Implement dice cross entropy loss

[yiheng-wang-nv]

Signed-off-by: yiheng-wang-nv vennw@nvidia.com

Fixes #1157 .

Status

Ready

Types of changes

• 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).
• New tests added to cover the changes.
• Integration tests passed locally by running ./runtests.sh --codeformat --coverage.
• Quick tests passed locally by running ./runtests.sh --quick.
• In-line docstrings updated.
• Documentation updated, tested make html command in the docs/ folder.

[wyli]

/black

[wyli]

thanks!

[sdesutter]

I was trying to import DiceCELoss as I found it in the documentation of MONAI, but I keep receiving an ImportError. Does this mean this is still in progress or might there be some kind of issue?

Thanks!

[yiheng-wang-nv]

I was trying to import DiceCELoss as I found it in the documentation of MONAI, but I keep receiving an ImportError. Does this mean this is still in progress or might there be some kind of issue?

Thanks!

Hi @sdesutter , could you post the error message here? The loss should be imported without errors as the corresponding unit test checked. I just downloaded the latest docker image ( docker pull projectmonai/monai:latest ) and tried to import this loss function and it also worked.

[sdesutter]

I was trying to import DiceCELoss as I found it in the documentation of MONAI, but I keep receiving an ImportError. Does this mean this is still in progress or might there be some kind of issue?
Thanks!

Hi @sdesutter , could you post the error message here? The loss should be imported without errors as the corresponding unit test checked. I just downloaded the latest docker image ( docker pull projectmonai/monai:latest ) and tried to import this loss function and it also worked.

The error I'm getting is:
ImportError: cannot import name 'DiceCELoss' from 'monai.losses' (/opt/anaconda3/lib/python3.7/site-packages/monai/losses/init.py)

[rijobro]

Inside of python, could you run:

import monai
monai.config.print_config()

I wonder if, for example, you are using an older version of the code.

[sdesutter]

Inside of python, could you run:

import monai
monai.config.print_config()

I wonder if, for example, you are using an older version of the code.

This is what I got:

MONAI version: 0.4.0
Numpy version: 1.17.2+intel.0
Pytorch version: 1.5.1
MONAI flags: HAS_EXT = False, USE_COMPILED = False
MONAI rev id: 0563a44

Optional dependencies:
Pytorch Ignite version: 0.3.0
Nibabel version: 3.0.2
scikit-image version: 0.15.0
Pillow version: 6.2.0
Tensorboard version: 2.0.2
gdown version: NOT INSTALLED or UNKNOWN VERSION.
TorchVision version: 0.6.1
ITK version: NOT INSTALLED or UNKNOWN VERSION.
tqdm version: 4.50.2
lmdb version: NOT INSTALLED or UNKNOWN VERSION.
psutil version: 5.6.3

[rijobro]

Ok, you're using version 0.4.0, and this feature was implemented after that. You'll probably want the master branch:

pip install git+https://github.com/Project-MONAI/MONAI#egg=MONAI

Once you have the master branch, the printed config output should have something like:

MONAI version: 0.4.0+22.ga4ef691

Here, 22 is the number of commits since the previous tag, 0.4.0, and ga4ef691 is the most recent git commit hash.

[sdesutter]

Ok, you're using version 0.4.0, and this feature was implemented after that. You'll probably want the master branch:

pip install git+https://github.com/Project-MONAI/MONAI#egg=MONAI

Once you have the master branch, the printed config output should have something like:

MONAI version: 0.4.0+22.ga4ef691

Here, 22 is the number of commits since the previous tag, 0.4.0, and ga4ef691 is the most recent git commit hash.

This worked, the import is working now.
Thanks a lot for your help!