def reduce3_to_bci_lh(reduce3labs):
class h32k:
pass
class h:
pass
class s:
pass
class bs:
pass
class bci:
pass
''' reduce3 to h32k'''
r3 = readdfs('lh.Yeo2011_17Networks_N1000_reduce3.dfs')
r3.labels = np.squeeze(reduce3labs.T)
'''h32k to full res FS'''
g_surf = nib.load('/big_disk/ajoshi/HCP_data/reference/100307/MNINonLinea\
r/Native/100307.L.very_inflated.native.surf.gii')
h.vertices = g_surf.darrays[0].data
h.faces = g_surf.darrays[1].data
h = interpolate_labels(r3, h)
''' native FS ref to native FS BCI'''
g_surf = nib.load('/big_disk/ajoshi/HCP_data/reference/100307/MNINon\
Linear/Native/100307.L.sphere.reg.native.surf.gii')
s.vertices = g_surf.darrays[0].data
s.faces = g_surf.darrays[1].data
s.labels = h.labels
''' map to bc sphere'''
bs.vertices, bs.faces = fsio.read_geometry('/big_disk/ajoshi/data/BCI\
_DNI_Atlas/surf/lh.sphere.reg')
bs = interpolate_labels(s, bs)
bci.vertices, bci.faces = fsio.read_geometry(
'/big_disk/ajoshi/data/BCI_DNI_A\
tlas/surf/lh.white')
bci.labels = bs.labels
# writedfs('BCI_orig_rh.dfs', bci)
bci.vertices, bci.faces = fsio.read_geometry(
'/big_disk/ajoshi/data/BCI_DNI_A\
tlas/surf/lh.inflated')
# view_patch(bci, bci.labels)
# writedfs('BCI_pial_rh.dfs.', bci)
bci.vertices, bci.faces = fsio.read_geometry('/big_disk/ajoshi/data/BCI_\
DNI_Atlas/surf/lh.white')
# writedfs('BCI_white_rh.dfs.', bci)
bci.vertices[:, 0] += 96 * 0.8
bci.vertices[:, 1] += 192 * 0.546875
bci.vertices[:, 2] += 192 * 0.546875
bci_bst = readdfs('/big_disk/ajoshi/data/BCI-DNI_brain_atlas/BCI-DNI_\
brain.left.inner.cortex.dfs')
bci_bst = interpolate_labels(bci, bci_bst)
labs = bci_bst.labels
return labs
def morph2dense(source_sphere,target_sphere,input_morph,path_output):
"""
This function maps a morphological file from a source surface to a target target surface.
Inputs:
*source_sphere: source surface.
*target_sphere: target surface.
*input_morph: morphological input file.
*path_output: path where output is saved.
created by Daniel Haenelt
Date created: 13-07-2019
Last modified: 13-07-2019
"""
import os
from nibabel.freesurfer.io import read_morph_data, write_morph_data, read_geometry
from scipy.interpolate import griddata
# make output folder
if not os.path.exists(path_output):
os.mkdir(path_output)
# transform morphological data to dense surfaces
pts_sphere_dense, _ = read_geometry(target_sphere)
pts_sphere, _ = read_geometry(source_sphere)
# get morphological data
morph = read_morph_data(input_morph)
# do the transformation
method = "nearest"
morph_dense = griddata(pts_sphere, morph, pts_sphere_dense, method)
# write dense morphological data
write_morph_data(os.path.join(path_output,os.path.basename(input_morph)), morph_dense)
def euclDist_infl(subject):
import numpy as np
import nibabel.freesurfer.io as fs
from scipy.spatial import distance_matrix
fsDir = '/afs/cbs.mpg.de/projects/mar004_lsd-lemon-preproc/freesurfer'
surfDir = '/afs/cbs.mpg.de/projects/mar005_lsd-lemon-surf/probands'
for hemi in ['lh', 'rh']:
# fsaverage5 coords on sphere
fsa5_sphere_coords = fs.read_geometry('%s/fsaverage5/surf/%s.sphere' % (fsDir, hemi))[0]
cort = fs.read_label('%s/fsaverage5/label/%s.cortex.label' % (fsDir, hemi))
# get corresponding nodes on subject sphere (find coords of high-dim subject surface closest to fsa5 nodes in sphere space)
subj_sphere_coords = fs.read_geometry('%s/%s/surf/%s.sphere' % (fsDir, subject, hemi))[0]
subj_indices = []
for node in cort:
dist2all = np.squeeze(distance_matrix(np.expand_dims(fsa5_sphere_coords[node], axis=0), subj_sphere_coords))
subj_indices.append(list(dist2all).index(min(dist2all)))
# pair-wise euclidean distance between included nodes on subject surface (midline)
subj_surf_coords = fs.read_geometry('%s/%s/surf/%s.inflated' % (fsDir, subject, hemi))[0]
euclDist = np.zeros((10242,10242))
euclDist[np.ix_(cort, cort)] = distance_matrix(subj_surf_coords[subj_indices,:],subj_surf_coords[subj_indices,:])
np.save('%s/%s/distance_maps/%s_%s_euclDist_inflated_fsa5' % (surfDir, subject, subject, hemi), euclDist)
def transfertodense(subjectid,vals,hemi,interptype='nearest',surftype='sphere'):
'''
def transfertodense(subjectid,vals,hemi,interptype='nearest',surftype='sphere'):
<subjectid> is like 'C0041'
<vals> is a column vector of values defined on the regular sphere surface (one hemi)
<hemi> is 'lh' or 'rh'
<interptype> is 'linear' or 'nearest', 'cubic'
<surftype> is 'sphere' (default),'inflated',etc...
