folder='example_data')
csv_file = path_to_visbrain_data(file='MIST_ROI.csv', folder='example_data')
# Read the .csv file :
arr = np.genfromtxt(csv_file, delimiter=';', dtype=str)
# Get column names, labels and index :
column_names = arr[0, :]
arr = np.delete(arr, 0, 0)
n_roi = arr.shape[0]
roi_index = arr[:, 0].astype(int)
roi_labels = arr[:, [1, 2]].astype(object)
# Build the struct array :
label = np.zeros(n_roi, dtype=[('label', object), ('name', object)])
label['label'] = roi_labels[:, 0]
label['name'] = roi_labels[:, 1]
# Get the volume and the hdr transformation :
vol, _, hdr = read_nifti(nifti_file, hdr_as_array=True)
# Define the ROI object and save it :
roi_custom = RoiObj('custom_roi', vol=vol, labels=label, index=roi_index,
hdr=hdr)
# Find thalamus entries :
idx_thalamus = roi_custom.where_is('THALAMUS')
colors = {55: 'slateblue', 56: 'olive', 63: 'darkred', 64: '#ab4642'}
roi_custom.select_roi(idx_thalamus, roi_to_color=colors)
sc.add_to_subplot(roi_custom, row=1, col=2, zoom=.5,
title='Plot dorsal and ventral thalamus with fixed colors')
###############################################################################
# .. note::
# Once your RoiObj is defined, you can save it using
# :class:`visbrain.objects.RoiObj.save`. Once the object is saved, you can
# reload it using the name you've used (here we've used the `custom_roi`
nifti_file = path_to_visbrain_data('MIST_ROI.nii.gz')
csv_file = path_to_visbrain_data('MIST_ROI.csv')
# Read the .csv file :
arr = np.genfromtxt(csv_file, delimiter=';', dtype=str)
# Get column names, labels and index :
column_names = arr[0, :]
arr = np.delete(arr, 0, 0)
n_roi = arr.shape[0]
roi_index = arr[:, 0].astype(int)
roi_labels = arr[:, [1, 2]].astype(object)
# Build the struct array :
label = np.zeros(n_roi, dtype=[('label', object), ('name', object)])
label['label'] = roi_labels[:, 0]
label['name'] = roi_labels[:, 1]
# Get the volume and the hdr transformation :
vol, _, hdr = read_nifti(nifti_file, hdr_as_array=True)
class TestRoiObj(_TestVolumeObject):
"""Test BrainObj."""
OBJ = roi_obj
def test_definition(self):
"""Test function definition."""
for k in ['aal', 'talairach', 'brodmann']:
RoiObj(k)
def test_get_labels(self):
"""Test function get_labels."""
import pandas as pd
vb = Brain(s_xyz=s_xyz)
# Print the list of volumes available :
print('Volumes available by default : ', vb.volume_list())
"""
If you don't have access to a Nifti file, download one of the volume available
at https://brainder.org/download/flair/.
"""
path_to_nifti1 = os.getenv("HOME") # Path to the Nifti file
file_nifti1 = 'GG-853-GM-0.7mm.nii.gz' # Name of the Nifti file
path1 = os.path.join(path_to_nifti1, file_nifti1)
"""
Load the Nifti file. The read_nifti function load the data and the
transformation to convert data into the MNI space :
"""
data1, header1, tf1 = read_nifti(path1)
# print(header1)
# Add the volume to the GUI :
vb.add_volume('Volume1', data1, transform=tf1)
print('Add Volume1 to the list of volumes : ', vb.volume_list())
# You can add multiple volumes :
path_to_nifti2 = os.getenv("HOME")
file_nifti2 = 'GG-853-WM-0.7mm.nii.gz'
path2 = os.path.join(path_to_nifti2, file_nifti2)
data2, header2, tf2 = read_nifti(path2)
vb.add_volume('Volume2', data2, transform=tf2)
print('Add Volume2 to the list of volumes : ', vb.volume_list())
# Set the cross-section to be centered on the last source :
def add_activation(self,
data=None,
vertices=None,
smoothing_steps=5,
file=None,
hemisphere=None,
hide_under=None,
n_contours=None,
cmap='viridis',
clim=None,
vmin=None,
vmax=None,
under='gray',
over='red'):
"""Add activation to the brain template.
This method can be used for :
* Add activations to specific vertices (`data` and `vertices`)
* Add an overlay (`file` input)
Parameters
----------
data : array_like | None
Vector array of data of shape (n_data,).
vertices : array_like | None
Vector array of vertex indices of shape (n_vtx).
Must be an array of integers. If hemisphere is 'left' or 'right'
indexation is done with respect to the specified hemisphere.
smoothing_steps : int | 20
Number of smoothing steps (smoothing is used if n_data < n_vtx).
