Single or multi-midality in testing phase

Hi there,

Could you please let me know about the testing phase? I mean the model can be given both modalities in testing phase, or just be tested by a single modality?

Regards, Maysam

Hi Maysam,

For each test-case, we present to your algorithm both CT and MR modalities from that patient, i.e. the CTA-MRA pair:

def predict(ct_image, mr_image) -> np.array:

, irrespective of whether your algorithm uses both arrays or not. In case you only need one of them, simply ignore the other modality input.

More importantly though, for each test case the output of your algorithm must be a prediction array either for the CT or the MR image (depending on the track) and either binary or multi-class (depending on the task). The output mask array must have the same shape as the input image array of the modality that you want to segment (i.e. the main modality of the track you submit for). This is important for the correct evaluation of your segmentation result.

We will also have more details in the github and tutorial pages soon. Hope it helps!

Hi Maysam,

We have just released the repo with the documentations and docker template for submission phase.

Please visit our repo 👉 "TopCoW_Algo_Submission" and it should help you with the submission process.

Feel free to leave an issue or let us know if you have further questions or feedback. Many thanks for your participation and we look forward to your submissions!