def cc_seg_same_regions(vol_path, lr_region_path_suffix):
"""
Segment cc according to lr same regions.
Parameters
----------
vol_path: volume (T1w) data path
lr_region_path_suffix: same region streamlines of both hemispheres
Return
------
Density of lr-overlap streamlines mapping to the corpus callosum
"""
img = nib.load(vol_path)
counts = np.zeros(img.shape)
roi_name_list = os.listdir(lr_region_path_suffix)
roi_fib_path = os.path.join(lr_region_path_suffix, roi_name_list[0])
fibs_points = apply_affine(npl.inv(img.affine), Fasciculus(roi_fib_path).xmin_nodes())
for n in fibs_points:
counts[int(n[0]), int(n[1]), int(n[2])] += 1
if len(roi_name_list) == 2:
roi_fib_path = os.path.join(lr_region_path_suffix, roi_name_list[1])
fibs_points = apply_affine(npl.inv(img.affine), Fasciculus(roi_fib_path).xmin_nodes())
for m in fibs_points:
if counts[int(m[0]), int(m[1]), int(m[2])] != 0:
counts[int(m[0]), int(m[1]), int(m[2])] += 1
return counts
def cc_seg_mp(vol_path, streamlines_path, labels):
"""
Make probabilistic map (pm)
Parameters
----------
vol_path: volume (T1w) data path
streamlimes_path: streamlines path
labels: label of each streamline
Return
------
pm: probabilisic map
"""
img = nib.load(vol_path)
dim4 = (len(set(labels)),)
cc_mp = np.zeros(img.shape + dim4, dtype=float)
fasciculus = Fasciculus(streamlines_path)
streamlines = fasciculus.get_data()
fibs_points = apply_affine(npl.inv(img.affine), fasciculus.xmin_nodes()).astype(int)
for i in range(len(streamlines)):
index_i = fibs_points == np.array([fibs_points[i][0], fibs_points[i][1], fibs_points[i][2]])
index_i = index_i.sum(axis=1)
inddex = index_i == 3
voxel_counts = np.sum(index_i == 3)
dim4_value = []
for label in set(labels):
index_lab = labels == label
arr = np.vstack((inddex, index_lab)).sum(axis=0)
dim4_value.append(np.sum(arr == 2) / voxel_counts)
cc_mp[fibs_points[i][0], fibs_points[i][1], fibs_points[i][2]] = dim4_value
return cc_mp
def endpoints_axis2cc(vol_path, streamlines_path, axis='y', mode='max'):
"""
Endpoints of streamlines mapping to cc.
Parameters
----------
vol_path: volume (T1w) data path
streamlines_path: streamlines path
axis: 'x', 'y', 'z'
Choose which direction to project
mode: 'max', 'min', 'mean'
Choose which typ to project
Return
------
Density left and right endpoints, respectively
"""
img = nib.load(vol_path)
counts_l = np.zeros(img.shape)
counts_r = np.zeros(img.shape)
fibs_points = apply_affine(npl.inv(img.affine), Fasciculus(streamlines_path).xmin_nodes()).astype(int)
streamlines = Fasciculus(streamlines_path).sort_streamlines()
if axis == 'x':
endpoints_l = np.array([fib[0][0] for fib in streamlines])
endpoints_r = np.array([fib[-1][0] for fib in streamlines])
elif axis == 'y':
endpoints_l = np.array([fib[0][1] for fib in streamlines])
endpoints_r = np.array([fib[-1][1] for fib in streamlines])
else:
endpoints_l = np.array([fib[0][2] for fib in streamlines])
endpoints_r = np.array([fib[-1][2] for fib in streamlines])
for i in range(len(streamlines)):
index_i = fibs_points == np.array([fibs_points[i][0], fibs_points[i][1], fibs_points[i][2]])
index_i = index_i.sum(axis=1)
inddex = index_i == 3
if mode == 'max':
counts_l[fibs_points[i][0], fibs_points[i][1], fibs_points[i][2]] = endpoints_l[inddex].max()
counts_r[fibs_points[i][0], fibs_points[i][1], fibs_points[i][2]] = endpoints_r[inddex].max()
elif mode == 'min':
counts_l[fibs_points[i][0], fibs_points[i][1], fibs_points[i][2]] = endpoints_l[inddex].min()
counts_r[fibs_points[i][0], fibs_points[i][1], fibs_points[i][2]] = endpoints_r[inddex].min()
else:
counts_l[fibs_points[i][0], fibs_points[i][1], fibs_points[i][2]] = endpoints_l[inddex].mean()
counts_r[fibs_points[i][0], fibs_points[i][1], fibs_points[i][2]] = endpoints_r[inddex].mean()
return counts_l, counts_r
def muti_bundle_registration(paths_file, pts=12):
"""
muti-bundle registration and consolidation
Parameters
----------
paths_file: list; muti_bundle file path
pts: each streamline is divided into sections
Return
------
