warning to truncate affine

monai/transforms/spatial/array.py

@@ -499,6 +499,9 @@ def __call__(
             warnings.warn("`data_array` is not of type MetaTensor, assuming affine to be identity.")
             # default to identity
             input_affine = np.eye(sr + 1, dtype=np.float64)
+        input_affine_shape = input_affine.shape[0]
+        if input_affine_shape != sr +1 :
+            warnings.warn(f"Expected `affine_matrix` of size ({sr+1},{sr+1}) for {sr}D input got ({input_affine_shape},{input_affine_shape}) . The affine matrix will be truncated for {sr}D input.")
         affine_ = to_affine_nd(sr, convert_data_type(input_affine, np.ndarray)[0])
 
         out_d = self.pixdim[:sr].copy()
@@ -3549,3 +3552,12 @@ def __call__(self, img: torch.Tensor, randomize: bool = True) -> torch.Tensor:
 
         else:
             return img
+
+
+img = MetaTensor(torch.rand(10,10,5), affine=torch.eye(4), meta={"original_channel_dim": torch.nan})
+sr = Spacing(pixdim=(2,2,2))
+
+double_spacing = sr(img)
+
+#expected shape (6,6,3) or an exception, but the results is (10,6,3), so first dim is treated as channel
+print(double_spacing.shape)

monai/transforms/spatial/functional.py

@@ -101,6 +101,8 @@ def spatial_resample(
     spatial_rank = min(len(img.shape) - 1, src_affine.shape[0] - 1, 3)
     if (not isinstance(spatial_size, int) or spatial_size != -1) and spatial_size is not None:
         spatial_rank = min(len(ensure_tuple(spatial_size)), 3)  # infer spatial rank based on spatial_size
+    if src_affine.shape[0] != spatial_rank +1 :
+        warnings.warn(f"Expected `affine_matrix` of size ({spatial_rank+1},{spatial_rank+1}) for {spatial_rank}D input got ({src_affine.shape[0]},{src_affine.shape[0]}) . The affine matrix will be truncated for {spatial_rank}D input.")
     src_affine = to_affine_nd(spatial_rank, src_affine).to(torch.float64)
     dst_affine = to_affine_nd(spatial_rank, dst_affine) if dst_affine is not None else src_affine
     dst_affine = convert_to_dst_type(dst_affine, src_affine)[0]