Skip to content

Latest commit

 

History

History
36 lines (29 loc) · 1.62 KB

File metadata and controls

36 lines (29 loc) · 1.62 KB

ProstateCancerRiskPrediction Algorithm

The source code for the algorithm container for ProstateCancerRiskPrediction, generated with evalutils version 0.4.2 using Python 3.9.

Here you can find an example on how to build your inference code for the prostate challenge.

The model uses two main inputs:

  • Prostate T2w MRI (axial-t2-prostate-mri), corresponding to the image in mha format.
  • Clinical data (psa-and-age.json), which includes the patient age and PSA.

Then, after risk prediction, the model has to provide two main output:

  • The risk score (prostate-cancer-risk-score.json) which can have the values "Low" or "High".
  • The risk score likelihood (prostate-cancer-risk-score-likelihood.json) which represents the probability given by the model.

In the main file you will find an example on how to read the input data and how to generate the output files. Outputs are generated randomly as an example.

Build the Docker image

To build the docker image open a terminal and change your directory to the ProstateCancerRiskPrediction folder. Then run the following command:

docker build -f Dockerfile -t <your_docker_image_name> .

Change <your_docker_image_name> by the name you want the Docker image to have.

Finally, to submit your Docker image to grand-challenge you need to encapsulate it in a file. The generate the file to upload, run the following command:

docker save -o <output_file_name>.tar.gz <your_docker_image_name>

With this command, a file named <output_file_name>.tar.gz will be generated. This is the file you have to upload to grand challenge when submitting a new algorithm