If None or 0, no smoothing is performed.
file : string | None
Full path to the overlay file. Can either be a nii.gz or gii file.
hemisphrere : {None, 'both', 'left', 'right'}
The hemisphere to use to add the overlay. If None, the method tries
to infer the hemisphere from the file name.
hide_under : float | None
Hide activations under a certain threshold.
n_contours : int | None
Display activations as contour.
cmap : string | 'viridis'
The colormap to use.
clim : tuple | None
The colorbar limits. If None, (data.min(), data.max()) will be used
instead.
vmin : float | None
Minimum threshold.
vmax : float | None
Maximum threshold.
under : string/tuple/array_like | 'gray'
The color to use for values under vmin.
over : string/tuple/array_like | 'red'
The color to use for values over vmax.
"""
kw = self._update_cbar_args(cmap, clim, vmin, vmax, under, over)
is_under = isinstance(hide_under, (int, float))
mask = np.zeros((len(self.mesh), ), dtype=bool)
data_vec = np.zeros((len(self.mesh), ), dtype=np.float32)
sm_data = np.zeros((len(self.mesh), ), dtype=np.float32)
self._default_cblabel = "Activation"
# ============================= METHOD =============================
if isinstance(data, np.ndarray):
# Hemisphere :
hemisphere, activ_vert = self._hemisphere_from_file(
hemisphere, file)
activ_vert_idx = np.where(activ_vert)[0]
is_do_smoothing = True
if vertices is None:
# Data are defined on a dense grid
assert len(activ_vert_idx) == len(data)
vertices = np.arange(len(activ_vert_idx))
is_do_smoothing = False
if smoothing_steps:
logger.warning(
'Data defined on a dense grid; ignore smoothing.')
else:
assert len(vertices) == len(data)
logger.info(" Add data to specific vertices.")
assert (data.ndim == 1) and (np.asarray(vertices).ndim == 1)
assert smoothing_steps is None or isinstance(smoothing_steps, int)
# Get smoothed vertices // data :
if hemisphere != 'both':
# Transform to indexing with respect to the whole brain
vert_whole = activ_vert_idx[vertices]
else:
vert_whole = vertices
if smoothing_steps and is_do_smoothing:
edges = mesh_edges(self.mesh._faces)
sm_mat = smoothing_matrix(vert_whole, edges, smoothing_steps)
sc = sm_mat * data # actual data smoothing
if hemisphere != 'both':
sc = sc[activ_vert]
else:
sc = np.zeros_like(sm_data[activ_vert])
sc[vertices] = data
elif isinstance(file, str):
assert os.path.isfile(file)
if '.nii' in file:
logger.info(" Add overlay from a NIFTI file")
from visbrain.io import read_nifti
# Load data using Nibabel :
sc, _, _ = read_nifti(file)
sc = sc.ravel(order="F")
elif '.gii' in file:
logger.info(" Add overlay from a GIFTI file")
is_nibabel_installed(raise_error=True)
import nibabel
nib = nibabel.load(file)
sc = nib.darrays[0].data.squeeze()
hemisphere = 'both' if len(sc) == len(self.mesh) else hemisphere
# Hemisphere :
_, activ_vert = self._hemisphere_from_file(hemisphere, file)
else:
raise ValueError("Unknown activation type.")
# Define the data to send to the vertices :
logger.info(" Data scaled between (%.3f, "
"%.3f)" % (sc.min(), sc.max()))
sm_data[activ_vert] = sc
data_vec[activ_vert] = self._data_to_contour(sc, clim, n_contours)
mask[activ_vert] = True
# Hide under :
if is_under:
mask[sm_data < hide_under] = False
# Clim :
clim = (sc.min(), sc.max()) if clim is None else clim
assert len(clim) == 2
kw['clim'] = clim
# Add overlay :
self.mesh.add_overlay(data_vec[mask], vertices=np.where(mask)[0], **kw)
def add_activation(self, data=None, vertices=None, smoothing_steps=5,
file=None, hemisphere=None, hide_under=None,
n_contours=None, cmap='viridis', clim=None, vmin=None,
vmax=None, under='gray', over='red'):
"""Add activation to the brain template.
This method can be used for :
* Add activations to specific vertices (`data` and `vertices`)
* Add an overlay (`file` input)
Parameters
----------
data : array_like | None
Vector array of data of shape (n_data,).
vertices : array_like | None
Vector array of vertex indices of shape (n_vtx).
Must be an array of integers. If hemisphere is 'left' or 'right'
indexation is done with respect to the specified hemisphere.
smoothing_steps : int | 20
Number of smoothing steps (smoothing is used if n_data < n_vtx).
