise ValueError("ERROR: No valid direction getter(s) specified.")
print("%s%s" % ("Step: ", step_curv_combinations[0]))
# Perform wm-gm interface seeding, using n_seeds at a time
seeds = utils.random_seeds_from_mask(
seeding_mask > 0,
seeds_count=n_seeds_per_iter,
seed_count_per_voxel=False,
affine=np.eye(4),
)
if len(seeds) < min_seeds:
print(
UserWarning(
f"<{min_seeds} valid seed points found in wm-gm interface..."))
return None
# print(seeds)
# Perform tracking
if track_type == "local":
streamline_generator = LocalTracking(dg,
tiss_classifier,
seeds,
np.eye(4),
max_cross=int(maxcrossing),
maxlen=int(max_length),
step_size=float(
step_curv_combinations[0]),
fixedstep=False,
return_all=True,
random_seed=42)
elif track_type == "particle":
streamline_generator = ParticleFilteringTracking(
dg,
tiss_classifier,
seeds,
np.eye(4),
max_cross=int(maxcrossing),
step_size=float(step_curv_combinations[0]),
maxlen=int(max_length),
pft_back_tracking_dist=pft_back_tracking_dist,
pft_front_tracking_dist=pft_front_tracking_dist,
pft_max_trial=20,
particle_count=particle_count,
return_all=True,
random_seed=42)
else:
raise ValueError("ERROR: No valid tracking method(s) specified.")
# Filter resulting streamlines by those that stay entirely
# inside the brain
try:
roi_proximal_streamlines = utils.target(streamline_generator,
np.eye(4),
B0_mask_data.astype('bool'),
include=True)
except BaseException:
print('No streamlines found inside the brain! ' 'Check registrations.')
return None
del mod_fit, seeds, tiss_classifier, streamline_generator, \
B0_mask_data, seeding_mask, dg
B0_mask.uncache()
atlas_img.uncache()
t1w2dwi.uncache()
gm_in_dwi.uncache()
vent_csf_in_dwi.uncache()
wm_in_dwi.uncache()
atlas_img.uncache()
tissue_img.uncache()
gc.collect()
# Filter resulting streamlines by roi-intersection
# characteristics
atlas_data = np.array(atlas_img.dataobj).astype("uint16")
# Build mask vector from atlas for later roi filtering
parcels = []
i = 0
intensities = [i for i in np.unique(atlas_data) if i != 0]
for roi_val in intensities:
parcels.append(atlas_data == roi_val)
i += 1
parcel_vec = list(np.ones(len(parcels)).astype("bool"))
try:
roi_proximal_streamlines = \
nib.streamlines.array_sequence.ArraySequence(
select_by_rois(
roi_proximal_streamlines,
affine=np.eye(4),
rois=parcels,
include=parcel_vec,
mode="any",
tol=roi_neighborhood_tol,
)
)
print("%s%s" % ("Filtering by: \nNode intersection: ",
len(roi_proximal_streamlines)))
except BaseException:
print('No streamlines found to connect any parcels! '
'Check registrations.')
return None
try:
roi_proximal_streamlines = nib.streamlines. \
array_sequence.ArraySequence(
[
s for s in roi_proximal_streamlines
if len(s) >= float(min_length)
]
)
print(f"Minimum fiber length >{min_length}mm: "
f"{len(roi_proximal_streamlines)}")
except BaseException:
print('No streamlines remaining after minimal length criterion.')
return None
if waymask is not None and os.path.isfile(waymask_tmp_path):
waymask_data = np.asarray(
nib.load(waymask_tmp_path).dataobj).astype("bool")
try:
roi_proximal_streamlines = roi_proximal_streamlines[utils.near_roi(
roi_proximal_streamlines,
np.eye(4),
waymask_data,
tol=int(round(roi_neighborhood_tol * 0.50, 1)),
mode="all")]
print("%s%s" %
("Waymask proximity: ", len(roi_proximal_streamlines)))
del waymask_data
except BaseException:
print('No streamlines remaining in waymask\'s vacinity.')
return None
hf.close()
del parcels, atlas_data
tmp_files = [tissues4d_tmp_path, waymask_tmp_path, recon_path_tmp_path]
for j in tmp_files:
if j is not None:
if os.path.isfile(j):
os.system(f"rm -f {j} &")
if len(roi_proximal_streamlines) > 0:
return ArraySequence(
[s.astype("float32") for s in roi_proximal_streamlines])
else:
return None
def track_ensemble(dwi_data,
target_samples,
atlas_data_wm_gm_int,
parcels,
mod_fit,
tiss_classifier,
sphere,
directget,
curv_thr_list,
step_list,
track_type,
maxcrossing,
max_length,
roi_neighborhood_tol,
min_length,
waymask,
n_seeds_per_iter=100,
pft_back_tracking_dist=2,
pft_front_tracking_dist=1,
particle_count=15):
"""
Perform native-space ensemble tractography, restricted to a vector of ROI masks.
dwi_data : array
4D array of dwi data.
target_samples : int
Total number of streamline samples specified to generate streams.
atlas_data_wm_gm_int : array
3D int32 numpy array of atlas parcellation intensities from Nifti1Image in T1w-warped native diffusion space,
restricted to wm-gm interface.
