Hi! In the challenge documentation, you specified that no external data or pre-trained networks are allowed. Can you clarify whether this also extends to loss functions that may rely on pre-trained networks? Additionally, can authors be added to the submitted paper if they were not signatories on the original data-sharing agreement (i.e. if they provided contributions without analysing any data)? Thank you!

Created by Levente Baljer levente1
Hi @levente1, Thank you for your thoughtful questions. Regarding the use of perceptual loss: Yes, you're allowed to use pre-trained networks for perceptual loss, provided that these models are standard, publicly available (e.g., via PyTorch), and were not trained using any ultra-low-field MRI data. Since the perceptual loss is used for feature comparison and not direct prediction or fine-tuning on our dataset, this usage aligns with our current rules. Regarding authorship on the short paper: We ask all participants to strictly follow the authorship criteria outlined by MICCAI. Individuals who did not directly contribute to the data analysis, modeling, or interpretation as defined in those guidelines should not be listed as authors. Acknowledging general advice or support in the Acknowledgements section is perfectly appropriate and encouraged. Let us know if you have any further questions. The ULF-EnC Challenge Organizers

Pre-trained networks and paper submission page is loading…