5422 update attentionunet parameters (#5423)

Signed-off-by: Wenqi Li <wenqil@nvidia.com>

Fixes #5422

### Description
the kernel size and strides are hard-coded:

https://github.com/Project-MONAI/MONAI/blob/a209b06438343830e561a0afd41b1025516a8977/monai/networks/nets/attentionunet.py#L151

this PR makes the values tunable.

`level` parameter is not used and removed in this PR.

### Types of changes
<!--- Put an `x` in all the boxes that apply, and remove the not
applicable items -->
- [x] Non-breaking change (fix or new feature that would not break
existing functionality).
- [ ] Breaking change (fix or new feature that would cause existing
functionality to change).
- [x] New tests added to cover the changes.
- [ ] Integration tests passed locally by running `./runtests.sh -f -u
--net --coverage`.
- [ ] Quick tests passed locally by running `./runtests.sh --quick
--unittests --disttests`.
- [ ] In-line docstrings updated.
- [ ] Documentation updated, tested `make html` command in the `docs/`
folder.

Signed-off-by: Wenqi Li <wenqil@nvidia.com>

Author: wyli

monai/networks/nets/attentionunet.py

@@ -143,12 +143,27 @@ def forward(self, g: torch.Tensor, x: torch.Tensor) -> torch.Tensor:
 
 
 class AttentionLayer(nn.Module):
-    def __init__(self, spatial_dims: int, in_channels: int, out_channels: int, submodule: nn.Module, dropout=0.0):
+    def __init__(
+        self,
+        spatial_dims: int,
+        in_channels: int,
+        out_channels: int,
+        submodule: nn.Module,
+        up_kernel_size=3,
+        strides=2,
+        dropout=0.0,
+    ):
         super().__init__()
         self.attention = AttentionBlock(
             spatial_dims=spatial_dims, f_g=in_channels, f_l=in_channels, f_int=in_channels // 2
         )
-        self.upconv = UpConv(spatial_dims=spatial_dims, in_channels=out_channels, out_channels=in_channels, strides=2)
+        self.upconv = UpConv(
+            spatial_dims=spatial_dims,
+            in_channels=out_channels,
+            out_channels=in_channels,
+            strides=strides,
+            kernel_size=up_kernel_size,
+        )
         self.merge = Convolution(
             spatial_dims=spatial_dims, in_channels=2 * in_channels, out_channels=in_channels, dropout=dropout
         )
@@ -174,7 +189,7 @@ class AttentionUnet(nn.Module):
         channels (Sequence[int]): sequence of channels. Top block first. The length of `channels` should be no less than 2.
         strides (Sequence[int]): stride to use for convolutions.
         kernel_size: convolution kernel size.
-        upsample_kernel_size: convolution kernel size for transposed convolution layers.
+        up_kernel_size: convolution kernel size for transposed convolution layers.
         dropout: dropout ratio. Defaults to no dropout.
     """
 
@@ -210,9 +225,9 @@ def __init__(
         )
         self.up_kernel_size = up_kernel_size
 
-        def _create_block(channels: Sequence[int], strides: Sequence[int], level: int = 0) -> nn.Module:
+        def _create_block(channels: Sequence[int], strides: Sequence[int]) -> nn.Module:
             if len(channels) > 2:
-                subblock = _create_block(channels[1:], strides[1:], level=level + 1)
+                subblock = _create_block(channels[1:], strides[1:])
                 return AttentionLayer(
                     spatial_dims=spatial_dims,
                     in_channels=channels[0],
@@ -227,17 +242,19 @@ def _create_block(channels: Sequence[int], strides: Sequence[int], level: int =
                         ),
                         subblock,
                     ),
+                    up_kernel_size=self.up_kernel_size,
+                    strides=strides[0],
                     dropout=dropout,
                 )
             else:
                 # the next layer is the bottom so stop recursion,
-                # create the bottom layer as the sublock for this layer
-                return self._get_bottom_layer(channels[0], channels[1], strides[0], level=level + 1)
+                # create the bottom layer as the subblock for this layer
+                return self._get_bottom_layer(channels[0], channels[1], strides[0])
 
         encdec = _create_block(self.channels, self.strides)
         self.model = nn.Sequential(head, encdec, reduce_channels)
 
-    def _get_bottom_layer(self, in_channels: int, out_channels: int, strides: int, level: int) -> nn.Module:
+    def _get_bottom_layer(self, in_channels: int, out_channels: int, strides: int) -> nn.Module:
         return AttentionLayer(
             spatial_dims=self.dimensions,
             in_channels=in_channels,
@@ -249,6 +266,8 @@ def _get_bottom_layer(self, in_channels: int, out_channels: int, strides: int, l
                 strides=strides,
                 dropout=self.dropout,
             ),
+            up_kernel_size=self.up_kernel_size,
+            strides=strides,
             dropout=self.dropout,
         )
 

tests/test_attentionunet.py

@@ -39,7 +39,7 @@ def test_attentionunet(self):
             shape = (3, 1) + (92,) * dims
             input = torch.rand(*shape)
             model = att.AttentionUnet(
-                spatial_dims=dims, in_channels=1, out_channels=2, channels=(3, 4, 5), strides=(2, 2)
+                spatial_dims=dims, in_channels=1, out_channels=2, channels=(3, 4, 5), up_kernel_size=5, strides=(1, 2)
             )
             output = model(input)
             self.assertEqual(output.shape[2:], input.shape[2:])