parcels : list
List of 3D boolean numpy arrays of atlas parcellation ROI masks from a Nifti1Image in T1w-warped native
diffusion space.
mod : obj
Connectivity reconstruction model.
tiss_classifier : str
Tissue classification method.
sphere : obj
DiPy object for modeling diffusion directions on a sphere.
directget : str
The statistical approach to tracking. Options are: det (deterministic), closest (clos), boot (bootstrapped),
and prob (probabilistic).
curv_thr_list : list
List of integer curvature thresholds used to perform ensemble tracking.
step_list : list
List of float step-sizes used to perform ensemble tracking.
track_type : str
Tracking algorithm used (e.g. 'local' or 'particle').
maxcrossing : int
Maximum number if diffusion directions that can be assumed per voxel while tracking.
max_length : int
Maximum fiber length threshold in mm to restrict tracking.
roi_neighborhood_tol : float
Distance (in the units of the streamlines, usually mm). If any
coordinate in the streamline is within this distance from the center
of any voxel in the ROI, the filtering criterion is set to True for
this streamline, otherwise False. Defaults to the distance between
the center of each voxel and the corner of the voxel.
min_length : int
Minimum fiber length threshold in mm.
waymask : str
Path to a Nifti1Image in native diffusion space to constrain tractography.
n_seeds_per_iter : int
Number of seeds from which to initiate tracking for each unique ensemble combination.
By default this is set to 200.
particle_count
pft_back_tracking_dist : float
Distance in mm to back track before starting the particle filtering
tractography. The total particle filtering tractography distance is
equal to back_tracking_dist + front_tracking_dist. By default this is set to 2 mm.
pft_front_tracking_dist : float
Distance in mm to run the particle filtering tractography after the
the back track distance. The total particle filtering tractography
distance is equal to back_tracking_dist + front_tracking_dist. By
default this is set to 1 mm.
particle_count : int
Number of particles to use in the particle filter.
Returns
-------
streamlines : ArraySequence
DiPy list/array-like object of streamline points from tractography.
"""
from colorama import Fore, Style
from dipy.tracking import utils
from dipy.tracking.streamline import Streamlines, select_by_rois
from dipy.tracking.local_tracking import LocalTracking, ParticleFilteringTracking
from dipy.direction import ProbabilisticDirectionGetter, BootDirectionGetter, ClosestPeakDirectionGetter, DeterministicMaximumDirectionGetter
if waymask:
waymask_data = nib.load(waymask).get_fdata().astype('bool')
# Commence Ensemble Tractography
parcel_vec = list(np.ones(len(parcels)).astype('bool'))
streamlines = nib.streamlines.array_sequence.ArraySequence()
ix = 0
circuit_ix = 0
stream_counter = 0
while int(stream_counter) < int(target_samples):
for curv_thr in curv_thr_list:
print("%s%s" % ('Curvature: ', curv_thr))
# Instantiate DirectionGetter
if directget == 'prob':
dg = ProbabilisticDirectionGetter.from_shcoeff(
mod_fit, max_angle=float(curv_thr), sphere=sphere)
elif directget == 'boot':
dg = BootDirectionGetter.from_data(dwi_data,
mod_fit,
max_angle=float(curv_thr),
sphere=sphere)
elif directget == 'clos':
dg = ClosestPeakDirectionGetter.from_shcoeff(
mod_fit, max_angle=float(curv_thr), sphere=sphere)
elif directget == 'det':
dg = DeterministicMaximumDirectionGetter.from_shcoeff(
mod_fit, max_angle=float(curv_thr), sphere=sphere)
else:
raise ValueError(
'ERROR: No valid direction getter(s) specified.')
for step in step_list:
print("%s%s" % ('Step: ', step))
# Perform wm-gm interface seeding, using n_seeds at a time
seeds = utils.random_seeds_from_mask(
atlas_data_wm_gm_int > 0,
seeds_count=n_seeds_per_iter,
seed_count_per_voxel=False,
affine=np.eye(4))
if len(seeds) == 0:
raise RuntimeWarning(
'Warning: No valid seed points found in wm-gm interface...'
)
print(seeds)
# Perform tracking
if track_type == 'local':
streamline_generator = LocalTracking(
dg,
tiss_classifier,
seeds,
np.eye(4),
max_cross=int(maxcrossing),
maxlen=int(max_length),
step_size=float(step),
return_all=True)
elif track_type == 'particle':
streamline_generator = ParticleFilteringTracking(
dg,
tiss_classifier,
seeds,
np.eye(4),
max_cross=int(maxcrossing),
step_size=float(step),
maxlen=int(max_length),
pft_back_tracking_dist=pft_back_tracking_dist,
pft_front_tracking_dist=pft_front_tracking_dist,
particle_count=particle_count,
return_all=True)
else:
raise ValueError(
'ERROR: No valid tracking method(s) specified.')
