Skip to content

Commit bc34263

Browse files
yiheng-wang-nvyiheng-wang-nv
authored
450 update AsDiscrete (#451)
* update asdiscrete Signed-off-by: Yiheng Wang <vennw@nvidia.com> * update postprocessing figures Signed-off-by: Yiheng Wang <vennw@nvidia.com> * fix version error of mutual info Signed-off-by: Yiheng Wang <vennw@nvidia.com> * update to use include in torchin Signed-off-by: Yiheng Wang <vennw@nvidia.com>
1 parent 2c63c42 commit bc34263

File tree

60 files changed

+158
-157
lines changed

Some content is hidden

Large Commits have some content hidden by default. Use the searchbox below for content that may be hidden.

60 files changed

+158
-157
lines changed

‎2d_classification/mednist_tutorial.ipynb

Lines changed: 3 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -375,7 +375,7 @@
375375
" [LoadImage(image_only=True), AddChannel(), ScaleIntensity(), EnsureType()])\n",
376376
"\n",
377377
"y_pred_trans = Compose([EnsureType(), Activations(softmax=True)])\n",
378-
"y_trans = Compose([EnsureType(), AsDiscrete(to_onehot=True, num_classes=num_class)])"
378+
"y_trans = Compose([EnsureType(), AsDiscrete(to_onehot=num_class)])"
379379
]
380380
},
381381
{
@@ -1328,7 +1328,7 @@
13281328
],
13291329
"metadata": {
13301330
"kernelspec": {
1331-
"display_name": "Python 3",
1331+
"display_name": "Python 3 (ipykernel)",
13321332
"language": "python",
13331333
"name": "python3"
13341334
},
@@ -1342,7 +1342,7 @@
13421342
"name": "python",
13431343
"nbconvert_exporter": "python",
13441344
"pygments_lexer": "ipython3",
1345-
"version": "3.8.10"
1345+
"version": "3.8.12"
13461346
}
13471347
},
13481348
"nbformat": 4,

‎2d_segmentation/torch/unet_evaluation_array.py

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -47,7 +47,7 @@ def main(tempdir):
4747
# sliding window inference for one image at every iteration
4848
val_loader = DataLoader(val_ds, batch_size=1, num_workers=1, pin_memory=torch.cuda.is_available())
4949
dice_metric = DiceMetric(include_background=True, reduction="mean", get_not_nans=False)
50-
post_trans = Compose([EnsureType(), Activations(sigmoid=True), AsDiscrete(threshold_values=True)])
50+
post_trans = Compose([EnsureType(), Activations(sigmoid=True), AsDiscrete(threshold=0.5)])
5151
saver = SaveImage(output_dir="./output", output_ext=".png", output_postfix="seg")
5252
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
5353
model = UNet(

‎2d_segmentation/torch/unet_evaluation_dict.py

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -54,7 +54,7 @@ def main(tempdir):
5454
# sliding window inference need to input 1 image in every iteration
5555
val_loader = DataLoader(val_ds, batch_size=1, num_workers=4, collate_fn=list_data_collate)
5656
dice_metric = DiceMetric(include_background=True, reduction="mean", get_not_nans=False)
57-
post_trans = Compose([EnsureType(), Activations(sigmoid=True), AsDiscrete(threshold_values=True)])
57+
post_trans = Compose([EnsureType(), Activations(sigmoid=True), AsDiscrete(threshold=0.5)])
5858
saver = SaveImage(output_dir="./output", output_ext=".png", output_postfix="seg")
5959
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
6060
model = UNet(

‎2d_segmentation/torch/unet_training_array.py

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -89,7 +89,7 @@ def main(tempdir):
8989
val_ds = ArrayDataset(images[-20:], val_imtrans, segs[-20:], val_segtrans)
9090
val_loader = DataLoader(val_ds, batch_size=1, num_workers=4, pin_memory=torch.cuda.is_available())
9191
dice_metric = DiceMetric(include_background=True, reduction="mean", get_not_nans=False)
92-
post_trans = Compose([EnsureType(), Activations(sigmoid=True), AsDiscrete(threshold_values=True)])
92+
post_trans = Compose([EnsureType(), Activations(sigmoid=True), AsDiscrete(threshold=0.5)])
9393
# create UNet, DiceLoss and Adam optimizer
9494
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
9595
model = monai.networks.nets.UNet(

