brain_deer.warp.io¶

The LPS <-> RAS boundary, and all warp/correction disk I/O.

This is the only module in brain_deer.warp that touches LPS. Everything it returns is RAS millimetres; everything it writes is in the on-disk convention of the format (LPS for ANTs fields and Slicer markups).

Two independent flips, both involutions (F @ F == I):

points / affines – diag(-1, -1, 1) on the xy world coordinates.
displacement vectors – the same sign flip on the three components.

nibabel normalises img.affine to RAS+ on read but leaves the displacement components untouched, so a naively-loaded ANTs field is RAS-grid / LPS-vectors with nothing to warn you. Loading here fixes the vectors; the affine is already RAS from nibabel.

Functions¶

`lps_to_ras_points`(→ FloatArray)	Flip `(N,3)` points between LPS and RAS (self-inverse).
`read_displacement_field`(...)	Load an ANTs-style displacement field as an RAS-mm `DeformationField`.
`write_displacement_field`(→ None)	Write an RAS-mm field to an ANTs-style LPS displacement NIfTI.
`read_correction_set`(...)	Read a WarpDrive session's `source`/`target` markups into RAS mm.
`write_correction_set`(→ None)	Write a correction set back out as paired Slicer markups (LPS on disk).

Module Contents¶

brain_deer.warp.io.lps_to_ras_points(points_lps: FloatArray) → FloatArray¶: Flip (N,3) points between LPS and RAS (self-inverse).

brain_deer.warp.io.read_displacement_field(path: str | pathlib.Path) → brain_deer.warp.field.DeformationField¶

Load an ANTs-style displacement field as an RAS-mm DeformationField.

Handles both (I,J,K,3) and ANTs’ 5-D (I,J,K,1,3) layout, and flips the LPS vector components to RAS. The grid affine is taken from nibabel, which has already normalised it to RAS+.

brain_deer.warp.io.write_displacement_field(field: brain_deer.warp.field.DeformationField, path: str | pathlib.Path, *, ants_5d: bool = True) → None¶

Write an RAS-mm field to an ANTs-style LPS displacement NIfTI.

ants_5d writes the (I,J,K,1,3) layout ANTs and Lead-DBS expect.

brain_deer.warp.io.read_correction_set(source_json: str | pathlib.Path, target_json: str | pathlib.Path) → brain_deer.warp.corrections.CorrectionSet¶

Read a WarpDrive session’s source/target markups into RAS mm.

The two files are paired by position: control point i of source is the origin, and control point i of target is where it should go.

brain_deer.warp.io.write_correction_set(cs: brain_deer.warp.corrections.CorrectionSet, source_json: str | pathlib.Path, target_json: str | pathlib.Path) → None¶: Write a correction set back out as paired Slicer markups (LPS on disk).