FAQ
This page is constantly being updated with question from potential challenge participants.
Can I use my own data to train my models for the challenge?
Response: Teams are free to use their own simulated data to train/pretrain their models, but teams are not allowed to use other sources of in vivo data other than what is being provided by the challenge organizers. One of the main goals of the challenge is to verify whether Edited-MRS reconstruction models trained on simulated data and potentially fine-tuned on small amounts of in vivo data translate well to in vivo test sets.
2. Do I need to participate in all three tracks of the challenge?
Response: Teams are welcome to participate in as many (or few) of the challenge tracks as they wish.
3. Will all tracks be submitted at the same time? What is the format?
Response: All tracks must be submitted as an H5 file (see the github repository for example submissions) by the deadline March 17th 2023 March 27th, 2023 @ 23:59 MST via email to one or both of the organizers Rodrigo (rodrigo.pommotber1@ucalgary.ca) or Hanna (hanna.bugler@ucalgary.ca).
Intermediate submissions are allowed with a maximum of 1 submission per 48 hour window per track (i.e. a team can submit "track01_simulated.H5" and submit "track02_invivo_homogeneous.H5" within 48 hours of each other) until March 15th 2023 March 25th, 2023.
4. Is in-person attendance at the 2023 IEEE ISBI Conference Mandatory?
Response: Our intent is to host a hybrid event at the IEEE ISBI 2023 conference. Teams who submit results which outperform the baseline will be invited to present their solution in-person. However, should travel to the conference not be possible, teams will be invited to make their presentation over Zoom (or equivalent).