Surface seeding for tractography

Surface seeding is a way to generate initial position for tractography from cortical surfaces position [1].

from fury.io import load_polydata
from fury.utils import (
    get_actor_from_polydata,
    get_polydata_triangles,
    get_polydata_vertices,
    normals_from_v_f,
)
import numpy as np

from dipy.data import get_fnames
from dipy.tracking.mesh import (
    random_coordinates_from_surface,
    seeds_from_surface_coordinates,
)
from dipy.viz import actor, window

Fetch and load a surface

brain_lh = get_fnames(name="fury_surface")
polydata = load_polydata(brain_lh)

Extract the triangles and vertices

triangles = get_polydata_triangles(polydata)
vts = get_polydata_vertices(polydata)

Display the surface

First, create an actor from the polydata, to display in the scene

scene = window.Scene()
surface_actor = get_actor_from_polydata(polydata)

scene.add(surface_actor)
scene.set_camera(position=(-500, 0, 0), view_up=(0.0, 0.0, 1))

# Uncomment the line below to show to display the window
# window.show(scene, size=(600, 600), reset_camera=False)
window.record(scene=scene, out_path="surface_seed1.png", size=(600, 600))

Initial cortical surface

Generate a list of seeding positions

Choose the number of seed

nb_seeds = 100000
nb_triangles = len(triangles)

Get a list of triangles indices and trilinear coordinates for each seed

tri_idx, trilin_co = random_coordinates_from_surface(nb_triangles, nb_seeds)

Get the 3d cartesian position from triangles indices and trilinear coordinates

seed_pts = seeds_from_surface_coordinates(triangles, vts, tri_idx, trilin_co)

Compute normal and get the normal direction for each seeds

normals = normals_from_v_f(vts, triangles)
seed_n = seeds_from_surface_coordinates(triangles, normals, tri_idx, trilin_co)

Create dot actor for seeds (blue)

seed_actors = actor.dot(seed_pts, colors=(0, 0, 1), dot_size=4.0)

Create line actors for seeds normals (green outside, red inside)

normal_length = 0.5
normal_in = np.tile(seed_pts[:, np.newaxis, :], (1, 2, 1))
normal_out = np.tile(seed_pts[:, np.newaxis, :], (1, 2, 1))
normal_in[:, 0] -= seed_n * normal_length
normal_out[:, 1] += seed_n * normal_length

normal_in_actor = actor.line(normal_in, colors=(1, 0, 0))
normal_out_actor = actor.line(normal_out, colors=(0, 1, 0))

Visualise seeds and normals along the surface

scene = window.Scene()
scene.add(surface_actor)
scene.add(seed_actors)
scene.add(normal_in_actor)
scene.add(normal_out_actor)
scene.set_camera(position=(-500, 0, 0), view_up=(0.0, 0.0, 1))

# Uncomment the line below to show to display the window
# window.show(scene, size=(600, 600), reset_camera=False)
window.record(scene=scene, out_path="surface_seed2.png", size=(600, 600))

Surface seeds with normal orientation

References

[1]

Etienne St-Onge, Alessandro Daducci, Gabriel Girard, and Maxime Descoteaux. Surface-enhanced tractography (SET). NeuroImage, 169:524–539, 2018. URL: https://doi.org/10.1016/j.neuroimage.2017.12.036, doi:10.1016/j.neuroimage.2017.12.036.