# Filter resulting streamlines by roi-intersection characteristics
roi_proximal_streamlines = Streamlines(
select_by_rois(streamline_generator,
affine=np.eye(4),
rois=parcels,
include=parcel_vec,
mode='any',
tol=roi_neighborhood_tol))
print("%s%s" %
('Qualifying Streamlines by node intersection: ',
len(roi_proximal_streamlines)))
roi_proximal_streamlines = nib.streamlines.array_sequence.ArraySequence(
[
s for s in roi_proximal_streamlines
if len(s) > float(min_length)
])
print("%s%s" %
('Qualifying Streamlines by minimum length criterion: ',
len(roi_proximal_streamlines)))
if waymask:
roi_proximal_streamlines = roi_proximal_streamlines[
utils.near_roi(roi_proximal_streamlines,
np.eye(4),
waymask_data,
tol=roi_neighborhood_tol,
mode='any')]
print("%s%s" %
('Qualifying Streamlines by waymask proximity: ',
len(roi_proximal_streamlines)))
# Repeat process until target samples condition is met
ix = ix + 1
for s in roi_proximal_streamlines:
stream_counter = stream_counter + len(s)
streamlines.append(s)
if int(stream_counter) >= int(target_samples):
break
else:
continue
# Cleanup memory
del seeds, roi_proximal_streamlines, streamline_generator
del dg
circuit_ix = circuit_ix + 1
print(
"%s%s%s%s%s" %
('Completed hyperparameter circuit: ', circuit_ix,
'...\nCumulative Streamline Count: ', Fore.CYAN, stream_counter))
print(Style.RESET_ALL)
print('\n')
return streamlines
We have created a bootstrapped probabilistic set of streamlines. If you repeat
the fiber tracking (keeping all inputs the same) you will NOT get exactly the
same set of streamlines. We can save the streamlines as a Trackvis file so it
can be loaded into other software for visualization or further analysis.
"""
save_trk("bootstrap_dg_CSD.trk", streamlines, affine, labels.shape)
"""
Example #2: Closest peak direction getter with CSD Model
"""
from dipy.direction import ClosestPeakDirectionGetter
pmf = csd_fit.odf(small_sphere).clip(min=0)
peak_dg = ClosestPeakDirectionGetter.from_pmf(pmf, max_angle=30.,
sphere=small_sphere)
peak_streamline_generator = LocalTracking(peak_dg, classifier, seeds, affine,
step_size=.5)
streamlines = Streamlines(peak_streamline_generator)
renderer.clear()
renderer.add(actor.line(streamlines, line_colors(streamlines)))
window.record(renderer, out_path='closest_peak_dg_CSD.png', size=(600, 600))
"""
.. figure:: closest_peak_dg_CSD.png
:align: center
**Corpus Callosum Closest Peak Deterministic Direction Getter**
We have created a set of streamlines using the closest peak direction getter,
def tracking(shm_file, mask_file, outdir, force_overwrite, particles,
step_size, max_lenght, max_angle, algorithm, wpid_seeds_info):
''' Tracking function that will run in parallel
Params:
shm_file: SHM file computed from the dwi file
mask_file: mask were to perform tractography
outdir: Directory were to save streamlines
force_overwrite: if True, existing files will be overwriten
step_size: size in mm of each step in the tracking
max_lenght: maximum lenght of each streamline
max_angle: maximum angle at each step of tracking
algoright: either 'probabilistic' or 'deterministic'
wpid_seeds_info: tuple which contains:
- wpid: The id of this worker
- seeds: One list for each seed with points to track from
- info: CIFTI information for each seed:
-mtype: A valid CIFTI MODELTYPE
-name: A valid CIFTI BRAINSTRUCTURE
-coord: Voxel or vertex to which the seed makes reference
-size: size of the CIFTI SURFACE (if applies)
Returns:
list of streamlines '''
import citrix
import streamlines as sl
from dipy.data import default_sphere
from dipy.tracking.local_tracking import LocalTracking
from dipy.tracking.stopping_criterion import BinaryStoppingCriterion
wpid, (seeds, cifti_info) = wpid_seeds_info
logging.debug("Worker {} started".format(wpid))
# Check if file exists
outfile = os.path.join(outdir, "stream_{}.trk".format(wpid))
if os.path.isfile(outfile) and not force_overwrite:
print("File already exists, use the -f flag to overwrite it")
return
shm = citrix.load(shm_file)
shm_data = shm.get_data()
mask_nib = citrix.load(mask_file)
mask = mask_nib.get_data()
if algorithm == 'deterministic':
directions = deterministic.from_shcoeff(shm_data, max_angle,
default_sphere)
else:
directions = probabilistic.from_shcoeff(shm_data, max_angle,
default_sphere)
stop_criterion = BinaryStoppingCriterion(mask)
percent = max(1, len(seeds) / 5)
streamlines = []
used_seeds = []
for i, s in enumerate(seeds):
if i % percent == 0:
logging.debug("{}, {}/{} seeds".format(wpid, i, len(seeds)))
# Repeat the seeds as long as needed
if len(s) == 3:
# It's one point
s = [s]
repeated_seeds = [ss for ss in s for _ in range(2 * particles)]
res = LocalTracking(directions,
stop_criterion,
repeated_seeds,
shm.affine,
step_size=step_size,
maxlen=max_lenght,
return_all=False)
for streamline in itertools.islice(res, particles * len(s)):
if streamline is not None and len(streamline) > 1:
streamlines.append(streamline)
if cifti_info[i][0] == 'CIFTI_MODEL_TYPE_SURFACE':
used_seeds.append(cifti_info[i][2])
else:
used_seeds.append([int(cf) for cf in cifti_info[i][2]])
streamlines = sl.Streamlines(streamlines, shm.affine, shm.shape[:3],
shm.header.get_zooms()[:3])
numpy.savetxt(os.path.join(outdir, "info_{}.txt".format(wpid)), used_seeds)
sl.io.save(streamlines, outfile)
logging.debug("Worker {} finished".format(wpid))
return
def track_ensemble(target_samples, atlas_data_wm_gm_int, parcels, mod_fit, tiss_classifier, sphere, directget,
curv_thr_list, step_list, track_type, maxcrossing, roi_neighborhood_tol, min_length, waymask,
B0_mask, max_length=1000, n_seeds_per_iter=500, pft_back_tracking_dist=2, pft_front_tracking_dist=1,
particle_count=15, min_separation_angle=20):
"""
Perform native-space ensemble tractography, restricted to a vector of ROI masks.
