update modules/inverse_transforms_and_test_time_augmentations.ipynb (#799)

* update based on new API

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

* fixes pep8

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

Author: wyli

modules/inverse_transforms_and_test_time_augmentations.ipynb

@@ -123,7 +123,6 @@
     "    Dataset,\n",
     "    pad_list_data_collate,\n",
     "    TestTimeAugmentation,\n",
-    "    decollate_batch,\n",
     ")\n",
     "from monai.inferers import sliding_window_inference\n",
     "from monai.losses import DiceLoss\n",
@@ -228,8 +227,7 @@
     "    def __call__(self, data):\n",
     "        d = dict(data)\n",
     "        im = d[self.label_key]\n",
-    "        _im = im.detach().cpu().numpy()\n",
-    "        q = np.sum((_im > 0).reshape(-1, _im.shape[-1]), axis=0)\n",
+    "        q = np.sum((im.array > 0).reshape(-1, im.array.shape[-1]), axis=0)\n",
     "        _slice = np.where(q == np.max(q))[0][0]\n",
     "        for key in self.keys:\n",
     "            d[key] = d[key][..., _slice]\n",
@@ -247,7 +245,7 @@
     "            fname = os.path.basename(\n",
     "                data[key + \"_meta_dict\"][\"filename_or_obj\"])\n",
     "            path = os.path.join(self.path, key, fname)\n",
-    "            nib.save(nib.Nifti1Image(data[key].detach().cpu().numpy(), np.eye(4)), path)\n",
+    "            nib.save(nib.Nifti1Image(data[key].array, np.eye(4)), path)\n",
     "            d[key] = path\n",
     "        return d\n",
     "\n",
@@ -443,7 +441,7 @@
     "def infer_seg(images, model, roi_size=(96, 96), sw_batch_size=4):\n",
     "    val_outputs = sliding_window_inference(\n",
     "        images, roi_size, sw_batch_size, model)\n",
-    "    return torch.stack([post_trans(i) for i in decollate_batch(val_outputs)])\n",
+    "    return pad_list_data_collate([post_trans(i) for i in val_outputs])\n",
     "\n",
     "\n",
     "# Create network, loss fn., etc.\n",