Per-neuron streamline workflow

When your streamline is a StreamlineField, the recommended workflow for a single neuron is to integrate the field once at the cell body, then apply that one fixed curve across the entire arbor.

Why not use the field directly per vertex?

The field describes the typical pia → white matter direction at each location, estimated from many neurons. An individual neuron's dendrites do not necessarily follow it — inverted layer-6 apical dendrites are the classic counterexample. If you measured every vertex against the local field direction, you would be assuming the arbor tracks the field, and you would lose the ability to see where it departs from the local axis.

Instead, you want to measure the arbor against a single reference axis: the cortical axis at its soma. That is what streamline_at_point gives you.

The workflow

from standard_transform import v1dd_ds

field = v1dd_ds.streamline()                    # a StreamlineField (default)
sl    = field.streamline_at_point(soma_xyz)     # a plain Streamline through the soma

# Now use `sl` like any fixed streamline across the whole neuron's coordinates:
verts_straight = sl.radial_points(soma_xyz, skel.vertices)   # straighten the arbor
depths         = sl.depth_along(skel.vertices)               # depth of each vertex

streamline_at_point also remembers the soma as the streamline's anchor, so you can hand the straightening straight to a container that expects a single-argument transform (e.g. Ossify's Cell.transform) with no wrapper:

cell.transform(sl.transformer(), inplace=True)   # anchor = the stored soma

streamline_at_point integrates the field at the soma and returns an ordinary Streamline (fixed shape), carrying the field's transform. From there everything in Depth & radial distance applies unchanged.

Transforming morphology objects

The examples above transform coordinate arrays. If you have a MeshParty object, note that the object-transform helpers are deprecated in favor of Ossify — see Morphology objects.

Checking confidence

The field is better determined in some regions than others (it depends on how much data informed each grid cell). Query the per-node confidence near a location with:

c = field.coverage_at(soma_xyz)

Low coverage is a flag that the streamline through that point rests on little data — useful for filtering cells in poorly-sampled regions. See Streamline field: method for what "confidence" means and how it is computed.