Running docker container on local device I get final outputs. But when upload algorithm from the github repository, it throws monai transform loadimage error.

Hello, I build the container on my local machine, like this:

> docker build -t docker_just_nnunet -f ./Dockerfile .

I also ran the docker container I built using this command:

> docker run -it --gpus all docker_just_nnunet

I get output "okay" result, when I run docker container on my local machine and it works. But I passed the link to github repository for loading my algorithm in challenge. But after passing the algorithm I get error as:

RuntimeError: applying transform <monai.transforms.io.dictionary.LoadImaged object at 0x7f9e1211df30>

I think the command I used for build and run the docker container is different from the build used by the evaluation checker bot, so the error is persisting!

Hi,

In your error messange, it shows: RuntimeError: Could not create IO object for reading file input/images/pelvic-fracture-ct/099.mha

Please refer to the algorithm template to handle the IO operation. The container is supposed to be run multiple times and will read the only mha file in a designated folder/docker volume. Please do not specify the file name.

Best,

Yudi

Hello, Sorry for the problem I've encountered. I have initialized the images like this,

images = sorted(glob(join(INPUT_PATH, '*.mha')))

So you mean to say I should just provide the path containing the image like this (without specifying the image filenames):

images = sorted(glob(INPUT_PATH))

And the algorithm takes each image one by one, and stores the result?

Thank you for support! It means a lot..

No worries!

https://github.com/YzzLiu/PENGWIN-example-algorithm/blob/main/PENGWIN-challenge-packages/preliminary-development-phase-ct/inference.py

pelvic_facture_ct = load_image_file_as_array(location=INPUT_PATH / "images/pelvic-fracture-ct")

This should give you a single image array correctly. The algorithm container will be re-initialized for every single case.

It is unclear to me why your error message shows 099.mha, which is not in the validation set. So I guess you had specified 099 in your code, which you should not.

Best,

Yudi

Thank you! that solved the problem.