‎2d_segmentation/torch/unet_training_dict.py

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -99,7 +99,7 @@ def main(tempdir):
9999
val_ds = monai.data.Dataset(data=val_files, transform=val_transforms)
100100
val_loader = DataLoader(val_ds, batch_size=1, num_workers=4, collate_fn=list_data_collate)
101101
dice_metric = DiceMetric(include_background=True, reduction="mean", get_not_nans=False)
102-
post_trans = Compose([EnsureType(), Activations(sigmoid=True), AsDiscrete(threshold_values=True)])
102+
post_trans = Compose([EnsureType(), Activations(sigmoid=True), AsDiscrete(threshold=0.5)])
103103
# create UNet, DiceLoss and Adam optimizer
104104
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
105105
model = monai.networks.nets.UNet(

‎3d_classification/ignite/densenet_training_dict.py

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -126,7 +126,7 @@ def prepare_batch(batch, device=None, non_blocking=False):
126126
# add evaluation metric to the evaluator engine
127127
val_metrics = {metric_name: ROCAUC()}
128128

129-
post_label = Compose([EnsureType(), AsDiscrete(to_onehot=True, num_classes=2)])
129+
post_label = Compose([EnsureType(), AsDiscrete(to_onehot=2)])
130130
post_pred = Compose([EnsureType(), Activations(softmax=True)])
131131
# Ignite evaluator expects batch=(img, label) and returns output=(y_pred, y) at every iteration,
132132
# user can add output_transform to return other values

‎3d_classification/torch/densenet_training_dict.py

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -81,7 +81,7 @@ def main():
8181
]
8282
)
8383
post_pred = Compose([EnsureType(), Activations(softmax=True)])
84-
post_label = Compose([EnsureType(), AsDiscrete(to_onehot=True, num_classes=2)])
84+
post_label = Compose([EnsureType(), AsDiscrete(to_onehot=2)])
8585

8686
# Define dataset, data loader
8787
check_ds = monai.data.Dataset(data=train_files, transform=train_transforms)

‎3d_segmentation/brats_segmentation_3d.ipynb

Lines changed: 3 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -455,7 +455,7 @@
455455
"dice_metric_batch = DiceMetric(include_background=True, reduction=\"mean_batch\")\n",
456456
"\n",
457457
"post_trans = Compose(\n",
458-
" [EnsureType(), Activations(sigmoid=True), AsDiscrete(threshold_values=True)]\n",
458+
" [EnsureType(), Activations(sigmoid=True), AsDiscrete(threshold=0.5)]\n",
459459
")\n",
460460
"\n",
461461
"\n",
@@ -815,7 +815,7 @@
815815
" to_tensor=True,\n",
816816
" ),\n",
817817
" Activationsd(keys=\"pred\", sigmoid=True),\n",
818-
" AsDiscreted(keys=\"pred\", threshold_values=True),\n",
818+
" AsDiscreted(keys=\"pred\", threshold=0.5),\n",
819819
"])"
820820
]
821821
},
@@ -899,7 +899,7 @@
899899
"name": "python",
900900
"nbconvert_exporter": "python",
901901
"pygments_lexer": "ipython3",
902-
"version": "3.8.10"
902+
"version": "3.8.12"
903903
}
904904
},
905905
"nbformat": 4,

‎3d_segmentation/challenge_baseline/run_net.py

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -172,7 +172,7 @@ def train(data_folder=".", model_folder="runs"):
172172

173173
# create evaluator (to be used to measure model quality during training
174174
val_post_transform = monai.transforms.Compose(
175-
[EnsureTyped(keys=("pred", "label")), AsDiscreted(keys=("pred", "label"), argmax=(True, False), to_onehot=True, num_classes=2)]
175+
[EnsureTyped(keys=("pred", "label")), AsDiscreted(keys=("pred", "label"), argmax=(True, False), to_onehot=2)]
176176
)
177177
val_handlers = [
178178
ProgressBar(),

‎3d_segmentation/ignite/unet_evaluation_array.py

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -65,7 +65,7 @@ def main(tempdir):
6565
roi_size = (96, 96, 96)
6666
sw_batch_size = 4
6767

