ericspod · Oct 3, 2026 · Aug 29, 2026 · Aug 29, 2026 · Aug 29, 2026 · Sep 13, 2026
@@ -266,12 +266,13 @@ def resample_if_needed(
                 The output data type of this method is always ``np.float32``.
         """
         orig_type = type(data_array)
-        data_array = convert_to_tensor(data_array, track_meta=True)
+        # Add the channel dimension before setting the affine so spatial_ndim is inferred correctly.
+        data_array = convert_to_tensor(data_array, track_meta=True)[None]
         if affine is not None:
             data_array.affine = convert_to_tensor(affine, track_meta=False)  # type: ignore
         resampler = SpatialResample(mode=mode, padding_mode=padding_mode, align_corners=align_corners, dtype=dtype)
         output_array = resampler(
-            data_array[None], dst_affine=target_affine, spatial_size=output_spatial_shape  # type: ignore
+            data_array, dst_affine=target_affine, spatial_size=output_spatial_shape  # type: ignore
         )
         # convert back at the end
         if isinstance(output_array, MetaTensor):

@@ -117,7 +117,9 @@ def spatial_resample(
     Args:
         img: data to be resampled, assuming `img` is channel-first.
         dst_affine: target affine matrix, if None, use the input affine matrix, effectively no resampling.
-        spatial_size: output spatial size, if the component is ``-1``, use the corresponding input spatial size.
+        spatial_size: output spatial size. Components set to ``-1`` or ``None`` use the corresponding input
+            spatial dimension. If the entire value is ``None``, the output size is computed automatically when
+            possible, otherwise the input spatial shape is used.
         mode: {``"bilinear"``, ``"nearest"``} or spline interpolation order 0-5 (integers).
             Interpolation mode to calculate output values.
             See also: https://pytorch.org/docs/stable/generated/torch.nn.functional.grid_sample.html
@@ -135,6 +137,13 @@ def spatial_resample(
         dtype_pt: data `dtype` for resampling computation.
         lazy: a flag that indicates whether the operation should be performed lazily or not
         transform_info: a dictionary with the relevant information pertaining to an applied transform.
+
+    Returns:
+        torch.Tensor: The resampled output tensor, with metadata preserved when metadata tracking is enabled.
+
+    Raises:
+        ValueError: If the affine or spatial dimensions are invalid, or if the output spatial size cannot be
+            computed.
     """
     original_spatial_shape = img.peek_pending_shape() if isinstance(img, MetaTensor) else img.shape[1:]
     src_affine: torch.Tensor = img.peek_pending_affine() if isinstance(img, MetaTensor) else torch.eye(4)
@@ -156,7 +165,7 @@ def spatial_resample(
     elif spatial_size is None and spatial_rank > 1:  # auto spatial size
         spatial_size, _ = compute_shape_offset(in_spatial_size, src_affine, dst_affine)  # type: ignore
     spatial_size = torch.tensor(
-        fall_back_tuple(ensure_tuple(spatial_size)[:spatial_rank], in_spatial_size, lambda x: x >= 0)
+        fall_back_tuple(ensure_tuple(spatial_size)[:spatial_rank], in_spatial_size, lambda x: x is not None and x >= 0)
     )
     extra_info = {
         "dtype": str(dtype_pt)[6:],  # remove "torch": torch.float32 -> float32

@@ -23,7 +23,7 @@
 from monai.data.utils import to_affine_nd
 from monai.transforms import SpatialResample
 from monai.utils import optional_import
-from tests.lazy_transforms_utils import test_resampler_lazy
+from tests.lazy_transforms_utils import test_resampler_lazy as check_resampler_lazy
 from tests.test_utils import TEST_DEVICES, TEST_NDARRAYS_ALL, assert_allclose, dict_product
 
 TESTS = []
@@ -148,7 +148,7 @@ def test_flips(self, img, device, data_param, expected_output):
             assert_allclose(out, expected_output, rtol=1e-2, atol=1e-2)
             assert_allclose(to_affine_nd(len(out.shape) - 1, out.affine), call_param["dst_affine"])
 
-            test_resampler_lazy(resampler, out, init_param=None, call_param=call_param)
+            check_resampler_lazy(resampler, out, init_param=None, call_param=call_param)
 
     @parameterized.expand(TEST_4_5_D)
     def test_4d_5d(self, new_shape, tile, device, dtype, expected_data):
@@ -165,7 +165,7 @@ def test_4d_5d(self, new_shape, tile, device, dtype, expected_data):
         assert_allclose(out, expected_data[None], rtol=1e-2, atol=1e-2)
         assert_allclose(out.affine, dst.to(torch.float32), rtol=1e-2, atol=1e-2)
 
-        test_resampler_lazy(resampler, out, init_param, call_param)
+        check_resampler_lazy(resampler, out, init_param, call_param)
 
     @parameterized.expand(TEST_DEVICES)
     def test_ill_affine(self, device):
@@ -199,7 +199,7 @@ def test_input_torch(self, new_shape, tile, device, dtype, expected_data, track_
         out = resampler(**call_param)
         assert_allclose(out, expected_data[None], rtol=1e-2, atol=1e-2)
 
-        test_resampler_lazy(resampler, out, init_param, call_param)
+        check_resampler_lazy(resampler, out, init_param, call_param)
 
         if track_meta:
             self.assertIsInstance(out, MetaTensor)
@@ -230,6 +230,24 @@ def test_unchange(self):
             assert_allclose(result, img, type_test=False)
         set_track_meta(True)
 
+    def test_none_spatial_size_rank_one(self):
+        """Verify that an unspecified rank-one size preserves the input shape and returns finite values."""
+        img = MetaTensor(torch.randn(1, 8))
+        result = SpatialResample()(img, spatial_size=None)
+
+        self.assertEqual(result.shape, img.shape)
+        self.assertIsInstance(result, MetaTensor)
+        self.assertTrue(torch.isfinite(result).all())
+
+    def test_partial_none_spatial_size(self):
+        """Verify that ``None`` dimensions fall back while specified dimensions produce the requested shape."""
+        img = MetaTensor(torch.randn(1, 3, 6, 7))
+        result = SpatialResample()(img, spatial_size=(None, 4, 5))
+
+        self.assertEqual(result.shape, (1, 3, 4, 5))
+        self.assertIsInstance(result, MetaTensor)
+        self.assertTrue(torch.isfinite(result).all())
+
 
 if __name__ == "__main__":
     unittest.main()