target_samples : int
Total number of streamline samples specified to generate streams.
atlas_data_wm_gm_int : array
3D int32 numpy array of atlas parcellation intensities from Nifti1Image in T1w-warped native diffusion space,
restricted to wm-gm interface.
parcels : list
List of 3D boolean numpy arrays of atlas parcellation ROI masks from a Nifti1Image in T1w-warped native
diffusion space.
mod : obj
Connectivity reconstruction model.
tiss_classifier : str
Tissue classification method.
sphere : obj
DiPy object for modeling diffusion directions on a sphere.
directget : str
The statistical approach to tracking. Options are: det (deterministic), closest (clos), boot (bootstrapped),
and prob (probabilistic).
curv_thr_list : list
List of integer curvature thresholds used to perform ensemble tracking.
step_list : list
List of float step-sizes used to perform ensemble tracking.
track_type : str
Tracking algorithm used (e.g. 'local' or 'particle').
maxcrossing : int
Maximum number if diffusion directions that can be assumed per voxel while tracking.
roi_neighborhood_tol : float
Distance (in the units of the streamlines, usually mm). If any
coordinate in the streamline is within this distance from the center
of any voxel in the ROI, the filtering criterion is set to True for
this streamline, otherwise False. Defaults to the distance between
the center of each voxel and the corner of the voxel.
min_length : int
Minimum fiber length threshold in mm.
waymask : str
Path to a Nifti1Image in native diffusion space to constrain tractography.
B0_mask : str
File path to B0 brain mask.
max_length : int
Maximum number of steps to restrict tracking.
n_seeds_per_iter : int
Number of seeds from which to initiate tracking for each unique ensemble combination.
By default this is set to 200.
particle_count
pft_back_tracking_dist : float
Distance in mm to back track before starting the particle filtering
tractography. The total particle filtering tractography distance is
equal to back_tracking_dist + front_tracking_dist. By default this is set to 2 mm.
pft_front_tracking_dist : float
Distance in mm to run the particle filtering tractography after the
the back track distance. The total particle filtering tractography
distance is equal to back_tracking_dist + front_tracking_dist. By
default this is set to 1 mm.
particle_count : int
Number of particles to use in the particle filter.
min_separation_angle : float
The minimum angle between directions [0, 90].
Returns
-------
streamlines : ArraySequence
DiPy list/array-like object of streamline points from tractography.
References
----------
.. [1] Takemura, H., Caiafa, C. F., Wandell, B. A., & Pestilli, F. (2016).
Ensemble Tractography. PLoS Computational Biology.
https://doi.org/10.1371/journal.pcbi.1004692
"""
import gc
import time
from colorama import Fore, Style
from dipy.tracking import utils
from dipy.tracking.streamline import Streamlines, select_by_rois
from dipy.tracking.local_tracking import LocalTracking, ParticleFilteringTracking
from dipy.direction import (ProbabilisticDirectionGetter, ClosestPeakDirectionGetter,
DeterministicMaximumDirectionGetter)
start = time.time()
B0_mask_data = nib.load(B0_mask).get_fdata()
if waymask:
waymask_data = np.asarray(nib.load(waymask).dataobj).astype('bool')
# Commence Ensemble Tractography
parcel_vec = list(np.ones(len(parcels)).astype('bool'))
streamlines = nib.streamlines.array_sequence.ArraySequence()
circuit_ix = 0
stream_counter = 0
while int(stream_counter) < int(target_samples):
for curv_thr in curv_thr_list:
print("%s%s" % ('Curvature: ', curv_thr))
# Instantiate DirectionGetter
if directget == 'prob':
dg = ProbabilisticDirectionGetter.from_shcoeff(mod_fit, max_angle=float(curv_thr), sphere=sphere,
min_separation_angle=min_separation_angle)
elif directget == 'clos':
dg = ClosestPeakDirectionGetter.from_shcoeff(mod_fit, max_angle=float(curv_thr), sphere=sphere,
min_separation_angle=min_separation_angle)
elif directget == 'det':
dg = DeterministicMaximumDirectionGetter.from_shcoeff(mod_fit, max_angle=float(curv_thr), sphere=sphere,
min_separation_angle=min_separation_angle)
else:
raise ValueError('ERROR: No valid direction getter(s) specified.')
for step in step_list:
print("%s%s" % ('Step: ', step))
# Perform wm-gm interface seeding, using n_seeds at a time
seeds = utils.random_seeds_from_mask(atlas_data_wm_gm_int > 0, seeds_count=n_seeds_per_iter,
seed_count_per_voxel=False, affine=np.eye(4))
if len(seeds) == 0:
raise RuntimeWarning('Warning: No valid seed points found in wm-gm interface...')
