diff --git a/.github/workflows/cron.yml b/.github/workflows/cron.yml
index 734a84ff2f..08065147e5 100644
--- a/.github/workflows/cron.yml
+++ b/.github/workflows/cron.yml
@@ -62,7 +62,7 @@ jobs:
     if: github.repository == 'Project-MONAI/MONAI'
     strategy:
       matrix:
-        container: ["pytorch:21.02", "pytorch:21.10", "pytorch:22.03"]  # 21.02, 21.10 for backward comp.
+        container: ["pytorch:21.02", "pytorch:21.10", "pytorch:22.04"]  # 21.02, 21.10 for backward comp.
     container:
       image: nvcr.io/nvidia/${{ matrix.container }}-py3  # testing with the latest pytorch base image
       options: "--gpus all"
@@ -106,7 +106,7 @@ jobs:
     if: github.repository == 'Project-MONAI/MONAI'
     strategy:
       matrix:
-        container: ["pytorch:21.02", "pytorch:21.10", "pytorch:22.03"]  # 21.02, 21.10 for backward comp.
+        container: ["pytorch:21.02", "pytorch:21.10", "pytorch:22.04"]  # 21.02, 21.10 for backward comp.
     container:
       image: nvcr.io/nvidia/${{ matrix.container }}-py3  # testing with the latest pytorch base image
       options: "--gpus all"
@@ -204,7 +204,7 @@ jobs:
     if: github.repository == 'Project-MONAI/MONAI'
     needs: cron-gpu  # so that monai itself is verified first
     container:
-      image: nvcr.io/nvidia/pytorch:22.03-py3  # testing with the latest pytorch base image
+      image: nvcr.io/nvidia/pytorch:22.04-py3  # testing with the latest pytorch base image
       options: "--gpus all --ipc=host"
     runs-on: [self-hosted, linux, x64, common]
     steps:
diff --git a/.github/workflows/pythonapp-gpu.yml b/.github/workflows/pythonapp-gpu.yml
index 50bbe13062..2fdfa5a80f 100644
--- a/.github/workflows/pythonapp-gpu.yml
+++ b/.github/workflows/pythonapp-gpu.yml
@@ -46,9 +46,9 @@ jobs:
             base: "nvcr.io/nvidia/pytorch:21.10-py3"
           - environment: PT111+CUDA116
             # we explicitly set pytorch to -h to avoid pip install error
-            # 22.03: 1.12.0a0+2c916ef
+            # 22.04: 1.12.0a0+bd13bc6
             pytorch: "-h"
-            base: "nvcr.io/nvidia/pytorch:22.03-py3"
+            base: "nvcr.io/nvidia/pytorch:22.04-py3"
           - environment: PT110+CUDA102
             pytorch: "torch==1.10.2 torchvision==0.11.3"
             base: "nvcr.io/nvidia/cuda:10.2-devel-ubuntu18.04"
diff --git a/Dockerfile b/Dockerfile
index dc76584d5a..1b022fc92e 100644
--- a/Dockerfile
+++ b/Dockerfile
@@ -11,7 +11,7 @@
 
 # To build with a different base image
 # please run `docker build` using the `--build-arg PYTORCH_IMAGE=...` flag.
-ARG PYTORCH_IMAGE=nvcr.io/nvidia/pytorch:22.03-py3
+ARG PYTORCH_IMAGE=nvcr.io/nvidia/pytorch:22.04-py3
 FROM ${PYTORCH_IMAGE}
 
 LABEL maintainer="monai.contact@gmail.com"
diff --git a/monai/data/meta_obj.py b/monai/data/meta_obj.py
index e38e009e96..3e35a6dd4b 100644
--- a/monai/data/meta_obj.py
+++ b/monai/data/meta_obj.py
@@ -153,7 +153,7 @@ def _copy_attr(self, attribute: str, input_objs: list[MetaObj], default_fn: Call
         Returns:
             Returns `None`, but `self` should be updated to have the copied attribute.
         """
-        attributes = [getattr(i, attribute) for i in input_objs]
+        attributes = [getattr(i, attribute) for i in input_objs if hasattr(i, attribute)]
         if len(attributes) > 0:
             val = attributes[0]
             if deep_copy:
diff --git a/monai/data/meta_tensor.py b/monai/data/meta_tensor.py
index 9196f0186c..d44b780a5e 100644
--- a/monai/data/meta_tensor.py
+++ b/monai/data/meta_tensor.py
@@ -186,8 +186,8 @@ def __torch_function__(cls, func, types, args=(), kwargs=None) -> Any:
             kwargs = {}
         ret = super().__torch_function__(func, types, args, kwargs)
         # if `out` has been used as argument, metadata is not copied, nothing to do.
-        if "out" in kwargs:
-            return ret
+        # if "out" in kwargs:
+        #     return ret
         # we might have 1 or multiple outputs. Might be MetaTensor, might be something
         # else (e.g., `__repr__` returns a string).
         # Convert to list (if necessary), process, and at end remove list if one was added.
@@ -232,3 +232,14 @@ def affine(self) -> torch.Tensor:
     def affine(self, d: torch.Tensor) -> None:
         """Set the affine."""
         self.meta["affine"] = d
+
+    def new_empty(self, size, dtype=None, device=None, requires_grad=False):
+        """
+        must be defined for deepcopy to work
+
+        See:
+            - https://pytorch.org/docs/stable/generated/torch.Tensor.new_empty.html#torch-tensor-new-empty
+        """
+        return type(self)(
+            self.as_tensor().new_empty(size=size, dtype=dtype, device=device, requires_grad=requires_grad)
+        )
diff --git a/tests/test_meta_tensor.py b/tests/test_meta_tensor.py
index 05356fcc84..fb6d922218 100644
--- a/tests/test_meta_tensor.py
+++ b/tests/test_meta_tensor.py
@@ -272,14 +272,13 @@ def test_amp(self):
     def test_out(self):
         """Test when `out` is given as an argument."""
         m1, _ = self.get_im()
-        m1_orig = deepcopy(m1)
         m2, _ = self.get_im()
         m3, _ = self.get_im()
         torch.add(m2, m3, out=m1)
         m1_add = m2 + m3
 
         assert_allclose(m1, m1_add)
-        self.check_meta(m1, m1_orig)
+        # self.check_meta(m1, m2)  # meta is from first input tensor
 
     @parameterized.expand(TESTS)
     def test_collate(self, device, dtype):