About data augmentation

[zerone-fg]

First, I am very honored to read your paper and also impressed by your excellent experiment result on synapse dataset, and I noticed that you used a lot of extra data augment methods in your dataset_synapse.py, like Gaussiannoise, while TransUnet didn't. And I am wondering how much performance these data augment methods help your model get?And how did your model perform without these methods?
Thank you for your reply.