def run_from_args(args):
img = nibabel.load(args.coord)
coord = img.get_data()
white = CorticalMesh.read(args.white)
dti = nibabel.load(args.dyad).get_data()
gdti = (coord * dti[..., None]).sum(-2)
if args.mask is None:
surf_mask = np.ones(white.nvertices, dtype='bool')
else:
surf_mask = nibabel.load(args.mask).darrays[0].data != 0
idx_vertex = None if args.idx_vertex is None else nibabel.load(args.idx_vertex).get_data()
if idx_vertex.ndim == 4:
idx_vertex = idx_vertex[..., 0]
distance = None if args.distance is None else nibabel.load(args.distance).get_data()
tofit = model(abs(gdti[..., 0]), white, img.affine, surf_mask, min_dist=args.min_dist, smooth_weight=args.weight,
watson=args.watson, idx_vertex=idx_vertex, dist=distance)
res = pymc3.find_MAP(model=tofit, fmin=optimize.fmin_cg)
if args.radial_index is not None:
model_ri = tofit.fastfn(tofit.model_ri_3d)(res)
obs_ri = tofit.fastfn(tofit.observed_ri_3d)(res)
nibabel.Nifti1Image(np.stack([model_ri, obs_ri], -1), img.affine).to_filename(args.radial_index)
bm = cifti.BrainModel.from_surface(np.where(surf_mask)[0], surf_mask.size, white.anatomy.cifti)
mat = np.stack([res['d0'], res['log_sigma'], np.exp(res['log_sigma'])])
cifti.write(args.output, mat, (cifti.Scalar.from_names(['offset', 'log_width', 'width']), bm))
def savemap(name, data, brain_mask, axes, directory="figures"):
if not os.path.exists(directory):
os.makedirs(directory)
file_path = f"{directory}/{name}.dtseries.nii"
# noinspection PyTypeChecker
if not isinstance(axes, collections.Iterable):
axes = (axes, )
cifti.write(file_path, data, axes + (brain_mask,))
return os.getcwd() + os.sep + file_path
def map_regions_pc_cole():
no_color = (1.0, 1.0, 1.0, 0.0)
tpt = tpt_cole
img, (lbl, brain) = cifti.read(tpt.file_full_path)
regions = lbl.label.item()
for h in [HierarchyName.EXTENDED_PERIPHERY_CORE, HierarchyName.RESTRICTED_PERIPHERY_CORE]:
hierarchy = tpt.net_hierarchy(h)
cp_out = {}
img_out = np.zeros(img.shape)
found_regions = {}
for index, (name, c) in regions.items():
if index == 0 or index > 360:
cp_out[index] = name, no_color
continue
net, lh = name.split("_")
net_parts = net.split("-")
net, rgn = ("".join(net_parts[:2]), net_parts[2]) if len(net_parts) == 3 else net_parts
is_p = net in hierarchy["P"]
is_c = net in (hierarchy["RC"] if "RC" in hierarchy else hierarchy["EC"])
if rgn not in found_regions:
new_index = len(found_regions) + 1
found_regions[rgn] = new_index
if is_p:
color = PC_colors_tuple[0]
elif is_c:
color = PC_colors_tuple[1]
else:
color = no_color
print(f"{net} without color in {lbl}")
cp_out[new_index] = rgn, color
else:
new_index = found_regions[rgn]
img_out[img == index] = new_index
# noinspection PyTypeChecker
cifti.write(f"figures/cole.{lbl}.dlabel.nii", img_out, (cifti.Label([cp_out]), brain))
os.system(f"wb_command -cifti-separate figures/cole.{lbl}.dlabel.nii COLUMN "
f"-label CORTEX_LEFT figures/cole.{lbl}.L.label.gii "
f"-label CORTEX_RIGHT figures/cole.{lbl}.R.label.gii")
os.system(f"wb_command -label-to-border "
f"{get_full_path(AssetCategory.HCP1200, 'anat.midthickness.32k.L.surf.gii')} "
f"figures/cole.{lbl}.L.label.gii figures/cole.{lbl}.L.border")
os.system(f"wb_command -label-to-border "
f"{get_full_path(AssetCategory.HCP1200, 'anat.midthickness.32k.R.surf.gii')} "
f"figures/cole.{lbl}.R.label.gii figures/cole.{lbl}.R.border")
def save_to_dtseries(filename, brain_model, mat):
'''
saves a dtseries.nii file given the data matrix and the brain model.