Interpolate to obtain <vals> defined on the dense sphere surface.
Note that griddata in matlab can only do 'nearest' for 3d interpolation
griddata in scipy can do 'nearest','linear','cubic' for 3d interpolation
'''
from scipy.Interpolate import griddata
import nibabel.freesurfer.io as fsio
# calc
surf1file = cvnpath('freesurfer')/subjectid/'surf'/f'{hemi}.{surftype}'
surf1file = cvnpath('freesurfer')/subjectid/'surf'/f'{hemi}.{surftype}DENSE'
# load surfaces (note that we skip the post-processing of vertices and faces since unnecessary for what we are doing)
surf1.vertices, _ = fsio.read_geometry(surf1file)
surf2.vertices, _ = fsio.read_geometry(surf2file)
# do it
return griddata(surf1.vertices[:,0],surf1.vertices[:,1],surf1.vertices[:,2],vals.flatten(), \
surf2.vertices[:,0],surf2.vertices[:,1],surf2.vertices[:,2],method=interptype)
def transfer_mesh_color(subject_id,
atlas='aparc',
reconall_folder='/data01/ayagoz/HCP_1200/FS_reconall/',
concon_mesh='/home/kurmukov/HCP/Dan_iso5.m'):
'''
Transfer mesh labels from subject sphere mesh, to concon sphere mesh
Parameters:
subject_id - int,
subject id
atlas - str,
atlas to transfer, possible values: aparc (Desikan-Killiany), aparc.a2009s (Destrieux Atlas).
Defined by free surfer https://surfer.nmr.mgh.harvard.edu/fswiki/CorticalParcellation
reconall_folder - str,
path to recon-all FS output
concon_mesh - str,
path to ConCon sphere mesh
'''
lh_vertices, _ = read_geometry(
f'{reconall_folder}{subject_id}/surf/lh.sphere.reg')
rh_vertices, _ = read_geometry(
f'{reconall_folder}{subject_id}/surf/rh.sphere.reg')
lh_labels = load_surf_data(
f'{reconall_folder}{subject_id}/label/lh.{atlas}.annot')
rh_labels = load_surf_data(
f'{reconall_folder}{subject_id}/label/rh.{atlas}.annot')
lh_vertices /= 100
rh_vertices /= 100
lh_vertices_CC, _ = load_mesh_boris(concon_mesh)
rh_vertices_CC, _ = load_mesh_boris(concon_mesh)
knn = KNeighborsClassifier(n_neighbors=5,
weights='uniform',
metric='minkowski')
knn.fit(lh_vertices, lh_labels)
lh_labels_CC = knn.predict(lh_vertices_CC)
knn.fit(rh_vertices, rh_labels)
rh_labels_CC = knn.predict(rh_vertices_CC)
rh_labels_CC[rh_labels_CC != -1] += np.max(lh_labels_CC)
labels_CC = np.concatenate([lh_labels_CC, rh_labels_CC])
return labels_CC
def surface_intersect(inner_surface_path, outer_surface_path):
inner_mesh = trimesh.Trimesh(
*read_geometry(inner_surface_path, read_metadata=False))
outer_mesh = trimesh.Trimesh(
*read_geometry(outer_surface_path, read_metadata=False))
ray_interceptor_inner = trimesh.ray.ray_pyembree.RayMeshIntersector(
inner_mesh)
inds = ray_interceptor_inner.intersects_location(
ray_origins=outer_mesh.vertices,
ray_directions=outer_mesh.vertex_normals,
multiple_hits=True)[1]
return len(inds) != 0
def calctransferfunctions(fslhfile, fsrhfile, sslhfile, ssrhfile):
'''
cvncalctransferfunctions(fslhfile,fsrhfile,sslhfile,ssrhfile)
<fslhfile>,<fsrhfile> are locations of the fsaverage spherical surfaces
<sslhfile>,<ssrhfile> are locations of other surfaces registered on the sphere to fsaverage
return functions that perform nearest-neigbor interpolation
to go back and forth between values defined on the surfaces.