# print(seeds)
# Perform tracking
if track_type == 'local':
streamline_generator = LocalTracking(dg, tiss_classifier, seeds, np.eye(4),
max_cross=int(maxcrossing), maxlen=int(max_length),
step_size=float(step), fixedstep=False, return_all=True)
elif track_type == 'particle':
streamline_generator = ParticleFilteringTracking(dg, tiss_classifier, seeds, np.eye(4),
max_cross=int(maxcrossing),
step_size=float(step),
maxlen=int(max_length),
pft_back_tracking_dist=pft_back_tracking_dist,
pft_front_tracking_dist=pft_front_tracking_dist,
particle_count=particle_count,
return_all=True)
else:
raise ValueError('ERROR: No valid tracking method(s) specified.')
# Filter resulting streamlines by those that stay entirely inside the brain
roi_proximal_streamlines = utils.target(streamline_generator, np.eye(4), B0_mask_data,
include=True)
# Filter resulting streamlines by roi-intersection characteristics
roi_proximal_streamlines = Streamlines(select_by_rois(roi_proximal_streamlines, affine=np.eye(4),
rois=parcels, include=parcel_vec,
mode='both_end',
tol=roi_neighborhood_tol))
print("%s%s" % ('Filtering by: \nnode intersection: ', len(roi_proximal_streamlines)))
if str(min_length) != '0':
roi_proximal_streamlines = nib.streamlines.array_sequence.ArraySequence([s for s in
roi_proximal_streamlines
if len(s) >=
float(min_length)])
print("%s%s" % ('Minimum length criterion: ', len(roi_proximal_streamlines)))
if waymask:
roi_proximal_streamlines = roi_proximal_streamlines[utils.near_roi(roi_proximal_streamlines,
np.eye(4),
waymask_data,
tol=roi_neighborhood_tol,
mode='any')]
print("%s%s" % ('Waymask proximity: ', len(roi_proximal_streamlines)))
out_streams = [s.astype('float32') for s in roi_proximal_streamlines]
streamlines.extend(out_streams)
stream_counter = stream_counter + len(out_streams)
# Cleanup memory
del seeds, roi_proximal_streamlines, streamline_generator, out_streams
gc.collect()
del dg
circuit_ix = circuit_ix + 1
print("%s%s%s%s%s%s" % ('Completed Hyperparameter Circuit: ', circuit_ix,
'\nCumulative Streamline Count: ', Fore.CYAN, stream_counter, "\n"))
print(Style.RESET_ALL)
print('Tracking Complete:\n', str(time.time() - start))
return streamlines
def run_tracking(step_curv_combinations, recon_path, n_seeds_per_iter,
directget, maxcrossing, max_length, pft_back_tracking_dist,
pft_front_tracking_dist, particle_count, roi_neighborhood_tol,
waymask, min_length, track_type, min_separation_angle, sphere,
tiss_class, tissues4d, cache_dir):
import gc
import os
import h5py
from dipy.tracking import utils
from dipy.tracking.streamline import select_by_rois
from dipy.tracking.local_tracking import LocalTracking, \
ParticleFilteringTracking
from dipy.direction import (ProbabilisticDirectionGetter,
ClosestPeakDirectionGetter,
DeterministicMaximumDirectionGetter)
from nilearn.image import index_img
from pynets.dmri.track import prep_tissues
from nibabel.streamlines.array_sequence import ArraySequence
from nipype.utils.filemanip import copyfile, fname_presuffix
recon_path_tmp_path = fname_presuffix(recon_path,
suffix=f"_{step_curv_combinations}",
newpath=cache_dir)
copyfile(recon_path, recon_path_tmp_path, copy=True, use_hardlink=False)
if waymask is not None:
waymask_tmp_path = fname_presuffix(waymask,
suffix=f"_{step_curv_combinations}",
newpath=cache_dir)
copyfile(waymask, waymask_tmp_path, copy=True, use_hardlink=False)
else:
waymask_tmp_path = None
tissue_img = nib.load(tissues4d)
# Order:
B0_mask = index_img(tissue_img, 0)
atlas_img = index_img(tissue_img, 1)
atlas_data_wm_gm_int = index_img(tissue_img, 2)
t1w2dwi = index_img(tissue_img, 3)
gm_in_dwi = index_img(tissue_img, 4)
vent_csf_in_dwi = index_img(tissue_img, 5)
wm_in_dwi = index_img(tissue_img, 6)
tiss_classifier = prep_tissues(t1w2dwi, gm_in_dwi, vent_csf_in_dwi,
wm_in_dwi, tiss_class, B0_mask)
B0_mask_data = np.asarray(B0_mask.dataobj).astype("bool")
atlas_data = np.array(atlas_img.dataobj).astype("uint16")
atlas_data_wm_gm_int_data = np.asarray(
atlas_data_wm_gm_int.dataobj).astype("bool").astype("int16")
# Build mask vector from atlas for later roi filtering
parcels = []
i = 0
intensities = [i for i in np.unique(atlas_data) if i != 0]
for roi_val in intensities:
parcels.append(atlas_data == roi_val)
i += 1
del atlas_data
parcel_vec = list(np.ones(len(parcels)).astype("bool"))
with h5py.File(recon_path_tmp_path, 'r+') as hf:
mod_fit = hf['reconstruction'][:].astype('float32')
hf.close()
print("%s%s" % ("Curvature: ", step_curv_combinations[1]))
# Instantiate DirectionGetter
if directget == "prob" or directget == "probabilistic":
dg = ProbabilisticDirectionGetter.from_shcoeff(
mod_fit,
max_angle=float(step_curv_combinations[1]),
sphere=sphere,
min_separation_angle=min_separation_angle,
)
elif directget == "clos" or directget == "closest":
dg = ClosestPeakDirectionGetter.from_shcoeff(
mod_fit,
max_angle=float(step_curv_combinations[1]),
sphere=sphere,
min_separation_angle=min_separation_angle,
)
elif directget == "det" or directget == "deterministic":
maxcrossing = 1
dg = DeterministicMaximumDirectionGetter.from_shcoeff(
mod_fit,
max_angle=float(step_curv_combinations[1]),
sphere=sphere,
min_separation_angle=min_separation_angle,
)
else:
raise ValueError("ERROR: No valid direction getter(s) specified.")