The series axis is generated so as to fit the size of the matrix (every row is a time point).
:param filename: the file to save
:param brain_model: a Cifti.Axis.BrainModel object
:param mat: the data matrix
:return:
'''
if len(np.shape(mat)) == 2:
assert mat.shape[1] == np.size(brain_model.arr)
elif len(np.shape(mat)) == 1:
mat = np.reshape(mat, [1, np.size(brain_model.arr)])
series = cifti.Series(start=0, step=1, size=mat.shape[0])
if not filename.endswith(".dtseries.nii"):
filename += ".dtseries.nii"
gb.curr_cifti_filenames.append(filename)
cifti.write(filename, mat, (series, brain_model))
return filename
def save_brainimg(imgpath, data, header):
"""
Save brain image identified by its suffix
suffix now support
Nifti: .nii.gz
freesurfer: .mgz, .mgh
cifti: .dscalar.nii, .dlabel.nii, .dtseries.nii
Parameters:
------------
imgpath: brain image path to be saved
data: brain image data matrix
header: brain image header
Returns:
--------
"""
imgname = os.path.basename(imgpath)
imgsuffix = imgname.split('.')[1:]
imgsuffix = '.'.join(imgsuffix)
if imgsuffix == 'nii.gz':
data = np.transpose(data, (1, 2, 3, 0))
outimg = nib.Nifti1Image(data, None, header)
nib.save(outimg, imgpath)
elif imgsuffix == 'mgz' or imgsuffix == 'mgh':
data = np.transpose(data, (1, 2, 3, 0))
outimg = nib.MGHImage(data, None, header)
nib.save(outimg, imgpath)
elif imgsuffix == 'dscalar.nii' or imgsuffix == 'dlabel.nii' or imgsuffix == 'dtseries.nii':
data = data[..., 0, 0]
map_name = [''] * data.shape[0]
bm_full = header[1]
cifti.write(imgpath, data,
(cifti.Scalar.from_names(map_names), bm_full))
else:
raise Exception(
'Not support this format of brain image data, please contact with author to update this function.'
)
def map_nets_sh2007():
tpt = tpt_sh
img, (lbl, brain) = cifti.read(tpt.file_full_path)
regions = lbl.label.item()
net_out = {}
img_out = np.zeros(img.shape)
nets = tpt.net_order
palette = tpt.net_colors
for index, (name, c) in regions.items():
if index == 0:
net_out[index] = name, c
continue
network = name.split("_")[2]
net_index = nets.index(network) + 1
if net_index not in net_out:
net_out[net_index] = network, palette[net_index - 1] + [1, ]
img_out[img == index] = net_index
# noinspection PyTypeChecker
cifti.write(f"figures/sh7nets.dlabel.nii", img_out, (cifti.Label([net_out]), brain))
def map_regions_pc_sh2007():
tpt = tpt_sh
img, (lbl, brain) = cifti.read(tpt.file_full_path)
regions = lbl.label.item()
hierarch = tpt.net_hierarchy(HierarchyName.PERIPHERY_CORE)
cp_out = {}
img_out = np.zeros(img.shape)
found_regions = {}
for index, (name, c) in regions.items():
if index == 0:
cp_out[index] = name, c
continue
parts = name.split("_")
lr, network, rgn, num = parts[1], parts[2], parts[-2], parts[-1]
cp = 0 if network in hierarch["P"] else 1
if rgn not in found_regions:
new_index = len(found_regions) + 1
found_regions[rgn] = new_index
cp_out[new_index] = rgn, PC_colors_tuple[cp]
else:
new_index = found_regions[rgn]
img_out[img == index] = new_index
# noinspection PyTypeChecker
cifti.write(f"figures/sh7cp.dlabel.nii", img_out, (cifti.Label([cp_out]), brain))
os.system(f"wb_command -cifti-separate figures/sh7cp.dlabel.nii COLUMN "
f"-label CORTEX_LEFT figures/sh7cp.L.label.gii "
f"-label CORTEX_RIGHT figures/sh7cp.R.label.gii")
os.system(f"wb_command -label-to-border "
f"{get_full_path(AssetCategory.HCP1200, 'anat.midthickness.32k.L.surf.gii')} "
f"figures/sh7cp.L.label.gii figures/sh7cp.L.border")
os.system(f"wb_command -label-to-border "
f"{get_full_path(AssetCategory.HCP1200, 'anat.midthickness.32k.R.surf.gii')} "
f"figures/sh7cp.R.label.gii figures/sh7cp.R.border")
def map_nets_cole():
tpt = tpt_cole
img, (lbl, brain) = cifti.read(tpt.file_full_path)
regions = lbl.label.item()
net_out = {}
img_out = np.zeros(img.shape)
nets = tpt.net_order
palette = tpt.net_colors
for index, (name, c) in regions.items():
if index == 0:
net_out[index] = name, c
continue
network, lh = name.split("_")
net_parts = network.split("-")
network, rgn = ("".join(net_parts[:2]), net_parts[2]) if len(net_parts) == 3 else net_parts
net_index = nets.index(network) + 1
if net_index not in net_out:
net_out[net_index] = network, palette[net_index - 1] + [1, ]
img_out[img == index] = net_index
# noinspection PyTypeChecker
cifti.write(f"figures/colenets.dlabel.nii", img_out, (cifti.Label([net_out]), brain))
dd = dd - np.mean(dd,axis=0)
next = (dd * normproj).A
hypsource[hemi] = next
zscore(hypsource[hemi],chunks_attr=None)
finalproj[hemi] = normproj
t2 = time.time()
print('Hyperalignment for %s hemisphere took %s minutes' % (hemi, ((t2-t1)/60.)))