new header: include id of each streamline that comes from different subjects
registration and consolidation bundle
"""
fas = Fasciculus(paths_file[0])
# print fas.get_header()
bundle_header = {'fasciculus_id': None}
sub1 = fas.get_data()
bundle_header['fasciculus_id'] = len(sub1) * [
int(paths_file[0].split('/')[9])
]
sub2 = Fasciculus(paths_file[1]).get_data()
subj2_aligned = bundle_registration(sub1, sub2, pts=pts)
bundle = fas.fib_merge(sub1, subj2_aligned)
bundle_header['fasciculus_id'] += (
len(bundle) - len(sub1)) * [int(paths_file[1].split('/')[9])]
# print bundle_header
# print len(bundle)
for index in range(len(paths_file))[2:]:
# print paths_file[index]
sub = Fasciculus(paths_file[index]).get_data()
sub_aligned = bundle_registration(sub1, sub, pts=pts)
lenth = len(bundle)
bundle = fas.fib_merge(bundle, sub_aligned)
bundle_header['fasciculus_id'] += (
len(bundle) - lenth) * [int(paths_file[index].split('/')[9])]
fas.update_header(bundle_header)
fas.set_data(nibas.ArraySequence(bundle))
return fas
def cc_seg_region(vol_path, roi_fib_paths_suffix):
"""
Segment cc according to roi fibs
Parameters
----------
vol_path: volume (T1w) data path
roi_fib_paths_suffix: all streamlines selected by rois (the sum of same kinds of areas)
Return
------
Density of all streamlines mapping to the corpus callosum
"""
img = nib.load(vol_path)
counts = np.zeros(img.shape)
roi_name_list = os.listdir(roi_fib_paths_suffix)
for roi_i in range(len(roi_name_list)):
roi_fib_path = os.path.join(roi_fib_paths_suffix, roi_name_list[roi_i])
fibs = apply_affine(npl.inv(img.affine), Fasciculus(roi_fib_path).xmin_nodes())
for n in fibs:
counts[int(n[0]), int(n[1]), int(n[2])] = roi_i + 1
return counts
from pyfat.core.dataobject import Fasciculus
from dipy.tracking.utils import length
from pyfat.algorithm.fiber_clustering import FibClustering
from pyfat.viz.colormap import create_random_colormap
from pyfat.viz.fiber_simple_viz_advanced import fiber_simple_3d_show_advanced
fib = '/home/brain/workingdir/data/dwi/hcp/preprocessed/' \
'response_dhollander/101107/Diffusion/1M_20_01_20dynamic250_SD_Stream_rhemi_occipital5.tck'
# fib = '/home/brain/workingdir/data/dwi/hcp/preprocessed/' \
# 'response_dhollander/101107/Diffusion/1M_20_01_20dynamic250_SD_Stream_occipital8_lr5.tck'
img_path = '/home/brain/workingdir/data/dwi/hcp/preprocessed/' \
'response_dhollander/101107/Structure/T1w_acpc_dc_restore_brain1.25.nii.gz'
img = nib.load(img_path)
fa = Fasciculus(fib)
streamlines = fa.get_data()
length_t = fa.get_lengths()
ind = length_t > 10
streamlines = streamlines[ind]
fa.set_data(streamlines)
fibcluster = FibClustering(fa)
print len(streamlines)
# 1
qb = QuickBundles(streamlines, 2)
clusters = qb.clusters()
print qb.clusters_sizes()
indexs = []
for i in range(len(clusters)):
if clusters[i]['N'] >= 400:
'response_dhollander/100408/Native/100408.R.white.native.surf.gii'
geo_path = [lh_white, rh_white]
tck_path = '/home/brain/workingdir/data/dwi/hcp/preprocessed/' \
'response_dhollander/100408/result/result20vs45/cc_20fib_1.5lr_new_hierarchical_single_cc_splenium.tck'
# tck_path = '/home/brain/workingdir/data/dwi/hcp/preprocessed/' \
# 'response_dhollander/100408/result/result20vs45/cc_20fib_only_node.tck'
# tck_path = '/home/brain/workingdir/data/dwi/hcp/preprocessed/' \
# 'response_dhollander/100408/result/result20vs45/cc_20fib_step20_new_sample5000.tck'
# load data
data_path = '/home/brain/workingdir/data/dwi/hcp/preprocessed/' \
'response_dhollander/100408/Structure/T1w_acpc_dc_restore_brain.nii.gz'
img = nib.load(data_path)
fasciculus = Fasciculus(tck_path)
fibcluster = FibClustering(fasciculus)
# length_clusters = fibcluster.length_seg()
# streamlines = fasciculus.sort_streamlines()
# fasciculus.set_data(streamlines)
streamlines = fasciculus.get_data()
print len(streamlines)
# d = fibcluster.bundle_seg(streamlines, dist_thre=15)
# print len(d[1])
index_streams = fibcluster.bundle_thre_seg(streamlines)
print len(index_streams[1])
centroids = fibcluster.bundle_centroids(streamlines,
cluster_thre=10,