# History
20180714 now accept pathlib path for all 4 input
'''
from nibabel.freesurfer.io import read_geometry
from scipy.interpolate import griddata
# load spherical surfaces (note that we skip the post-processing of vertices and faces since unnecessary for what we are doing)
fslh_vertices, _ = read_geometry(str(fslhfile))
fsrh_vertices, _ = read_geometry(str(fsrhfile))
sslh_vertices, _ = read_geometry(str(sslhfile))
ssrh_vertices, _ = read_geometry(str(ssrhfile))
# define the functions
tempix = griddata(fslh_vertices,
np.arange(fslh_vertices.shape[0]) + 1,
sslh_vertices,
method='nearest')
tfunFSSSlh = lambda x: x[tempix].flatten().astype('float')
tempix = griddata(fslh_vertices,
np.arange(fslh_vertices.shape[0]) + 1,
ssrh_vertices,
method='nearest')
tfunFSSSrh = lambda x: x[tempix].flatten().astype('float')
tempix = griddata(sslh_vertices,
np.arange(sslh_vertices.shape[0]) + 1,
fslh_vertices,
method='nearest')
tfunSSFSlh = lambda x: x[tempix].flatten().astype('float')
tempix = griddata(ssrh_vertices,
np.arange(ssrh_vertices.shape[0]) + 1,
fsrh_vertices,
method='nearest')
tfunSSFSrh = lambda x: x[tempix].flatten().astype('float')
return tfunFSSSlh, tfunFSSSrh, tfunSSFSlh, tfunSSFSrh
def __init__(self, files):
if isinstance(files, tuple):
white = read_geometry(files[0])
pial = read_geometry(files[1])
vertices = (pial[0] - white[0]) / 2
faces = pial[1]
else:
mid = read_geometry(files)
vertices = mid[0]
faces = mid[1]
self.vertices = vertices
self.faces = faces
self.mesh = Trimesh(vertices=self.vertices, faces=self.faces)
def calculate_area(filename_surf, filename_area=""):
"""
The function calculates vertex-wise surface area. The code is taken from the octave code
surf2area.m from Anderson Winkler found in his github repository (https://github.com/
andersonwinkler/areal). Consider a triangular face ABC with corner points
a = [x_A, y_A, z_A]'
b = [x_B, y_B, z_B]'
c = [x_C, y_C, z_C]'
The area for this triangle is given by the normed cross product A = |u x v|/2 with u = a - c and
v = b - c. This is a face-wise surface area representation. To convert this to a vertex-wise
representation, we assign each vertex one third of the sum of the areas of all faces that meet
at that vertex. Cf. Anderson Winkler et al. Measuring and comparing brain cortical surface area
and other areal quantities, Neuroimage 61(4), p. 1428-1443 (2012).
Inputs:
*file_surf: input file geometry on which surface area is calculated.
*file_area: file name of the surface area file.
Outputs:
*dpv: vertex-wise surface area.
created by Daniel Haenelt
Date created: 01-11-2018
Last modified: 17-12-2018
"""
import numpy as np
from numpy.linalg import norm
from nibabel.freesurfer.io import write_morph_data, read_geometry
# Read the surface file
vtx, fac = read_geometry(filename_surf)
nV = len(vtx)
nF = len(fac)
# compute area per face (DPF)
facvtx = np.concatenate([vtx[fac[:, 0]], vtx[fac[:, 1]], vtx[fac[:, 2]]],
axis=1)
facvtx0 = facvtx[:, 0:6] - np.concatenate(
[facvtx[:, 6:9], facvtx[:, 6:9]], axis=1) # place 3rd vtx at origin
cp = np.cross(facvtx0[:, 0:3], facvtx0[:, 3:6], axisa=1,
axisb=1) # cross product
dpf = norm(cp, axis=1) / 2 # half of the norm
print("Total area (facewise): " + str(np.sum(dpf)))
# compute area per vertex (DPV)
dpv = np.zeros(nV)
# for speed, divide the dpf by 3
dpf = dpf / 3
# redistribute
for f in range(nF):
dpv[fac[f, :]] = dpv[fac[f, :]] + dpf[f]
print("Total area (vertexwise): " + str(np.sum(dpv)))
# save dpv
if filename_area:
write_morph_data(filename_area, dpv)
# return vertex-wise surface area
return dpv
def hemisphere(request, image="", hemisphere=""):
fileUrl = f"https://neurovault.org/api/images/{image}"
try:
with urllib.request.urlopen(fileUrl) as url:
fileData = json.loads(url.read().decode())
except (urllib.error.HTTPError, ValueError):
fileData = None
# Map external file to internal file:
surface_file = fileData[f"surface_{hemisphere}_file"]
giftiParser = GiftiImageParser(buffer_size=35000000)
giftiParser.parse(string=urlopen(surface_file).read())
giftiObject = giftiParser.img
colors = giftiObject.darrays[0].data
if hemisphere not in ["left", "right"]:
exit(1)
elif hemisphere == "left":
hemi_short = "lh"
else:
hemi_short = "rh"
fs_base = os.path.join(settings.STATIC_ROOT, "fs/")
verts, faces = fsio.read_geometry(
os.path.join(fs_base, "%s.pial" % hemi_short))
bytestream = bytes(fv_scalar_to_collada(verts, faces, colors).getvalue())
filename = f"{hemisphere}.dae"
file = ContentFile(bytestream, filename)
response = HttpResponse(file, content_type="application/xml")
response["Content-Disposition"] = "attachment; filename=" + filename
return response
def read(self, surface_path, use_center_surface):
vertices, triangles, metadata = read_geometry(surface_path,
read_metadata=True)
self.logger.info("From the file %s the extracted metadata is %s",
surface_path, metadata)
if use_center_surface:
cras = [0, 0, 0]
self.logger.info(
"The --center_ras flag was specified, so the ras centering point is %s",
cras)
else:
if CENTER_RAS_FS_SURF in metadata:
cras = metadata[CENTER_RAS_FS_SURF]
self.logger.info(
"The ras centering point for surface %s is %s",
surface_path, cras)
else:
cras = [0, 0, 0]
self.logger.warning(
"Could not read the ras centering point from surface %s header. "
"The cras will be %s", surface_path, cras)