# Apply projections to other tasks:
# tasks = ['rfMRI_REST2','tfMRI_WM','tfMRI_GAMBLING','tfMRI_LANGUAGE','tfMRI_MOTOR','tfMRI_RELATIONAL','tfMRI_SOCIAL']
tasks = ['tfMRI_RELATIONAL']
for task in tasks:
for ses in ['LR','RL']:
if task == 'rfMRI_REST2':
cname = '%s/%s/%s_%s/%s_%s_10k_Atlas_MSMAll_clean_24nuisance.reg_lin.trend_filt_sm6' % (tempdir,sub,task,ses,task,ses)
else:
cname = '%s/%s/%s_%s/%s_%s_Atlas_MSMAll_10k_anat_cortex' % (tempdir,sub,task,ses,task,ses)
tmpimg =cifti.read('%s.dtseries.nii' % cname)
tmpwrite = tmpimg[0].copy()
nvols = tmpimg[0].shape[0]
for hemi in ['L','R']:
nverts = len(ind[hemi])
blank = np.zeros([nvols,nverts])
img = tmpimg[0][:,ind[hemi]]
img = img[:,maskimg[hemi]]
himg1 = img.dot(finalproj[hemi])
blank[:,maskimg[hemi]] = himg1
tmpwrite[:,ind[hemi]] = blank
cifti.write('%s/%s/%s_%s/%s_%s_Atlas_MSMAll_10k_R1LRRL_hyp.dtseries.nii' % (tempdir,sub,task,ses,task,ses),tmpwrite,tmpimg[1])
hypsource[hemi] = next
zscore(hypsource[hemi], chunks_attr=None)
finalproj[hemi] = normproj
t2 = time.time()
print('Hyperalignment for %s hemisphere took %s minutes' %
(hemi, ((t2 - t1) / 60.)))
# Apply projections to other tasks:
tasks = [
'rfMRI_REST2', 'tfMRI_WM', 'tfMRI_GAMBLING', 'tfMRI_LANGUAGE',
'tfMRI_MOTOR', 'tfMRI_RELATIONAL', 'tfMRI_SOCIAL'
]
for task in tasks:
for ses in ['LR', 'RL']:
cname = '%s/%s/%s_%s/%s_%s_10k_Atlas_MSMAll_clean_24nuisance.reg_lin.trend_filt_sm6' % (
tempdir, sub, task, ses, task, ses)
tmpimg = cifti.read('%s.dtseries.nii' % cname)
tmpwrite = tmpimg[0].copy()
nvols = tmpimg[0].shape[0]
for hemi in ['L', 'R']:
nverts = len(ind[hemi])
blank = np.zeros([nvols, nverts])
img = tmpimg[0][:, ind[hemi]]
img = img[:, maskimg[hemi]]
himg1 = img.dot(finalproj[hemi])
blank[:, maskimg[hemi]] = himg1
tmpwrite[:, ind[hemi]] = blank
cifti.write(
'%s/%s/%s_%s/%s_%s_10k_Atlas_MSMAll_clean_24nuisance.reg_lin.trend_filt_sm6_R1LRRL_hyp.dtseries.nii'
% (tempdir, sub, task, ses, task, ses), tmpwrite, tmpimg[1])