If None or 0, no smoothing is performed.
file : string | None
Full path to the overlay file. Can either be a nii.gz or gii file.
hemisphrere : {None, 'both', 'left', 'right'}
The hemisphere to use to add the overlay. If None, the method tries
to infer the hemisphere from the file name.
hide_under : float | None
Hide activations under a certain threshold.
n_contours : int | None
Display activations as contour.
cmap : string | 'viridis'
The colormap to use.
clim : tuple | None
The colorbar limits. If None, (data.min(), data.max()) will be used
instead.
vmin : float | None
Minimum threshold.
vmax : float | None
Maximum threshold.
under : string/tuple/array_like | 'gray'
The color to use for values under vmin.
over : string/tuple/array_like | 'red'
The color to use for values over vmax.
"""
kw = self._update_cbar_args(cmap, clim, vmin, vmax, under, over)
is_under = isinstance(hide_under, (int, float))
mask = np.zeros((len(self.mesh),), dtype=bool)
data_vec = np.zeros((len(self.mesh),), dtype=np.float32)
sm_data = np.zeros((len(self.mesh),), dtype=np.float32)
self._default_cblabel = "Activation"
# ============================= METHOD =============================
if isinstance(data, np.ndarray):
# Hemisphere :
hemisphere, activ_vert = self._hemisphere_from_file(hemisphere,
file)
activ_vert_idx = np.where(activ_vert)[0]
is_do_smoothing = True
if vertices is None:
# Data are defined on a dense grid
assert len(activ_vert_idx) == len(data)
vertices = np.arange(len(activ_vert_idx))
is_do_smoothing = False
if smoothing_steps:
logger.warning(
'Data defined on a dense grid; ignore smoothing.')
else:
assert len(vertices) == len(data)
logger.info(" Add data to specific vertices.")
assert (data.ndim == 1) and (np.asarray(vertices).ndim == 1)
assert smoothing_steps is None or isinstance(smoothing_steps, int)
# Get smoothed vertices // data :
if hemisphere != 'both':
# Transform to indexing with respect to the whole brain
vert_whole = activ_vert_idx[vertices]
else:
vert_whole = vertices
if smoothing_steps and is_do_smoothing:
edges = mesh_edges(self.mesh._faces)
sm_mat = smoothing_matrix(vert_whole, edges, smoothing_steps)
sc = sm_mat * data # actual data smoothing
if hemisphere != 'both':
sc = sc[activ_vert]
else:
sc = np.zeros_like(sm_data[activ_vert])
sc[vertices] = data
elif isinstance(file, str):
assert os.path.isfile(file)
if '.nii' in file:
logger.info(" Add overlay from a NIFTI file")
from visbrain.io import read_nifti
# Load data using Nibabel :
sc, _, _ = read_nifti(file)
sc = sc.ravel(order="F")
elif '.gii' in file:
logger.info(" Add overlay from a GIFTI file")
is_nibabel_installed(raise_error=True)
import nibabel
nib = nibabel.load(file)
sc = nib.darrays[0].data.squeeze()
hemisphere = 'both' if len(sc) == len(self.mesh) else hemisphere
# Hemisphere :
_, activ_vert = self._hemisphere_from_file(hemisphere, file)
else:
raise ValueError("Unknown activation type.")
# Define the data to send to the vertices :
logger.info(" Data scaled between (%.3f, "
"%.3f)" % (sc.min(), sc.max()))
sm_data[activ_vert] = sc
data_vec[activ_vert] = self._data_to_contour(sc, clim, n_contours)
mask[activ_vert] = True
# Hide under :
if is_under:
mask[sm_data < hide_under] = False
# Clim :
clim = (sc.min(), sc.max()) if clim is None else clim
assert len(clim) == 2
kw['clim'] = clim
# Add overlay :
self.mesh.add_overlay(data_vec[mask], vertices=np.where(mask)[0], **kw)
Control the cross-section panel and the volume. This script use a custom
nifti volume downloadable at : https://brainder.org/download/flair/
The nibabel package should also be installed.
.. image:: ../../picture/picbrain/ex_crossec_and_volume.png
"""
import os
from visbrain import Brain
from visbrain.io import read_nifti
"""Import the volume and the associated affine transformation
"""
volume_name = 'GG-853-WM-0.7mm'
path_to_nifti = os.getenv("HOME")
file_nifti = 'GG-853-WM-0.7mm.nii.gz'
path = os.path.join(path_to_nifti, file_nifti)
data, header, tf = read_nifti(path)
vb = Brain()
# Add the volume to the GUI
vb.add_volume(volume_name, data, transform=tf)
"""Set the cross-section to be centered on the slice (70, 171, 80) with the
gist_stern colormap.
"""
vb.cross_sections_control(center=(70, 171, 80),
volume=volume_name,
cmap='gist_stern')
"""Display the 3D volume using the translucent rendering method.
"""
vb.volume_control(volume_name, cmap='TransFire', rendering='translucent')
"""Finally, rotate the scene and hide the main brain.