return Surface(vertices,
triangles,
area_mask=None,
center_ras=cras,
generic_metadata=metadata)
def hemisphere(request, image='', hemisphere=''):
# query neurovault image
fileUrl = f"https://neurovault.org/api/images/{image}"
try:
with urllib.request.urlopen(fileUrl) as url:
fileData = json.loads(url.read().decode())
except (urllib.error.HTTPError, ValueError):
fileData = None
surface = fileData[f"surface_{hemisphere}_file"]
dom = parse(urllib.request.urlopen(surface))
zip_base64 = dom.getElementsByTagName('Data')[0].childNodes[0].data
zip = base64.b64decode(zip_base64.encode('ascii'))
unzip = zlib.decompress(zip)
colors = numpy.fromstring(unzip, dtype='float32')
if hemisphere not in ["left","right"]:
print("Bad hemisphere input")
exit(1)
elif hemisphere == "left":
hemi_short = "lh"
else:
hemi_short = "rh"
fs_base = os.path.join(settings.BASE_DIR, 'staticfiles/fs/')
verts,faces = fsio.read_geometry(os.path.join(fs_base,"%s.pial" % hemi_short))
bytestream = bytes(fv_scalar_to_collada(verts,faces,colors).getvalue())
filename = f"{hemisphere}.dae"
file = ContentFile(bytestream, filename)
response = HttpResponse(file, content_type='application/xml')
response['Content-Disposition'] = 'attachment; filename=' + filename
return response
def spherically_project_surface(insurf, outsurf):
""" (string) -> None
takes path to insurf, spherically projects it, outputs it to outsurf
"""
surf = fs.read_geometry(insurf, read_metadata=True)
projected = sphericalProject(surf[0], surf[1])
fs.write_geometry(outsurf, projected[0], projected[1], volume_info=surf[2])
def get_pial_meshes(age=None,
template=None,
face_count=20000,
subjects_dir=None):
if template is None:
if age is not None:
template = f"ANTS{age}-0Months3T"
else:
raise ValueError("The age or the template must be specified.")
if subjects_dir is None:
subjects_dir = Path(os.environ["SUBJECTS_DIR"])
mesh_pattern = "{}/{}/surf/{}.pial"
vertices = {}
faces = {}
for hemi in ["lh", "rh"]:
vertices_hemi, faces_hemi = read_geometry(
mesh_pattern.format(subjects_dir, template, hemi))
open3d_mesh = open3d.geometry.TriangleMesh(
vertices=open3d.utility.Vector3dVector(vertices_hemi),
triangles=open3d.utility.Vector3iVector(faces_hemi))
mesh = open3d_mesh.simplify_quadric_decimation(int(face_count / 2))
vertices[hemi] = np.asarray(mesh.vertices)
faces[hemi] = np.asarray(mesh.triangles)
return vertices, faces
def mrtrix_mesh2vox(surface_path, template_path, temp_dir, output_prefix):
"""
Create a partial volume map from a surface and a reference template using mrtrix mesh2voxel command.
:param surface_path: path to the surface file
:param template_path: path to the template file
:param temp_dir: path to temporary directory to which temporary files are saved
:param output_prefix: prefix to output file
"""
# Adapt affine translation using metadata
template = nib.load(template_path)
_, _, meta = read_geometry(surface_path, read_metadata=True)
template = nib.as_closest_canonical(template)
affine = template.affine.copy()
affine[:-1, -1] = template.affine[:-1, -1] - meta['cras']
new_template = nib.Nifti1Image(template.dataobj, affine)
new_template_path = temp_dir / 'template.mgz'
nib.save(new_template, new_template_path)
# Reconstruct volume from mesh
subprocess.run([
'mesh2voxel', surface_path, new_template_path,
temp_dir / f'{output_prefix}_output.mgz'
])
# Save the reconstructed volume with the right affine
output = nib.load(temp_dir / f'{output_prefix}_output.mgz')
new_output = nib.Nifti1Image(output.dataobj, template.affine)
# nib.save(new_output, output_path)
return new_output
def main():
# define specific freesurfer polydata surface to read
hemi = 'lh' # lh or rh for left or right hemispheres
surf = 'inflated' # all surfaces exported by freesurfer, e.g., pial, sphere, smoothwm, curv, inflated...
subj = 5
fs_file = '{}.{}'.format(hemi, surf)
fs_dir = os.environ[
'OneDrive'] + r'\data\nexstim_coord\freesurfer\ppM1_S{}\surf'.format(
subj)
fs_path = os.path.join(fs_dir, fs_file)
vertices, faces, volume_info = fsio.read_geometry(fs_path,
read_metadata=True)
# create the methods to convert the nifti points to vtk object
polydata = vtk.vtkPolyData()
points = vtk.vtkPoints()
polys = vtk.vtkCellArray()
scalars = vtk.vtkFloatArray()
colors = vtk.vtkNamedColors()
# load the point, cell, and data attributes
for n, xi in enumerate(vertices):
points.InsertPoint(n, xi)
scalars.InsertTuple1(n, n)
for fc in faces:
polys.InsertNextCell(make_vtk_id_list(fc))
# assign the pieces to the vtkPolyData
polydata.SetPoints(points)
polydata.SetPolys(polys)
polydata.GetPointData().SetScalars(scalars)
# visualize
mapper = vtk.vtkPolyDataMapper()
mapper.SetInputData(polydata)
mapper.SetScalarRange(polydata.GetScalarRange())
actor = vtk.vtkActor()
actor.SetMapper(mapper)
ren = vtk.vtkRenderer()
ren.AddActor(actor)
ren.SetBackground(colors.GetColor3d("Cornsilk"))
ren_win = vtk.vtkRenderWindow()
ren_win.AddRenderer(ren)
ren_win.SetSize(800, 800)
# create the window interactor
iren = vtk.vtkRenderWindowInteractor()
iren.SetRenderWindow(ren_win)
iren.Initialize()
ren_win.Render()
iren.Start()
close_window(iren)
def get_b0_orientation(surf_in,
vol_in,
write_output=False,
path_output="",
name_output=""):
"""
This function computes the angle between surface normals and B0-direction per vertex.