68-
post_trans = Compose([EnsureType(), Activations(sigmoid=True), AsDiscrete(threshold_values=True)])
68+
post_trans = Compose([EnsureType(), Activations(sigmoid=True), AsDiscrete(threshold=0.5)])
6969
save_image = SaveImage(output_dir="tempdir", output_ext=".nii.gz", output_postfix="seg")
7070

7171
def _sliding_window_processor(engine, batch):

‎3d_segmentation/ignite/unet_evaluation_dict.py

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -72,7 +72,7 @@ def main(tempdir):
7272
roi_size = (96, 96, 96)
7373
sw_batch_size = 4
7474

75-
post_trans = Compose([EnsureType(), Activations(sigmoid=True), AsDiscrete(threshold_values=True)])
75+
post_trans = Compose([EnsureType(), Activations(sigmoid=True), AsDiscrete(threshold=0.5)])
7676
save_image = SaveImage(output_dir="tempdir", output_ext=".nii.gz", output_postfix="seg")
7777

7878
def _sliding_window_processor(engine, batch):

‎3d_segmentation/ignite/unet_training_array.py

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -151,8 +151,8 @@ def main(tempdir):
151151
# add evaluation metric to the evaluator engine
152152
val_metrics = {metric_name: MeanDice()}
153153

154-
post_pred = Compose([EnsureType(), Activations(sigmoid=True), AsDiscrete(threshold_values=True)])
155-
post_label = Compose([EnsureType(), AsDiscrete(threshold_values=True)])
154+
post_pred = Compose([EnsureType(), Activations(sigmoid=True), AsDiscrete(threshold=0.5)])
155+
post_label = Compose([EnsureType(), AsDiscrete(threshold=0.5)])
156156

157157
# Ignite evaluator expects batch=(img, seg) and returns output=(y_pred, y) at every iteration,
158158
# user can add output_transform to return other values

‎3d_segmentation/ignite/unet_training_dict.py

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -180,8 +180,8 @@ def prepare_batch(batch, device=None, non_blocking=False):
180180
# add evaluation metric to the evaluator engine
181181
val_metrics = {metric_name: MeanDice()}
182182

183-
post_pred = Compose([EnsureType(), Activations(sigmoid=True), AsDiscrete(threshold_values=True)])
184-
post_label = Compose([EnsureType(), AsDiscrete(threshold_values=True)])
183+
post_pred = Compose([EnsureType(), Activations(sigmoid=True), AsDiscrete(threshold=0.5)])
184+
post_label = Compose([EnsureType(), AsDiscrete(threshold=0.5)])
185185

186186
# Ignite evaluator expects batch=(img, seg) and returns output=(y_pred, y) at every iteration,
187187
# user can add output_transform to return other values

‎3d_segmentation/spleen_segmentation_3d.ipynb

Lines changed: 5 additions & 5 deletions
Original file line numberDiff line numberDiff line change
@@ -471,8 +471,8 @@
471471
"best_metric_epoch = -1\n",
472472
"epoch_loss_values = []\n",
473473
"metric_values = []\n",
474-
"post_pred = Compose([EnsureType(), AsDiscrete(argmax=True, to_onehot=True, num_classes=2)])\n",
475-
"post_label = Compose([EnsureType(), AsDiscrete(to_onehot=True, num_classes=2)])\n",
474+
"post_pred = Compose([EnsureType(), AsDiscrete(argmax=True, to_onehot=2)])\n",
475+
"post_label = Compose([EnsureType(), AsDiscrete(to_onehot=2)])\n",
476476
"\n",
477477
"for epoch in range(max_epochs):\n",
478478
" print(\"-\" * 10)\n",
@@ -720,8 +720,8 @@
720720
" nearest_interp=False,\n",
721721
" to_tensor=True,\n",
722722
" ),\n",
723-
" AsDiscreted(keys=\"pred\", argmax=True, to_onehot=True, num_classes=2),\n",
724-
" AsDiscreted(keys=\"label\", to_onehot=True, num_classes=2),\n",
723+
" AsDiscreted(keys=\"pred\", argmax=True, to_onehot=2),\n",
724+
" AsDiscreted(keys=\"label\", to_onehot=2),\n",
725725
"])"
726726
]
727727
},
@@ -799,7 +799,7 @@
799799
"name": "python",
800800
"nbconvert_exporter": "python",
801801
"pygments_lexer": "ipython3",
802-
"version": "3.8.10"
802+
"version": "3.8.12"
803803
}
804804
},
805805
"nbformat": 4,