print("%s%s" % ("Step: ", step_curv_combinations[0]))
# Perform wm-gm interface seeding, using n_seeds at a time
seeds = utils.random_seeds_from_mask(
atlas_data_wm_gm_int_data > 0,
seeds_count=n_seeds_per_iter,
seed_count_per_voxel=False,
affine=np.eye(4),
)
if len(seeds) == 0:
print(
UserWarning("No valid seed points found in wm-gm "
"interface..."))
return None
# print(seeds)
# Perform tracking
if track_type == "local":
streamline_generator = LocalTracking(
dg,
tiss_classifier,
seeds,
np.eye(4),
max_cross=int(maxcrossing),
maxlen=int(max_length),
step_size=float(step_curv_combinations[0]),
fixedstep=False,
return_all=True,
)
elif track_type == "particle":
streamline_generator = ParticleFilteringTracking(
dg,
tiss_classifier,
seeds,
np.eye(4),
max_cross=int(maxcrossing),
step_size=float(step_curv_combinations[0]),
maxlen=int(max_length),
pft_back_tracking_dist=pft_back_tracking_dist,
pft_front_tracking_dist=pft_front_tracking_dist,
particle_count=particle_count,
return_all=True,
)
else:
try:
raise ValueError("ERROR: No valid tracking method(s) specified.")
except ValueError:
import sys
sys.exit(0)
# Filter resulting streamlines by those that stay entirely
# inside the brain
try:
roi_proximal_streamlines = utils.target(streamline_generator,
np.eye(4),
B0_mask_data,
include=True)
except BaseException:
print('No streamlines found inside the brain! ' 'Check registrations.')
return None
# Filter resulting streamlines by roi-intersection
# characteristics
try:
roi_proximal_streamlines = \
nib.streamlines.array_sequence.ArraySequence(
select_by_rois(
roi_proximal_streamlines,
affine=np.eye(4),
rois=parcels,
include=parcel_vec,
mode="%s" % ("any" if waymask is not None else
"both_end"),
tol=roi_neighborhood_tol,
)
)
print("%s%s" % ("Filtering by: \nNode intersection: ",
len(roi_proximal_streamlines)))
except BaseException:
print('No streamlines found to connect any parcels! '
'Check registrations.')
return None
try:
roi_proximal_streamlines = nib.streamlines. \
array_sequence.ArraySequence(
[
s for s in roi_proximal_streamlines
if len(s) >= float(min_length)
]
)
print(f"Minimum fiber length >{min_length}mm: "
f"{len(roi_proximal_streamlines)}")
except BaseException:
print('No streamlines remaining after minimal length criterion.')
return None
if waymask is not None and os.path.isfile(waymask_tmp_path):
from nilearn.image import math_img
mask = math_img("img > 0.0075", img=nib.load(waymask_tmp_path))
waymask_data = np.asarray(mask.dataobj).astype("bool")
try:
roi_proximal_streamlines = roi_proximal_streamlines[utils.near_roi(
roi_proximal_streamlines,
np.eye(4),
waymask_data,
tol=roi_neighborhood_tol,
mode="all")]
print("%s%s" %
("Waymask proximity: ", len(roi_proximal_streamlines)))
except BaseException:
print('No streamlines remaining in waymask\'s vacinity.')
