Preprocessor stuck on running HD-Bet

[chocolatetoast-chiu]

Hi, I'm trying to use the BraTS preprocessor, and it is getting stuck on the "running HD-Bet." Does anyone face this issue before when using this preprocessor? Thank you! I will provide the required information to help you figure it out!

Attachment from the output command line

Thank you!

[neuronflow]

thanks for your interest in btk. Can you provide information about your hardware? Also run the nvidia docker toolkit hello-world please and post the output here.

[chocolatetoast-chiu]

Sure! Thanks for the quick reply! Here is the docker hello-world response:

nvidia-smi response:

running a base CUDA container:

Hardware information: I used cloud computing instance under our university cloud environment:

1. OS: Linux ubuntu focal 20.04
2. RAM: 16GB
3. vCPUs: 8
4. GPU: NVIDIA A100-SXM

[neuronflow]

Sorry, I believe there might be a misunderstanding, please run:

docker run --rm --gpus all nvidia/cuda:11.0-base nvidia-smi

You might need sudo depending on your docker configuration.

from: https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/install-guide.html

[neuronflow]

closing until more details are provided, feel free to reopen.

[azounek]

Hi, I have the same problem. I am running ubuntu 22.04.1 with CUDA 12.0 on WSL2 in Windows 11 and the Toolkit is also stuck on "running HD-Bet".

@chocolatetoast-chiu Did you manage to find a solution in the meantime?

The output of the nvidia-smi command looks fine so I am not sure what went wrong.

Please let me know if I should provide more details. Thank you !

[neuronflow]

Sorry, I overlooked your msg @azounek . Did you find a solution int he meantime? Potentially the same issue as here: https://github.com/neuronflow/BraTS-Toolkit/issues/29

[neuronflow]

Perhaps a good test would be whether HD-BET alone can run on your machines: https://github.com/MIC-DKFZ/HD-BET

@FabianIsensee do you have an idea what could be the issue here?

[chocolatetoast-chiu]

Sorry for not following this issue for a long time!

I did solve this issue by setting the mode to GPU.

        # execute it
	prep.single_preprocess(
		t1File=t1File,
		t1cFile=t1cFile,
		t2File=t2File,
		flaFile=flaFile,
		outputFolder=outputDir + examName,
		# this way the brain extraction will run on cpu
		mode="gpu",
		confirm=False,
		skipUpdate=False,
		gpuid="0",
	)

If the mode was set to 'cpu', then I'm not able to finish my preprocessing. I don't know the exact reason why.

[neuronflow]

Interesting. We have a new preprocessing pipeline now that might solve your problems. Let me know if you want to serve as a beta tester.

[neuronflow]

closed until further details about the cpu mode are provided.