Skip to content

Add spacing to surface distances calculations #6137

Closed

Description

@gasperpodobnik

Is your feature request related to a problem? Please describe.
The current implementation of Hausdorff distance and other surface distance metrics only works for images with isotropic spacing.

Describe the solution you'd like
Consider adding a new parameter spacing to the metric call in order to pass it as sampling=spacing to both distance_transform_edt and distance_transform_cdt functions inside the get_surface_distance function (

def get_surface_distance(seg_pred: np.ndarray, seg_gt: np.ndarray, distance_metric: str = "euclidean") -> np.ndarray:
). This will allow users to specify the voxel spacing during distance transform computation, which can be useful for handling different image resolutions.
The same is done in DeepMind's implementation (https://github.com/deepmind/surface-distance/blob/master/surface_distance/metrics.py#L254).

Activity

  1. added a commit that references this issue on Apr 12, 2023
    ef1285a
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions