Dear Organisers, Could you please clarify: For the **test phase**, 1. whether the central slab or the full volume would be used for evaluation? I get contradictory instructions in this post (https://www.synapse.org/Synapse:syn72060672/discussion/threadId=13179?replyid=36306) and this document (https://github.com/MRIxFields/MRIxFields2026/blob/main/Submission/testing-2026/doc/participant_guide.md). 2. How many cases would be involved in each task? Again contradictory information in this page (https://github.com/MRIxFields/MRIxFields2026/blob/main/Submission/testing-2026/doc/participant_guide.md#workload) and this page (https://www.synapse.org/Synapse:syn72060672/wiki/639760). The former says 5/5/2 cases for task 1/2/3, but the latter says 20 cases. 3. The runtime limit for task 3 is 20 hours in total, regardless of how many cases? Your help would be highly appreciated.

Created by YULIANG HUANG hyliang
Thank you for your questions. 1. For the test phase, the full volume (all slices) will be evaluated. This differs from the validation phase, where only the central slab was used due to computational constraints. 2. All 20 test cases will be evaluated, but in a designed way. The evaluation workflow is slightly more involved, as illustrated in our example figure: https://www.synapse.org/Synapse:syn72060672/wiki/639760. 3. For all three tasks, the runtime limit is 20 hours in total to complete inference on all assigned test cases. The case allocation for each task is shown in the example figure, and we have updated both the relevant webpage and the directory tree to make this clearer. We hope this clarifies the evaluation procedure. Best regards, MRIxFields2026 Organizing Committee

Clarification about test phase submission page is loading…