return None
out_streams = [s.astype("float32") for s in roi_proximal_streamlines]
del dg, seeds, roi_proximal_streamlines, streamline_generator, \
atlas_data_wm_gm_int_data, mod_fit, B0_mask_data
os.remove(recon_path_tmp_path)
gc.collect()
try:
return ArraySequence(out_streams)
except BaseException:
return None
def dwi_dipy_run(dwi_dir,
node_size,
dir_path,
conn_model,
parc,
atlas_select,
network,
wm_mask=None):
from dipy.reconst.dti import TensorModel, quantize_evecs
from dipy.reconst.csdeconv import ConstrainedSphericalDeconvModel, recursive_response
from dipy.tracking.local import LocalTracking, ActTissueClassifier
from dipy.tracking import utils
from dipy.direction import peaks_from_model
from dipy.tracking.eudx import EuDX
from dipy.data import get_sphere, default_sphere
from dipy.core.gradients import gradient_table
from dipy.io import read_bvals_bvecs
from dipy.tracking.streamline import Streamlines
from dipy.direction import ProbabilisticDirectionGetter, ClosestPeakDirectionGetter, BootDirectionGetter
from nibabel.streamlines import save as save_trk
from nibabel.streamlines import Tractogram
##
dwi_dir = '/Users/PSYC-dap3463/Downloads/bedpostx_s002'
img_pve_csf = nib.load(
'/Users/PSYC-dap3463/Downloads/002_all/tmp/reg_a/t1w_vent_csf_diff_dwi.nii.gz'
)
img_pve_wm = nib.load(
'/Users/PSYC-dap3463/Downloads/002_all/tmp/reg_a/t1w_wm_in_dwi_bin.nii.gz'
)
img_pve_gm = nib.load(
'/Users/PSYC-dap3463/Downloads/002_all/tmp/reg_a/t1w_gm_mask_dwi.nii.gz'
)
labels_img = nib.load(
'/Users/PSYC-dap3463/Downloads/002_all/tmp/reg_a/dwi_aligned_atlas.nii.gz'
)
num_total_samples = 10000
tracking_method = 'boot' # Options are 'boot', 'prob', 'peaks', 'closest'
procmem = [2, 4]
##
if parc is True:
node_size = 'parc'
dwi_img = "%s%s" % (dwi_dir, '/dwi.nii.gz')
nodif_brain_mask_path = "%s%s" % (dwi_dir, '/nodif_brain_mask.nii.gz')
bvals = "%s%s" % (dwi_dir, '/bval')
bvecs = "%s%s" % (dwi_dir, '/bvec')
dwi_img = nib.load(dwi_img)
data = dwi_img.get_data()
[bvals, bvecs] = read_bvals_bvecs(bvals, bvecs)
gtab = gradient_table(bvals, bvecs)
gtab.b0_threshold = min(bvals)
sphere = get_sphere('symmetric724')
# Loads mask and ensures it's a true binary mask
mask_img = nib.load(nodif_brain_mask_path)
mask = mask_img.get_data()
mask = mask > 0
# Fit a basic tensor model first
model = TensorModel(gtab)
ten = model.fit(data, mask)
fa = ten.fa
# Tractography
if conn_model == 'csd':
print('Tracking with csd model...')
elif conn_model == 'tensor':
print('Tracking with tensor model...')
else:
raise RuntimeError("%s%s" % (conn_model, ' is not a valid model.'))
# Combine seed counts from voxel with seed counts total
wm_mask_data = img_pve_wm.get_data()
wm_mask_data[0, :, :] = False
wm_mask_data[:, 0, :] = False
wm_mask_data[:, :, 0] = False
seeds = utils.seeds_from_mask(wm_mask_data,
density=1,
affine=dwi_img.get_affine())
seeds_rnd = utils.random_seeds_from_mask(ten.fa > 0.02,
seeds_count=num_total_samples,
seed_count_per_voxel=True)
seeds_all = np.vstack([seeds, seeds_rnd])
# Load tissue maps and prepare tissue classifier (Anatomically-Constrained Tractography (ACT))
background = np.ones(img_pve_gm.shape)
background[(img_pve_gm.get_data() + img_pve_wm.get_data() +
img_pve_csf.get_data()) > 0] = 0
include_map = img_pve_gm.get_data()
include_map[background > 0] = 1
exclude_map = img_pve_csf.get_data()
act_classifier = ActTissueClassifier(include_map, exclude_map)
if conn_model == 'tensor':
ind = quantize_evecs(ten.evecs, sphere.vertices)
streamline_generator = EuDX(a=fa,
ind=ind,
seeds=seeds_all,
odf_vertices=sphere.vertices,
a_low=0.05,
step_sz=.5)
elif conn_model == 'csd':
print('Tracking with CSD model...')
response = recursive_response(
gtab,
data,
mask=img_pve_wm.get_data().astype('bool'),
sh_order=8,
peak_thr=0.01,
init_fa=0.05,
init_trace=0.0021,
iter=8,
convergence=0.001,
parallel=True)
csd_model = ConstrainedSphericalDeconvModel(gtab, response)
if tracking_method == 'boot':
dg = BootDirectionGetter.from_data(data,
csd_model,
max_angle=30.,
sphere=default_sphere)
elif tracking_method == 'prob':
try:
print(
'First attempting to build the direction getter directly from the spherical harmonic representation of the FOD...'
)
csd_fit = csd_model.fit(
data, mask=img_pve_wm.get_data().astype('bool'))
dg = ProbabilisticDirectionGetter.from_shcoeff(
csd_fit.shm_coeff, max_angle=30., sphere=default_sphere)
except:
print(
'Sphereical harmonic not available for this model. Using peaks_from_model to represent the ODF of the model on a spherical harmonic basis instead...'
