Can we use additional dataset and pretrained models in the network training?

I just wonder whether it is allowed to use some pretrained models during the network training. For example, the pretrained weight on some other brain lesion segmentation tasks or just on ImageNet. Thanks.

Hi, According to our ATLAS participation policies (https://zenodo.org/record/6362388#.YqBrWJNBzmF), you can use other publicly available datasets and open pre-trained nets for this task. Best, EDLR