meddlr.metrics.functional.assd#
- meddlr.metrics.functional.assd(y_pred, y_true, spacing=None, connectivity=1, crop=True)#
Computes average symmetric surface distance.
- Parameters
y_pred (torch.Tensor) – Predicted categorical segmentation. Expected shape:
(B, C, H, ...).y_true (torch.Tensor) – Ground truth categorical segmentation. Expected shape:
(B, C, H, ...).
- Returns
Dice score. Shape:
(B, C).- Return type
torch.Tensor