Inputs:
*surf_in: input of surface mesh.
*vol_in: input of corresponding nifti volume.
*write output: write out to disk (boolean).
*path_output: path where to save output.
*name_output: basename of output file.
Outputs:
*theta: angle in radians.
created by Daniel Haenelt
Date created: 31-07-2020
Last modified: 31-07-2020
"""
import os
import numpy as np
import nibabel as nb
from nibabel.affines import apply_affine
from nibabel.freesurfer.io import read_geometry, write_morph_data
from lib.io.get_filename import get_filename
from lib.surface.vox2ras import vox2ras
from lib_gbb.normal import get_normal
# make subfolders
if write_output and not os.path.exists(path_output):
os.makedirs(path_output)
# get hemi from surface filename
_, hemi, _ = get_filename(surf_in)
# load surface
vtx, fac = read_geometry(surf_in)
# get transformation matrix
_, r2v = vox2ras(vol_in) # ras-tkr -> voxel
v2s = nb.load(vol_in).affine # voxel -> scanner-ras
M = v2s.dot(r2v)
# apply affine transformation
vtx = apply_affine(M, vtx)
# get surface normals
n = get_normal(vtx, fac)
# get angle between b0 and surface normals in radians
theta = np.arccos(np.dot(n, [0, 0, 1]))
# write output
if write_output:
write_morph_data(os.path.join(path_output, hemi + "." + name_output),
theta)
return theta
def parse_fs(self, surface_path):
(vertices, triangles, metadata) = read_geometry(surface_path, read_metadata=True)
cras = metadata['cras']
logger = get_logger(__name__)
logger.info("From the file %s the extracted metadata is %s" % (surface_path, metadata))
return Surface(vertices, triangles, cras, metadata)
def write_surface(image, filename):
extension = filename.split('.')[-1]
if extension == 'mha':
sitk.WriteImage(image, filename)
elif extension == 'annot':
print(type(image[1]))
io.write_annot(filename, image[0], image[1], image[2])
elif extension == 'label':
raise ValueError('Reader for extensions \'label\' not yet implemented')
elif extension in ['inflated', 'pial', 'white']:
io.read_geometry(filename, image[0], image[1])
else:
return io.write_morph_data(filename, image)
def euclDist(subject):
import numpy as np
import nibabel.freesurfer.io as fs
from scipy.spatial import distance_matrix
fsDir = '/afs/cbs.mpg.de/projects/mar004_lsd-lemon-preproc/freesurfer'
surfDir = '/afs/cbs.mpg.de/projects/mar005_lsd-lemon-surf/probands'
for hemi in ['lh', 'rh']:
# fsaverage5 coords on sphere
fsa5_sphere_coords = fs.read_geometry('%s/fsaverage5/surf/%s.sphere' %
(fsDir, hemi))[0]
cort = fs.read_label('%s/fsaverage5/label/%s.cortex.label' %
(fsDir, hemi))
# get corresponding nodes on subject sphere (find coords of high-dim subject surface closest to fsa5 nodes in sphere space)
subj_sphere_coords = fs.read_geometry('%s/%s/surf/%s.sphere' %
(fsDir, subject, hemi))[0]
subj_indices = []
for node in cort:
dist2all = np.squeeze(
distance_matrix(
np.expand_dims(fsa5_sphere_coords[node], axis=0),
subj_sphere_coords))
subj_indices.append(list(dist2all).index(min(dist2all)))
# pair-wise euclidean distance between included nodes on subject surface (midline)
subj_surf_coords_pial = fs.read_geometry('%s/%s/surf/%s.pial' %
(fsDir, subject, hemi))[0]
subj_surf_coords_wm = fs.read_geometry('%s/%s/surf/%s.smoothwm' %
(fsDir, subject, hemi))[0]
subj_surf_coords = (subj_surf_coords_pial + subj_surf_coords_wm) / 2.
euclDist = np.zeros((10242, 10242))
euclDist[np.ix_(cort, cort)] = distance_matrix(
subj_surf_coords[subj_indices, :],
subj_surf_coords[subj_indices, :])
np.save(
'%s/%s/distance_maps/%s_%s_euclDist_fsa5' %
(surfDir, subject, subject, hemi), euclDist)
def fetch_template_surface(
template: str,
join: bool = True,
layer: Optional[str] = None,
data_dir: Optional[Union[str, Path]] = None,
) -> Union[BSPolyData, Tuple[BSPolyData, BSPolyData]]:
"""Loads surface templates.
