Add anisotropic edge-case tests for SegResNetDS shape validation

tests/test_segresnet_ds.py

@@ -0,0 +1,136 @@
+# 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 torch
+from parameterized import parameterized
+
+from monai.networks.nets import SegResNetDS
+
+
+class TestSegResNetDSShapeLogic(unittest.TestCase):
+    """Tests for shape_factor() and is_valid_shape() in SegResNetDS."""
+
+    # ---- shape_factor, isotropic (resolution=None) ----
+    @parameterized.expand(
+        [
+            # (spatial_dims, blocks_down, expected_factor)
+            (2, [1, 2, 2, 4], [8, 8]),
+            (3, [1, 2, 2, 4], [8, 8, 8]),
+            (3, [1, 2, 4], [4, 4, 4]),
+        ]
+    )
+    def test_shape_factor_isotropic(self, spatial_dims, blocks_down, expected):
+        """
+        Test shape_factor() calculation for isotropic (resolution=None) configurations.
+
+        Args:
+            spatial_dims: Number of spatial dimensions (2 or 3).
+            blocks_down: List of integers defining the downsampling blocks.
+            expected: Expected divisor factors per spatial dimension.
+        """
+        model = SegResNetDS(
+            spatial_dims=spatial_dims, in_channels=1, out_channels=1, blocks_down=blocks_down, resolution=None
+        )
+        actual = [int(x) for x in model.shape_factor()]
+        self.assertEqual(actual, expected)
+
+    # ---- shape_factor, anisotropic (resolution set) ----
+    @parameterized.expand(
+        [
+            # (spatial_dims, blocks_down, resolution, expected_factor)
+            (3, [1, 2, 2, 4], [1, 1, 5], [8, 8, 2]),
+            (3, [1, 2, 2, 4], [1, 2, 3], [8, 4, 4]),
+        ]
+    )
+    def test_shape_factor_anisotropic(self, spatial_dims, blocks_down, resolution, expected):
+        """
+        Test shape_factor() calculation for anisotropic (resolution set) configurations.
+
+        Args:
+            spatial_dims: Number of spatial dimensions.
+            blocks_down: List of integers defining the downsampling blocks.
+            resolution: List of resolutions for anisotropic scaling.
+            expected: Expected divisor factors per spatial dimension.
+        """
+        model = SegResNetDS(
+            spatial_dims=spatial_dims, in_channels=1, out_channels=1, blocks_down=blocks_down, resolution=resolution
+        )
+        actual = [int(x) for x in model.shape_factor()]
+        self.assertEqual(actual, expected)
+
+    # ---- is_valid_shape, valid inputs ----
+    @parameterized.expand(
+        [
+            # (spatial_dims, blocks_down, resolution, input_shape)
+            (2, [1, 2, 2, 4], None, (1, 1, 16, 16)),
+            (3, [1, 2, 2, 4], None, (1, 1, 16, 16, 16)),
+            (3, [1, 2, 2, 4], [1, 1, 5], (1, 1, 16, 16, 16)),
+            (3, [1, 2, 2, 4], [1, 2, 3], (1, 1, 16, 16, 16)),
+        ]
+    )
+    def test_is_valid_shape_true(self, spatial_dims, blocks_down, resolution, shape):
+        """
+        Test is_valid_shape() returns True for inputs with valid shapes.
+
+        Args:
+            spatial_dims: Number of spatial dimensions.
+            blocks_down: List of integers defining the downsampling blocks.
+            resolution: List of resolutions for anisotropic scaling.
+            shape: Input tensor shape to validate.
+        """
+        model = SegResNetDS(
+            spatial_dims=spatial_dims, in_channels=1, out_channels=1, blocks_down=blocks_down, resolution=resolution
+        )
+        x = torch.zeros(shape)
+        self.assertTrue(model.is_valid_shape(x))
+
+    # ---- is_valid_shape, invalid inputs ----
+    @parameterized.expand(
+        [
+            (3, [1, 2, 2, 4], None, (1, 1, 15, 16, 16)),  # 15 not divisible by 8
+            (3, [1, 2, 2, 4], None, (1, 1, 7, 7, 7)),  # 7 not divisible by 8
+            (3, [1, 2, 2, 4], [1, 1, 5], (1, 1, 16, 15, 16)),  # 15 not divisible by 8
+            (3, [1, 2, 2, 4], [1, 2, 3], (1, 1, 16, 16, 15)),  # 15 not divisible by 4
+        ]
+    )
+    def test_is_valid_shape_false(self, spatial_dims, blocks_down, resolution, shape):
+        """
+        Test is_valid_shape() returns False for inputs with invalid shapes.
+
+        Args:
+            spatial_dims: Number of spatial dimensions.
+            blocks_down: List of integers defining the downsampling blocks.
+            resolution: List of resolutions for anisotropic scaling.
+            shape: Input tensor shape to validate.
+        """
+        model = SegResNetDS(
+            spatial_dims=spatial_dims, in_channels=1, out_channels=1, blocks_down=blocks_down, resolution=resolution
+        )
+        x = torch.zeros(shape)
+        self.assertFalse(model.is_valid_shape(x))
+
+    # ---- integration: forward pass raises on invalid shape ----
+    def test_forward_raises_on_invalid_shape(self):
+        """
+        Test that the forward pass raises ValueError when the input shape is invalid.
+        """
+        model = SegResNetDS(spatial_dims=3, in_channels=1, out_channels=1, blocks_down=[1, 2, 2, 4], resolution=None)
+        x = torch.zeros(1, 1, 15, 16, 16)  # 15 not divisible by 8
+        with self.assertRaises(ValueError):
+            model(x)
+
+
+if __name__ == "__main__":
+    unittest.main()