If I am understanding the challenge protocol correctly:

The inference.py should evaluate a single .mha CT image file that is to be copied by the organizers to:
    /input/images/thoracic-abdominal-ct/example1_image.mha

The inference.py should output:
     .mha file containing the [1,2,3] mask to:
        /output/images/abdominal-organ-segmentation/example1_image.mha
    .mha file containing the {0.0-1.0} kidney confidence heat map to:
        /output/images/kidney-confidence/example1_image.mha
    .mha file containing the {0.0-1.0} pancreas confidence heat map to:
        /output/images/pancreas-confidence/example1_image.mha
    .mha file containing the {0.0-1.0} liver confidence heat map to:
        /output/images/liver-confidence/example1_image.mha

My understanding is that all of the output files should have the exact same name "example1_image.mha" as the single .mha file that was copied by the organizers into /input/images/thoracic-abdominal-ct/

Please let me know if I am understanding the challenge protocol correctly.