Parameters
----------
template : str
Name of the surface template. Valid values are "fslr32k", "fsaverage",
"fsaverage3", "fsaverage4", "fsaverage5", "fsaverage6", "civet41k",
"civet164k".
join : bool, optional
If true, returns surfaces as a single object, if false, returns an
object per hemisphere, by default True.
layer : str, optional
Name of the cortical surface of interest. Valid values are "white",
"smoothwm", "pial", "inflated", "sphere" for fsaverage surfaces;
"midthickness", "inflated", "vinflated" for "fslr32k"; "mid", "white"
for CIVET surfaces; and "sphere" for "civet41k". If None,
defaults to "pial" or "midthickness", by default None.
data_dir : str, Path, optional
Directory to save the data, by default
$HOME_DIR/brainstat_data/surface_data.
Returns
-------
BSPolyData or tuple of BSPolyData
Output surface(s). If a tuple, then the first element is the left
hemisphere.
"""
data_dir = Path(
data_dir) if data_dir else data_directories["SURFACE_DATA_DIR"]
surface_files = _fetch_template_surface_files(template, data_dir, layer)
if template[:9] == "fsaverage":
surfaces_fs = [read_geometry(file) for file in surface_files]
surfaces = [
build_polydata(surface[0], surface[1]) for surface in surfaces_fs
]
elif template == "fslr32k":
surfaces = [read_surface(file) for file in surface_files]
else:
surfaces_obj = [read_civet(file) for file in surface_files]
surfaces = [
build_polydata(surface[0], surface[1]) for surface in surfaces_obj
]
if join:
return combine_surfaces(surfaces[0], surfaces[1])
else:
return surfaces[0], surfaces[1]
def load_surface(lr):
all_coords = []
for surf_type in ['white', 'pial']:
coords, faces = read_geometry(
'/data_dir/freesurfer/'
'subjects/fsaverage/surf/lh.{surf_type}'.format(
surf_type=surf_type))
all_coords.append(coords)
coords = np.array(all_coords).astype(np.float).mean(axis=0)
surf = Surface(coords, faces)
return surf
def load_freesurfer_geometry(filename, to='mesh', warn=False):
'''
load_freesurfer_geometry(filename) yields the data stored at the freesurfer geometry file given
by filename. The optional argument 'to' may be used to change the kind of data that is
returned.
The following are valid settings for the 'to' keyword argument:
* 'mesh' (the default) yields a mesh object
* 'tess' yields a tess object (discarding coordinates)
* 'raw' yields a tuple of numpy arrays, identical to the read_geometry return value.
'''
if not warn:
with warnings.catch_warnings():
warnings.filterwarnings('ignore',
category=UserWarning,
module='nibabel')
(xs, fs, info) = fsio.read_geometry(filename, read_metadata=True)
else:
(xs, fs, info) = fsio.read_geometry(filename, read_metadata=True)
# see if there's chirality data here...
filename = os.path.split(filename)[1]
filename = filename.lower()
if filename.startswith('lh'): info['chirality'] = 'lh.'
elif filename.startswith('rh'): info['chirality'] = 'rh.'
# parse it into something
to = to.lower()
if to in ['mesh', 'auto', 'automatic']:
return geo.Mesh(fs, xs, meta_data=info)
elif to in ['tess', 'tesselation']:
return geo.Tesselation(fs, meta_data=info)
elif to in ['coords', 'coordinates']:
return xs
elif to in ['triangles', 'faces']:
return fs
elif to in ['meta', 'meta_data']:
return info
elif to =='raw':
return (xs, fs)
else:
raise ValueError('Could not understand \'to\' argument: %s' % to)
def main():
parser = create_parser()
args = parser.parse_args()
print 'fname:', args.fname
g = fio.read_geometry(args.fname)
m = Trimesh(g[0], g[1])
if args.outname is None:
# (name, ext) = os.path.splitext(args.fname)
outname = '%s.stl' % args.fname
else:
outname = args.outname
print 'outname:', outname
export_mesh(m, outname)
def main(): parser = create_parser() args = parser.parse_args() print 'fname:', args.fname g = fio.read_geometry(args.fname) m = Trimesh(g[0], g[1]) if args.outname is None: # (name, ext) = os.path.splitext(args.fname) outname = '%s.stl' % args.fname else: outname = args.outname print 'outname:', outname export_mesh(m, outname)
def main():
parser = create_parser()
args = parser.parse_args()
print 'fname:', args.fname
g = fio.read_geometry(args.fname)
if args.outname is None:
outname = '%s_vtk' % args.fname
else:
outname = args.outname
print 'outname:', outname
rnd = (np.random.random((g[0].shape[0])) * 255).astype(np.uint8)
pointData = {'Colors': (rnd, rnd, rnd)}
meshToVTK(outname, g[0], g[1], pointData)
def main():
parser = create_parser()
args = parser.parse_args()
print 'fname:', args.fname
g = fio.read_geometry(args.fname)
if args.outname is None:
outname = '%s_vtk' % args.fname
else:
outname = args.outname
print 'outname:', outname
rnd = (np.random.random((g[0].shape[0])) * 255).astype(np.uint8)
pointData = {'Colors': (rnd, rnd, rnd)}
meshToVTK(outname, g[0], g[1], pointData)
def read_freesurfer(cls, filename):
"""Read a triangular format Freesurfer surface mesh.