‎3d_segmentation/spleen_segmentation_3d_lightning.ipynb

Lines changed: 3 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -241,8 +241,8 @@
241241
" norm=Norm.BATCH,\n",
242242
" )\n",
243243
" self.loss_function = DiceLoss(to_onehot_y=True, softmax=True)\n",
244-
" self.post_pred = Compose([EnsureType(), AsDiscrete(argmax=True, to_onehot=True, num_classes=2)])\n",
245-
" self.post_label = Compose([EnsureType(), AsDiscrete(to_onehot=True, num_classes=2)])\n",
244+
" self.post_pred = Compose([EnsureType(), AsDiscrete(argmax=True, to_onehot=2)])\n",
245+
" self.post_label = Compose([EnsureType(), AsDiscrete(to_onehot=2)])\n",
246246
" self.dice_metric = DiceMetric(include_background=False, reduction=\"mean\", get_not_nans=False)\n",
247247
" self.best_val_dice = 0\n",
248248
" self.best_val_epoch = 0\n",
@@ -704,7 +704,7 @@
704704
"name": "python",
705705
"nbconvert_exporter": "python",
706706
"pygments_lexer": "ipython3",
707-
"version": "3.8.10"
707+
"version": "3.8.12"
708708
}
709709
},
710710
"nbformat": 4,

‎3d_segmentation/torch/unet_evaluation_array.py

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -52,7 +52,7 @@ def main(tempdir):
5252
# sliding window inference for one image at every iteration
5353
val_loader = DataLoader(val_ds, batch_size=1, num_workers=1, pin_memory=torch.cuda.is_available())
5454
dice_metric = DiceMetric(include_background=True, reduction="mean", get_not_nans=False)
55-
post_trans = Compose([EnsureType(), Activations(sigmoid=True), AsDiscrete(threshold_values=True)])
55+
post_trans = Compose([EnsureType(), Activations(sigmoid=True), AsDiscrete(threshold=0.5)])
5656
saver = SaveImage(output_dir="./output", output_ext=".nii.gz", output_postfix="seg")
5757
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
5858
model = UNet(

‎3d_segmentation/torch/unet_evaluation_dict.py

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -60,7 +60,7 @@ def main(tempdir):
6060
# sliding window inference need to input 1 image in every iteration
6161
val_loader = DataLoader(val_ds, batch_size=1, num_workers=4, collate_fn=list_data_collate)
6262
dice_metric = DiceMetric(include_background=True, reduction="mean", get_not_nans=False)
63-
post_trans = Compose([EnsureType(), Activations(sigmoid=True), AsDiscrete(threshold_values=True)])
63+
post_trans = Compose([EnsureType(), Activations(sigmoid=True), AsDiscrete(threshold=0.5)])
6464
saver = SaveImage(output_dir="./output", output_ext=".nii.gz", output_postfix="seg")
6565
# try to use all the available GPUs
6666
devices = [torch.device("cuda" if torch.cuda.is_available() else "cpu")]

‎3d_segmentation/torch/unet_inference_dict.py

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -84,7 +84,7 @@ def main(tempdir):
8484
# to ensure a smooth output, then execute `AsDiscreted` transform
8585
to_tensor=True, # convert to PyTorch Tensor after inverting
8686
),
87-
AsDiscreted(keys="pred", threshold_values=True),
87+
AsDiscreted(keys="pred", threshold=0.5),
8888
SaveImaged(keys="pred", meta_keys="pred_meta_dict", output_dir="./out", output_postfix="seg", resample=False),
8989
])
9090

‎3d_segmentation/torch/unet_training_array.py

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -81,7 +81,7 @@ def main(tempdir):
8181
val_ds = ImageDataset(images[-20:], segs[-20:], transform=val_imtrans, seg_transform=val_segtrans)
8282
val_loader = DataLoader(val_ds, batch_size=1, num_workers=4, pin_memory=torch.cuda.is_available())
8383
dice_metric = DiceMetric(include_background=True, reduction="mean", get_not_nans=False)
84-
post_trans = Compose([EnsureType(), Activations(sigmoid=True), AsDiscrete(threshold_values=True)])
84+
post_trans = Compose([EnsureType(), Activations(sigmoid=True), AsDiscrete(threshold=0.5)])
8585

