Spatial properties of brain regions

When a reference space is specified, brain regions can expose a range of spatial properties.

We start by selecting an atlas and a space

import siibra
atlas = siibra.atlases.get("human")
# print the spaces defined in the atlas
print(atlas.spaces)

InstanceTable:
 - BIGBRAIN_MICROSCOPIC_TEMPLATE_HISTOLOGY
 - FREESURFER_FSAVERAGE
 - MNI_152_ICBM_2009C_NONLINEAR_ASYMMETRIC
 - MNI_COLIN_27

Choose a space

space = atlas.spaces.get("icbm 2009c asym")
print(space.name)
# and a region
v1_left = atlas.get_region("v1 left", parcellation='julich 2.9')
print(v1_left)

MNI 152 ICBM 2009c Nonlinear Asymmetric
Area hOc1 (V1, 17, CalcS) left

The Region.spatial_props() method computes the centroid and volume of a brain region. Note that a brain region might in general consist of multiple separated components in the space. Also note that in siibra, spatial properties are always represented in millimeter units of the physical coordinate system of the reference space, not in voxel units.

from pprint import pprint
props = v1_left.spatial_props(space=space)
pprint(props)

SpatialProp(cog=None, components=[SpatialPropCmpt(centroid=<Point((-8.515510862347028, -82.38838819846315, 2.00540745659805), space='minds/core/referencespace/v1.0.0/dafcffc5-4826-4bf1-8ff6-46b8a31ff8e2', sigma_mm=0.0)>, volume=10541.0)], space=<Space(identifier='minds/core/referencespace/v1.0.0/dafcffc5-4826-4bf1-8ff6-46b8a31ff8e2', name='MNI 152 ICBM 2009c Nonlinear Asymmetric', species='Homo sapiens')>)

The returned centroid is siibra.Point object. Such spatial primitives are covered in more detail in Locations in reference spaces. For now, we just acknowledge that this minimizes misinterpretation of the coordinates, since a siibra Point is explicitly linked to its space.

centroid = props.components[0].centroid
print(centroid)
centroid.space.name

Point in MNI 152 ICBM 2009c Nonlinear Asymmetric [-8.515510862347028,-82.38838819846315,2.00540745659805]

'MNI 152 ICBM 2009c Nonlinear Asymmetric'

We can also generate a binary mask of the region in a given space and maptype, which gives us a Nifti1Image object as provided by nibabel, and which we can directly visualize using plotting functions like the ones in nilearn:

mask = v1_left.get_regional_map(space, maptype="labelled")
from nilearn import plotting
plotting.plot_roi(mask.fetch(), title=f"Mask of {v1_left.name} in {space.name}")

<nilearn.plotting.displays._slicers.OrthoSlicer object at 0x7f72a9184250>