Parameters
----------
filename : str
Filename for the file with the triangular data
Returns
-------
surface : TriSurface
"""
vertices, faces = read_geometry(filename)
return cls(vertices=vertices, faces=faces)
def read(fn):
'''General read function for surfaces
For now only supports ascii (as used in AFNI's SUMA), Caret
and freesurfer formats
'''
if fn.endswith('.asc'):
from mvpa2.support.nibabel import surf_fs_asc
return surf_fs_asc.read(fn)
elif fn.endswith('.coord'):
from mvpa2.support.nibabel import surf_caret
return surf_caret.read(fn)
else:
import nibabel.freesurfer.io as fsio
coords, faces = fsio.read_geometry(fn)
return Surface(coords, faces)
def fs_to_dae( args ):
#load in FS mesh
verts,faces = fsio.read_geometry( args.input )
#dumb copypasta for mesh face normals
norms = np.zeros( verts.shape, dtype=verts.dtype )
tris = verts[faces]
n = np.cross( tris[::,1 ] - tris[::,0] , tris[::,2 ] - tris[::,0] )
norm_sizes = np.sqrt(n[:,0]**2 + n[:,1]**2 + n[:,2]**2)
for i in range(3):
n[:,i] = n[:,i] / norm_sizes
del norm_sizes
#map back to vertices
norms[ faces[:,0] ] += n
norms[ faces[:,1] ] += n
norms[ faces[:,2] ] += n
del n
norm_sizes = np.sqrt(norms[:,0]**2 + norms[:,1]**2 + norms[:,2]**2)
for i in range(3):
norms[:,i] = norms[:,i] / norm_sizes
#color
if not args.color:
color = np.ones(norms.shape) * 0.4
else:
scalars = fsio.read_morph_data(args.color)
color = color_func(scalars)
#make trimesh
mesh = trimesh.Trimesh(\
verts,\
faces,\
vertex_normals=norms,\
vertex_colors=color)
mesh.export(file_obj=args.output, file_type="collada")
#mesh.export(file_obj=args.output, file_type="obj")
#mesh.export_gltf(file_obj=args.output)
return
def plot_normal_direction(input_surf, axis=2):
""" Plot normal direction
This function plots the direction from the white surface towards the pial
surface based on the white surface vertex normals along one axis.
Parameters
----------
input_surf : str
Filename of source mesh (white surface).
axis : int, optional
Axis for distance calculation in ras space (0,1,2). The default is 2.
Returns
-------
None.
Notes
-------
created by Daniel Haenelt
Date created: 13-12-2019
Last modified: 05-10-2020
"""
# fixed parameter
line_threshold = 0.05 # if direction is along one axis, omit line if length is below threshold
# load geometry
vtx_source, fac_source = read_geometry(input_surf)
# get surface normals per vertex
norm = get_normal(vtx_source, fac_source)
# get distance along one axis
r_dist = norm[:,axis].copy()
# get directions
r_dist[r_dist > line_threshold] = 1
r_dist[r_dist < -line_threshold] = -1
r_dist[np.abs(r_dist) != 1] = 0
# write output
header = nb.freesurfer.mghformat.MGHHeader()
output = nb.freesurfer.mghformat.MGHImage(r_dist, np.eye(4), header)
nb.save(output, input_surf+"_plot_normal_dir"+str(axis)+".mgh")
def __init__(self, fsaverage_dir, hemisphere, atlas):
# Load patient labels for vertex based on <atlas> parcels (MACRO)
(self.vertices, self.colortable, self.labels) = fio.read_annot(
f"{fsaverage_dir}/label/{hemisphere}.{atlas}.annot")
# Load geometry of the brain
(self.coords, self.faces) = fio.read_geometry(f"{fsaverage_dir}/surf/{hemisphere}.sphere.reg")
# an array of list: at position i-th there is the list of faces that the vertex i touches
self.vertex_to_faces = np.empty((len(self.vertices),), dtype=object)
for i in range(len(self.vertex_to_faces)):
self.vertex_to_faces[i] = []
# For each face, add the face i to the vertex i-th list of faces
for i, f in enumerate(self.faces):
for j in range(3):
self.vertex_to_faces[f[j]].append(i)
def read_surface(filename):
extension = filename.split('.')[-1]
if extension == 'mha':
img = sitk.ReadImage(filename)
return sitk.GetArrayFromImage(img)
elif extension == 'annot':
raise ValueError('Reader for extensions \'annot\' not yet implemented')
elif extension == 'label':
raise ValueError('Reader for extensions \'label\' not yet implemented')
elif extension in ['inflated', 'pial', 'white']:
coords, faces = io.read_geometry(filename)
return coords, faces
else:
return io.read_morph_data(filename)
def read(self, surface_path, use_center_surface):
vertices, triangles, metadata = read_geometry(
surface_path, read_metadata=True)
self.logger.info(
"From the file %s the extracted metadata is %s", surface_path, metadata)
if use_center_surface:
cras = [0, 0, 0]
self.logger.info(
"The --center_ras flag was specified, so the ras centering point is %s", cras)
else:
if CENTER_RAS_FS_SURF in metadata:
cras = metadata[CENTER_RAS_FS_SURF]
self.logger.info(
"The ras centering point for surface %s is %s", surface_path, cras)
else:
cras = [0, 0, 0]
self.logger.warning("Could not read the ras centering point from surface %s header. "