8686
# create UNet, DiceLoss and Adam optimizer
8787
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")

‎3d_segmentation/torch/unet_training_dict.py

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -104,7 +104,7 @@ def main(tempdir):
104104
val_ds = monai.data.Dataset(data=val_files, transform=val_transforms)
105105
val_loader = DataLoader(val_ds, batch_size=1, num_workers=4, collate_fn=list_data_collate)
106106
dice_metric = DiceMetric(include_background=True, reduction="mean", get_not_nans=False)
107-
post_trans = Compose([EnsureType(), Activations(sigmoid=True), AsDiscrete(threshold_values=True)])
107+
post_trans = Compose([EnsureType(), Activations(sigmoid=True), AsDiscrete(threshold=0.5)])
108108
# create UNet, DiceLoss and Adam optimizer
109109
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
110110
model = monai.networks.nets.UNet(

‎3d_segmentation/unet_segmentation_3d_catalyst.ipynb

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -377,7 +377,7 @@
377377
"\n",
378378
"dice_metric = DiceMetric(include_background=True, reduction=\"mean\")\n",
379379
"post_trans = Compose(\n",
380-
" [EnsureType(), Activations(sigmoid=True), AsDiscrete(threshold_values=True)]\n",
380+
" [EnsureType(), Activations(sigmoid=True), AsDiscrete(threshold=0.5)]\n",
381381
")"
382382
]
383383
},
@@ -678,7 +678,7 @@
678678
"name": "python",
679679
"nbconvert_exporter": "python",
680680
"pygments_lexer": "ipython3",
681-
"version": "3.8.10"
681+
"version": "3.8.12"
682682
}
683683
},
684684
"nbformat": 4,

‎3d_segmentation/unet_segmentation_3d_ignite.ipynb

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -330,7 +330,7 @@
330330
"# add evaluation metric to the evaluator engine\n",
331331
"val_metrics = {metric_name: MeanDice()}\n",
332332
"post_pred = Compose(\n",
333-
" [EnsureType(), Activations(sigmoid=True), AsDiscrete(threshold_values=True)]\n",
333+
" [EnsureType(), Activations(sigmoid=True), AsDiscrete(threshold=0.5)]\n",
334334
")\n",
335335
"post_label = Compose([EnsureType(), AsDiscrete(threshold_values=True)])\n",
336336
"# Ignite evaluator expects batch=(img, seg) and\n",
@@ -541,7 +541,7 @@
541541
"name": "python",
542542
"nbconvert_exporter": "python",
543543
"pygments_lexer": "ipython3",
544-
"version": "3.8.10"
544+
"version": "3.8.12"
545545
}
546546
},
547547
"nbformat": 4,

‎3d_segmentation/unetr_btcv_segmentation_3d.ipynb

Lines changed: 4 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -681,8 +681,8 @@
681681
"\n",
682682
"max_iterations = 25000\n",
683683
"eval_num = 500\n",
684-
"post_label = AsDiscrete(to_onehot=True, num_classes=14)\n",
685-
"post_pred = AsDiscrete(argmax=True, to_onehot=True, num_classes=14)\n",
684+
"post_label = AsDiscrete(to_onehot=14)\n",
685+
"post_pred = AsDiscrete(argmax=True, to_onehot=14)\n",
686686
"dice_metric = DiceMetric(include_background=True, reduction=\"mean\", get_not_nans=False)\n",
687687
"global_step = 0\n",
688688
"dice_val_best = 0.0\n",
@@ -833,7 +833,7 @@
833833
],
834834
"metadata": {
835835
"kernelspec": {
836-
"display_name": "Python 3",
836+
"display_name": "Python 3 (ipykernel)",
837837
"language": "python",
838838
"name": "python3"
839839
},
@@ -847,7 +847,7 @@
847847
"name": "python",
848848
"nbconvert_exporter": "python",
849849
"pygments_lexer": "ipython3",
850-
"version": "3.7.10"
850+
"version": "3.8.12"
851851
}
852852
},
853853
"nbformat": 4,

