Participants are allowed to use additional publicly available annotated datasets and publicly available pre-trained models. These pre-trained models should not be generated using previous BraTS challenge datasets. Participants should not use data from other BraTS sub-challenges to train their models unless explicitly allowed on the challenge webpage. For example, participants developing method for Meningioma segmentation should not use data from pediatric challenge. In addition to this there are some challenge specific rules given below: For Segmentation - **BraTS-Africa Challenge**: Participants are allowed to use data from BraTS Adult Glioma Challenge. For** BraTS Generalizability Across Tumors (BraTS-GoAT) challenge**: participants will ONLY be allowed to use the data provided through the BraTS-GoAT sub-challenge website. The use of any additional data, either private or publicly available (including previous BraTS challenges), will result in the announcement of immediate disqualification from current and future BraTS challenges in the challenge forum. If participants are using private data (from their own institutions) for extending the provided data, it should be mentioned in the manuscript and an additional MLCube must be submitted during the testing phase of the challenge. This is due to our intentions to provide a fair comparison among the participating methods.

Created by Ujjwal Baid ujjwalbaid
Thank you very much!
Yes. Authors will have to report this in their manuscript.
So for clarification, there will be three different submissions? - MLCube using the current rules, so ONLY the BraTs dataset of that challenge - MLCube using BraTS + publicly available data/pretrained models - MLCube using BraTS + publicly available data/pretrained models + private data
Hi, We are allowing this as an additional MLCube submission for those who are using the private datasets. Unfortunately, the submission deadline will not be extended.
Will the July 31st deadline be extended then? While I commend this change, as it allows for better-performing models and will enable new techniques, retraining models within this short timeframe seems difficult.
Yes, this applies to BraTS-PED as well. Appropriate changes will be made in webpages and in Arxiv manuscripts soon. If you are using private data for extending the provided data, it should be mentioned in the manuscript and an additional MLCube must be submitted during the testing phase of the challenge.
Does this apply to BraTS-PED as well? Their paper explicitly states this is not allowed. Source: Participants are prohibited from training their algorithm on any additional public and/or private data (from their own institutions) besides the provided BraTS-PEDs data, or using models pretrained on any other dataset. This restriction is imposed to allow for a fair comparison among the participating methods. However, participants can use additional public and/or private data only for publication of their scientific papers, if they also provide a report of their results on the data from the BraTS-PEDs challenge alone, and discuss potential differences in the obtained results. See https://arxiv.org/abs/2404.15009 ,[Direct text link](https://ar5iv.labs.arxiv.org/html/2404.15009#:~:text=Participants%20are%20prohibited,the%20obtained%20results.)

Clarification on usage of Challenge datasets, pre-trained models page is loading…