)
peaks = peaks_from_model(
csd_model,
data,
default_sphere,
.5,
25,
mask=img_pve_wm.get_data().astype('bool'),
return_sh=True,
parallel=True,
nbr_processes=procmem[0])
dg = ProbabilisticDirectionGetter.from_shcoeff(
peaks.shm_coeff, max_angle=30., sphere=default_sphere)
elif tracking_method == 'peaks':
dg = peaks_from_model(model=csd_model,
data=data,
sphere=default_sphere,
relative_peak_threshold=.5,
min_separation_angle=25,
mask=img_pve_wm.get_data().astype('bool'),
parallel=True,
nbr_processes=procmem[0])
elif tracking_method == 'closest':
csd_fit = csd_model.fit(data,
mask=img_pve_wm.get_data().astype('bool'))
pmf = csd_fit.odf(default_sphere).clip(min=0)
dg = ClosestPeakDirectionGetter.from_pmf(pmf,
max_angle=30.,
sphere=default_sphere)
streamline_generator = LocalTracking(dg,
act_classifier,
seeds_all,
affine=dwi_img.affine,
step_size=0.5)
del dg
try:
del csd_fit
except:
pass
try:
del response
except:
pass
try:
del csd_model
except:
pass
streamlines = Streamlines(streamline_generator, buffer_size=512)
save_trk(Tractogram(streamlines, affine_to_rasmm=dwi_img.affine),
'prob_streamlines.trk')
tracks = [sl for sl in streamlines if len(sl) > 1]
labels_data = labels_img.get_data().astype('int')
labels_affine = labels_img.affine
conn_matrix, grouping = utils.connectivity_matrix(
tracks,
labels_data,
affine=labels_affine,
return_mapping=True,
mapping_as_streamlines=True,
symmetric=True)
conn_matrix[:3, :] = 0
conn_matrix[:, :3] = 0
return conn_matrix
def track_ensemble(target_samples, atlas_data_wm_gm_int, parcels, parcel_vec, mod_fit,
tiss_classifier, sphere, directget, curv_thr_list, step_list, track_type, maxcrossing, max_length,
n_seeds_per_iter=200):
from colorama import Fore, Style
from dipy.tracking import utils
from dipy.tracking.streamline import Streamlines, select_by_rois
from dipy.tracking.local import LocalTracking, ParticleFilteringTracking
from dipy.direction import ProbabilisticDirectionGetter, BootDirectionGetter, ClosestPeakDirectionGetter, DeterministicMaximumDirectionGetter
# Commence Ensemble Tractography
streamlines = nib.streamlines.array_sequence.ArraySequence()
ix = 0
circuit_ix = 0
stream_counter = 0
while int(stream_counter) < int(target_samples):
for curv_thr in curv_thr_list:
print("%s%s" % ('Curvature: ', curv_thr))
# Instantiate DirectionGetter
if directget == 'prob':
dg = ProbabilisticDirectionGetter.from_shcoeff(mod_fit, max_angle=float(curv_thr),
sphere=sphere)
elif directget == 'boot':
dg = BootDirectionGetter.from_shcoeff(mod_fit, max_angle=float(curv_thr),
sphere=sphere)
elif directget == 'closest':
dg = ClosestPeakDirectionGetter.from_shcoeff(mod_fit, max_angle=float(curv_thr),
sphere=sphere)
elif directget == 'det':
dg = DeterministicMaximumDirectionGetter.from_shcoeff(mod_fit, max_angle=float(curv_thr),
sphere=sphere)
else:
raise ValueError('ERROR: No valid direction getter(s) specified.')
for step in step_list:
print("%s%s" % ('Step: ', step))
# Perform wm-gm interface seeding, using n_seeds at a time
seeds = utils.random_seeds_from_mask(atlas_data_wm_gm_int > 0, seeds_count=n_seeds_per_iter,
seed_count_per_voxel=False, affine=np.eye(4))
if len(seeds) == 0:
raise RuntimeWarning('Warning: No valid seed points found in wm-gm interface...')
print(seeds)
# Perform tracking
if track_type == 'local':
streamline_generator = LocalTracking(dg, tiss_classifier, seeds, np.eye(4),
max_cross=int(maxcrossing), maxlen=int(max_length),
step_size=float(step), return_all=True)
elif track_type == 'particle':
streamline_generator = ParticleFilteringTracking(dg, tiss_classifier, seeds, np.eye(4),
max_cross=int(maxcrossing),
step_size=float(step),
maxlen=int(max_length),
pft_back_tracking_dist=2,
pft_front_tracking_dist=1,
particle_count=15, return_all=True)
else:
raise ValueError('ERROR: No valid tracking method(s) specified.')
# Filter resulting streamlines by roi-intersection characteristics
streamlines_more = Streamlines(select_by_rois(streamline_generator, parcels, parcel_vec.astype('bool'),
mode='any', affine=np.eye(4), tol=8))
# Repeat process until target samples condition is met
ix = ix + 1
for s in streamlines_more:
stream_counter = stream_counter + len(s)
streamlines.append(s)
if int(stream_counter) >= int(target_samples):
break
else:
continue
circuit_ix = circuit_ix + 1
print("%s%s%s%s%s" % ('Completed hyperparameter circuit: ', circuit_ix, '...\nCumulative Streamline Count: ',
Fore.CYAN, stream_counter))
print(Style.RESET_ALL)
print('\n')
return streamlines