Hi! I have a question regarding what our models/algorithms are expected to **output**. Can it be a single nifti file (regardless of task) with all the segmentation classes inside it? I.e., with all structures 1 through 8 (left hipp, right hipp, ..., right lentiform). That is, does the evaluation code do the task of "separating" the correct classes? Or should I separate explicitly (with different files) hippocampus and basal ganglia? EDIT: Also, I would like to know how the container runs during the test phase. Is there a container run per test case (ciso scan in task 2a/2b), or does the container run one time in which it is supposed to process the N test cases found in the "/input" folder? Finally: On [this page](https://www.synapse.org/Synapse:syn68633106/wiki/633393) about how to submit for the test phase there is a mention to _MLCubes_. However, on [this other page](https://www.synapse.org/Synapse:syn68633106/wiki/633398) also about submitting to the test phase, there is only mention to Docker. Are both accepted ways of submission? I am familiar with Docker, not with MLCube, so if I could do it w Docker it would be better for me. Thanks, Vicent

Created by Vicent Caselles Ballester vcasellesb
Dear @LISA_mri_challenge, Thanks for your reply. I have checked that wiki page you sent, from which I understand that I should provide two files as output -- one ending in _baga_, the other ending in _hipp_. I just wanted to note that for the validation phase I did not do it like this, and I still got decent results. I apologize because I'm probably being too obtuse. I just want to make sure I understand before the testing phase starts. Best, Vicent
Hi Vicent, Please check here https://www.synapse.org/Synapse:syn68633106/wiki/633398. Best, The LISA 2025 Challenge Organizers
Thanks for your reply and support! Thus, just to make sure, the naming convention of the output files should be: LISA\_HF\_XXXX\_seg\_prediction.nii.gz Where XXXX is the case identifier (4 digits). Best, Vicent
Hi Vicent, For the output format, it should be exactly the same as in the validation phase. If your model’s output passes the validation phase checks, then it will also be accepted in the test phase without modification. For the container execution, in the test phase the platform runs your container once. The complete set of test cases is mounted under /input, and your code should iterate through all cases in that folder and write the results into /output. For the submission method, we only accept Docker images. Please submit your solution as a Docker container; MLCube is not required. Best, The LISA 2025 Challenge Organizers

Question about expected output of docker container run page is loading…