Surface timeseries

Display a surface timeseries using data from nilearn

Tags: experimental

from importlib.metadata import version

from nilearn import datasets, surface
from packaging.version import parse

import napari

if parse(version("numpy")) >= parse('1.24') and parse(version("nilearn")) < parse('0.10.1'):
    raise RuntimeError(
        'Incompatible numpy version. '
        'You must have numpy less than 1.24 for nilearn 0.10.1 and below to '
        'work and download the example data'
    )


# Fetch datasets - this will download dataset if datasets are not found
nki_dataset = datasets.fetch_surf_nki_enhanced(n_subjects=1)
fsaverage = datasets.fetch_surf_fsaverage()

# Load surface data and resting state time series from nilearn
brain_vertices, brain_faces = surface.load_surf_data(fsaverage['pial_left'])
brain_vertex_depth = surface.load_surf_data(fsaverage['sulc_left'])
timeseries = surface.load_surf_data(nki_dataset['func_left'][0])
# nilearn provides data as n_vertices x n_timepoints, but napari requires the
# vertices axis to be placed last to match NumPy broadcasting rules
timeseries = timeseries.transpose((1, 0))

# create an empty viewer
viewer = napari.Viewer(ndisplay=3)

# add the mri
viewer.add_surface((brain_vertices, brain_faces, brain_vertex_depth), name='base')
viewer.add_surface((brain_vertices, brain_faces, timeseries),
                    colormap='turbo', opacity=0.9,
                    contrast_limits=[-1.5, 3.5], name='timeseries')

if __name__ == '__main__':
    napari.run()