‎3d_segmentation/unetr_btcv_segmentation_3d_lightning.ipynb

Lines changed: 4 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -415,8 +415,8 @@
415415
" ).to(device)\n",
416416
"\n",
417417
" self.loss_function = DiceCELoss(to_onehot_y=True, softmax=True)\n",
418-
" self.post_pred = AsDiscrete(argmax=True, to_onehot=True, num_classes=14)\n",
419-
" self.post_label = AsDiscrete(to_onehot=True, num_classes=14)\n",
418+
" self.post_pred = AsDiscrete(argmax=True, to_onehot=14)\n",
419+
" self.post_label = AsDiscrete(to_onehot=14)\n",
420420
" self.dice_metric = DiceMetric(\n",
421421
" include_background=False, reduction=\"mean\", get_not_nans=False\n",
422422
" )\n",
@@ -771,7 +771,7 @@
771771
],
772772
"metadata": {
773773
"kernelspec": {
774-
"display_name": "Python 3",
774+
"display_name": "Python 3 (ipykernel)",
775775
"language": "python",
776776
"name": "python3"
777777
},
@@ -785,7 +785,7 @@
785785
"name": "python",
786786
"nbconvert_exporter": "python",
787787
"pygments_lexer": "ipython3",
788-
"version": "3.7.10"
788+
"version": "3.8.12"
789789
}
790790
},
791791
"nbformat": 4,

‎acceleration/automatic_mixed_precision.ipynb

Lines changed: 3 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -352,8 +352,8 @@
352352
" optimizer = torch.optim.Adam(model.parameters(), 1e-4)\n",
353353
" scaler = torch.cuda.amp.GradScaler() if amp else None\n",
354354
"\n",
355-
" post_pred = Compose([EnsureType(), AsDiscrete(argmax=True, to_onehot=True, num_classes=2)])\n",
356-
" post_label = Compose([EnsureType(), AsDiscrete(to_onehot=True, num_classes=2)])\n",
355+
" post_pred = Compose([EnsureType(), AsDiscrete(argmax=True, to_onehot=2)])\n",
356+
" post_label = Compose([EnsureType(), AsDiscrete(to_onehot=2)])\n",
357357
"\n",
358358
" dice_metric = DiceMetric(include_background=False, reduction=\"mean\", get_not_nans=False)\n",
359359
"\n",
@@ -870,7 +870,7 @@
870870
"name": "python",
871871
"nbconvert_exporter": "python",
872872
"pygments_lexer": "ipython3",
873-
"version": "3.8.10"
873+
"version": "3.8.12"
874874
}
875875
},
876876
"nbformat": 4,

‎acceleration/dataset_type_performance.ipynb

Lines changed: 3 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -209,8 +209,8 @@
209209
" loss_function = DiceLoss(to_onehot_y=True, softmax=True)\n",
210210
" optimizer = torch.optim.Adam(model.parameters(), 1e-4)\n",
211211
"\n",
212-
" post_pred = Compose([EnsureType(), AsDiscrete(argmax=True, to_onehot=True, num_classes=2)])\n",
213-
" post_label = Compose([EnsureType(), AsDiscrete(to_onehot=True, num_classes=2)])\n",
212+
" post_pred = Compose([EnsureType(), AsDiscrete(argmax=True, to_onehot=2)])\n",
213+
" post_label = Compose([EnsureType(), AsDiscrete(to_onehot=2)])\n",
214214
"\n",
215215
" dice_metric = DiceMetric(include_background=True, reduction=\"mean\", get_not_nans=False)\n",
216216
"\n",
@@ -753,7 +753,7 @@
753753
"name": "python",
754754
"nbconvert_exporter": "python",
755755
"pygments_lexer": "ipython3",
756-
"version": "3.8.10"
756+
"version": "3.8.12"
757757
}
758758
},
759759
"nbformat": 4,

‎acceleration/distributed_training/brats_training_ddp.py

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -283,7 +283,7 @@ def main_worker(args):
283283
dice_metric = DiceMetric(include_background=True, reduction="mean")
284284
dice_metric_batch = DiceMetric(include_background=True, reduction="mean_batch")
285285

286-
post_trans = Compose([EnsureType(), Activations(sigmoid=True), AsDiscrete(threshold_values=True)])
286+
post_trans = Compose([EnsureType(), Activations(sigmoid=True), AsDiscrete(threshold=0.5)])
287287

288288
# start a typical PyTorch training
289289
best_metric = -1

0 commit comments

Comments
 (0)