dist_thre=10)
subject_id = "100408"
def cc_seg_pipeline(tck_path, mask, geo_path, thr=5., roi_id=None, step=0):
"""
Extract occipital streamlines and separate hemisphere streamlines
Parameters
----------
tck_path: list; streamline file path (.tck)
mask: surface mask (xh.aparc.annot)
geo_path: geometry data path (subject_id.L.white.native.surf.gii)
thr: distance of streamline terminus to surface
Return
------
occipital streamlines
"""
# load data
fasciculus = Fasciculus(tck_path)
# extract streamlines from left to right hemisphere
fibselection = FibSelection(fasciculus)
L_temp_need0 = fibselection.endpoint_dissimilarity()
fasciculus.set_data(L_temp_need0)
L_temp_need1 = fibselection.single_point_mid_sag()
fasciculus.set_data(L_temp_need1)
del L_temp_need0, L_temp_need1
gc.collect()
# separation of lr_seed fib
fib = fasciculus.hemi_fib_separation()
# choose the mask: Occipital
mask_fib = []
for i in range(len(fib)):
vertices, colortable, label = nib.freesurfer.read_annot(mask[i])
gii_data = nib.load(geo_path[i]).darrays
coords, faces = gii_data[0].data, gii_data[1].data
label_value = np.array(len(coords) * [0])
if not roi_id:
label_value[vertices == 11] = 11 # lateraloccipital
label_value[vertices == 13] = 13 # lingual
label_value[vertices == 5] = 5 # cuneus
label_value[vertices == 21] = 21 # pericalcarine
else:
for value in roi_id:
label_value[vertices == value] = value
# extract fib using mask
streamlines = fib[i]
streamlines = fasciculus.sort_streamlines(streamlines)
if i == 0:
s0 = [s[0] for s in streamlines]
else:
s0 = [s[-1] for s in streamlines]
stream_terminus = np.array(s0)
dist = cdist(coords[label_value > 0], stream_terminus)
stream_index = np.array(len(streamlines) * [False])
for j in range(len(dist[:])):
temp_index = np.array(dist[j] <= thr)
stream_index += temp_index
rois_streamlines = streamlines[stream_index]
mask_fib.append(rois_streamlines)
# lr fib merge
fib_merge = fasciculus.fib_merge(mask_fib[0], mask_fib[1])
del mask_fib
gc.collect()
fib_hemi = fasciculus.separation_fib_to_hemi(fib_merge)
del fib_merge
gc.collect()
return fib_hemi
# !/usr/bin/python
# -*- coding: utf-8 -*-
from pyfat.core.dataobject import Fasciculus
from pyfat.algorithm.fiber_extract import FibSelection
# load data
# file = '/home/brain/workingdir/data/dwi/hcp/preprocessed/' \
# 'response_dhollander/100206/Diffusion/100k_sift_1M45006_dynamic250.tck'
file = '/home/brain/workingdir/data/dwi/hcp/preprocessed/' \
'response_dhollander/100408/result/result20vs45/Prob01_plenium_fib_vis.tck'
fasciculus = Fasciculus(file)
# extract CC
fibselection = FibSelection(fasciculus)
L_temp_need0 = fibselection.endpoint_dissimilarity()
fasciculus.set_data(L_temp_need0)
L_temp_need1 = fibselection.single_point_mid_sag()
fasciculus.set_data(L_temp_need1)
L_temp_need2 = fibselection.lr_step()
fasciculus.set_data(L_temp_need2)
out_path = '/home/brain/workingdir/data/dwi/hcp/preprocessed/' \
'response_dhollander/100408/result/result20vs45/Prob01_plenium_fib_vis_pure.tck'
fasciculus.save2tck(out_path)
# if __name__ == '__main__':
# fib_path = '/home/brain/workingdir/data/dwi/hcp/preprocessed/' \
# 'response_dhollander/100408/result/result20vs45/Prob01_plenium_fib.tck'
# out_path = '/home/brain/workingdir/data/dwi/hcp/preprocessed/' \
# 'response_dhollander/100408/result/result20vs45/Prob01_plenium_fib_vis.tck'
# fasciculus = Fasciculus(fib_path)
# !/usr/bin/python
# -*- coding: utf-8 -*-
from pyfat.io.load import load_tck
from pyfat.core.dataobject import Fasciculus
from pyfat.algorithm.fiber_extract import FibSelection
source_path = '/home/brain/workingdir/data/dwi/hcp/preprocessed/' \
'response_dhollander/101107/Diffusion/1M_20_01_20dynamic250_SD_Stream_rh_occipital8.tck'
fasciculus = Fasciculus(source_path)
fibs = FibSelection(fasciculus)
fib = fibs.lr_rat(5)
fasciculus.set_data(fib)
out_path = '/home/brain/workingdir/data/dwi/hcp/preprocessed/' \
'response_dhollander/101107/Diffusion/1M_20_01_20dynamic250_SD_Stream_rh_occipital8_lr5.tck'
fasciculus.save2tck(out_path)