fix(metrics): stop passing deprecated always_return_as_numpy internally (Fixes #9059)

monai/metrics/utils.py

@@ -417,7 +417,7 @@ def get_edge_surface_distance(
     edges_spacing = None
     if use_subvoxels:
         edges_spacing = spacing if spacing is not None else ([1] * len(y_pred.shape))
-    edge_results = get_mask_edges(y_pred, y, crop=(mask is None), spacing=edges_spacing, always_return_as_numpy=False)
+    edge_results = get_mask_edges(y_pred, y, crop=(mask is None), spacing=edges_spacing)
     edges_pred, edges_gt = edge_results[0], edge_results[1]
     edges_pred_any = bool(edges_pred.any())
     edges_gt_any = bool(edges_gt.any())

tests/metrics/test_metrics_internal_deprecation.py

@@ -0,0 +1,65 @@
+# Copyright (c) MONAI Consortium
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#     http://www.apache.org/licenses/LICENSE-2.0
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+
+from __future__ import annotations
+
+import unittest
+import warnings
+
+import torch
+
+from monai.metrics import HausdorffDistanceMetric, SurfaceDistanceMetric
+from monai.metrics.utils import get_mask_edges
+from monai.utils import optional_import
+
+_, has_scipy = optional_import("scipy.ndimage", name="binary_erosion")
+
+_DEPRECATED_ARG_MESSAGE = ".*always_return_as_numpy.*"
+
+
+@unittest.skipUnless(has_scipy, "Requires scipy.")
+class TestMetricsNoInternalDeprecationWarnings(unittest.TestCase):
+    """`SurfaceDistanceMetric` and `HausdorffDistanceMetric` must not trigger deprecation warnings of their own.
+
+    Both metrics route through `get_edge_surface_distance`, which calls `get_mask_edges`. If that helper passes the
+    deprecated `always_return_as_numpy` argument, every metric computation warns about an argument the caller never
+    passed and cannot suppress. Both metrics compute mask edges via `scipy.ndimage`, so the tests need scipy.
+    """
+
+    def setUp(self):
+        self.pred = torch.zeros(1, 1, 32, 32)
+        self.pred[..., :16, :] = 1
+        self.gt = torch.zeros(1, 1, 32, 32)
+        self.gt[..., :20, :] = 1
+
+    def test_deprecated_arg_warns_when_passed_explicitly(self):
+        """Control: passing the deprecated argument directly does emit a `FutureWarning`.
+
+        Ensures the no-warning checks below are meaningful and not passing vacuously.
+        """
+        with self.assertWarnsRegex(FutureWarning, "always_return_as_numpy"):
+            get_mask_edges(self.pred, self.gt, always_return_as_numpy=False)
+
+    def test_metrics_emit_no_deprecated_arg_warning(self):
+        """Neither metric emits the `always_return_as_numpy` `FutureWarning`.
+
+        The warning is escalated to an error, so any internal use of the deprecated argument fails the test.
+        """
+        for metric in (SurfaceDistanceMetric(), HausdorffDistanceMetric()):
+            with self.subTest(metric=type(metric).__name__):
+                with warnings.catch_warnings():
+                    warnings.filterwarnings("error", message=_DEPRECATED_ARG_MESSAGE, category=FutureWarning)
+                    metric(self.pred, self.gt)
+
+
+if __name__ == "__main__":
+    unittest.main()