"The cras will be %s", surface_path, cras)
return Surface(vertices, triangles, area_mask=None,
center_ras=cras, generic_metadata=metadata)
def connectivity_2_tvb_fs(subID, subFolder, SC_matrix, reconallFolder='recon_all'):
# Create the results folder
if not os.path.exists(subFolder + 'results/'):
os.mkdir(subFolder + 'results/')
# Load the SC matrix
SC = io.loadmat(subFolder + '/mrtrix_68/tracks_68/' + SC_matrix)
weights = SC['SC_cap_agg_bwflav2']
delay = SC['SC_dist_agg_mean']
# Load the required things compiuted previously by FREESURFER
lh_vert, lh_faces = fs.read_geometry(subFolder + '/' + reconallFolder + '/surf/lh.pial')
rh_vert, rh_faces = fs.read_geometry(subFolder + '/' + reconallFolder + '/surf/rh.pial')
cortexMesh = {'vertices': np.vstack((lh_vert, rh_vert)),
'faces': np.vstack((lh_faces, rh_faces + np.shape(lh_vert)[0]))}
# Calculate vertex-normals
cortexMesh['vertexNormals'] = calcVertNormals(cortexMesh['vertices'], cortexMesh['faces'])
# Load annotation tables
lh_labels, lh_ctab, lh_names = fs.read_annot(subFolder + '/' + reconallFolder + '/label/lh.aparc.annot')
rh_labels, rh_ctab, rh_names = fs.read_annot(subFolder + '/' + reconallFolder + '/label/rh.aparc.annot')
# Remove the CC i.e. correct labeling
lh_labels[lh_labels > 3] -= 1
rh_labels[rh_labels > 3] -= 1
# Combine into single vectors
rh_labels += np.max(lh_labels)
rh_labels[rh_labels == np.min(rh_labels)] = -1
cortexMesh['labels'] = np.hstack((lh_labels, rh_labels))
# Store label name-strings
tmp = lh_names[1:4] + lh_names[5:]
tmp_lh = ['lh_' + s for s in tmp]
tmp_rh = ['rh_' + s for s in tmp]
cortexMesh['labelNames'] = tmp_lh + tmp_rh
# Do the TVB mesh clean
cortexMesh['vertices'], cortexMesh['faces'], cortexMesh['vertexNormals'], cortexMesh['labels'] = removeFB(
cortexMesh['vertices'], cortexMesh['faces'], cortexMesh['vertexNormals'], cortexMesh['labels'])
# Now finally start storing things....
# ############
# Define the filenames
filenames = ['weights.txt', 'centres.txt', 'tract.txt', 'orientation.txt', 'area.txt', 'cortical.txt', 'hemisphere.txt']
# 1.) Weights
np.savetxt(subFolder + 'results/' + filenames[0], weights, delimiter=' ', fmt='%1i')
# 2.) Position
# Calc region centers
# centers = np.zeros((weights.shape[0], 3))
with open(subFolder + 'results/' + filenames[1], 'w') as f:
for i in range(weights.shape[0]):
# First get all vertices corresponding to a certain region
regionVertices = cortexMesh['vertices'][cortexMesh['labels'] == i + 1]
# Compute the mean of each region
tmp = np.mean(regionVertices, axis=0)
# Now look for the nearest neighbors
idx = np.sum(np.abs(cortexMesh['vertices'] - tmp), axis=1).argmin()
# Define the nearest vertex as center
# centers[i, :] = cortexMesh['vertices'][idx, :]
center = cortexMesh['vertices'][idx, :]
# Write file
f.write('{0} {1} {2} {3}\n'.format(cortexMesh['labelNames'][i], str(center[0]), str(center[1]), str(center[2])))
f.close()
# 3.) Tract
np.savetxt(subFolder + 'results/' + filenames[2], delay, delimiter=' ', fmt='%1i')
# 4.) Orientation
with open(subFolder + 'results/' + filenames[3], 'w') as f:
for i in range(weights.shape[0]):
# Get all vertex-normals correspodning to the vertices of the current region
regionVertexNormals = cortexMesh['vertexNormals'][cortexMesh['labels'] == i + 1]
# Compute mean vector
orientation = np.mean(regionVertexNormals, axis=0)
# Normalize it
orientation /= np.sqrt(orientation[0]**2 + orientation[1]**2 + orientation[2]**2)
# Write to file
f.write('{0} {1} {2}\n'.format(str(orientation[0]), str(orientation[1]), str(orientation[2])))
f.close()
# 5.) Area
# I'm not quite sure how to get to the exact value for the surface in mm^2
# so for now i just count the surface vertices corresponding to each region
# EDIT: According to the TVB Dokumentation, this attribute is not mandatory
# for the Input!
with open(subFolder + 'results/' + filenames[4], 'w') as f:
for i in range(weights.shape[0]):
area = np.count_nonzero(cortexMesh['labels'] == i)
f.write('{0}\n'.format(str(area)))
f.close()
# 6.) Cortical
# Since in the default atlas all areas are cortical
cortical = np.ones((68, 1))
np.savetxt(subFolder + 'results/' + filenames[5], cortical, delimiter=' ', fmt='%1i')
# 7.) Hemisphere
# Again hard coded for Desikan-Killany Mask!
# TODO: Make this flexible!
hemisphere = np.vstack((np.zeros((34, 1)), np.ones((34, 1))))
np.savetxt(subFolder + 'results/' + filenames[6], hemisphere, delimiter=' ', fmt='%1i')
# Assemble the Zip-File
zf = zipfile.ZipFile(subFolder + 'results/' + subID + '_Connectivity.zip', mode='w')
for fname in filenames:
zf.write(subFolder + 'results/' + fname)
os.remove(subFolder + 'results/' + fname)
zf.close()
