search

HiLab-git / SSL4MIS

Semi Supervised Learning for Medical Image Segmentation, a collection of literature reviews and code implementations.
MIT License
2.19k stars 387 forks source link

BRaTS2019 settings #31

Closed ghost closed 2 years ago

ghost commented 2 years ago

Hi,

Thanks for your fantastic work. One question is why we preprocess the BRaST2019 into a binary problem. According to your paper (https://arxiv.org/pdf/2105.09511v3.pdf), the dataset has four classes (including b.g.). Did anyone do it before? Please share the reference.

Cheers,

Luoxd1996 commented 2 years ago

Hi, Thanks for your attention. You can read the original paper of the BraTS dataset and other relevant works or our lab papers. It's a common setting. Best, Xiangde.

ghost commented 2 years ago

Hi,

I've searched the BRaTS2019 related paper, and for example, in this link: https://paperswithcode.com/sota/brain-tumor-segmentation-on-brats-2019. From your paper Medical Image Segmentation Using Squeeze-and-Expansion Transformers, with the code for the BraTS proceeding in here: https://github.com/askerlee/segtran/blob/462b206570d68ed10d8b98d740f6c920753d0958/code/dataloaders/brats_processing.py#L36 I think you've built four labels (including b.g.) in this approach. But I'm confused why you simply merge them together in here, https://github.com/HiLab-git/SSL4MIS/blob/e8c8b5c5ad36d22748ab7c5745e275f612b1186a/code/dataloaders/brats_proprecessing.py#L105.

Could you please describe the reason for such the difference (or just simply drop me a paper link)?

Cheers

Luoxd1996 commented 2 years ago

Hi, Same response, please read the original paper of the BraTS dataset in TMI 2015 carefully. Best, Xiangde.

ghost commented 2 years ago

Hi,

Thanks for narrowing down the search region from "other relevant works or our lab papers" to the BRaTS 2015 original TMI paper; however, I'm asking for the semi-supervised approach that utilizes the binary labelled setting for the BRaTS2019 dataset (as a ref).

For anyone who also has this concern, this submission: https://arxiv.org/pdf/2112.02508.pdf could be an example of such a semi-supervised setting.

Cheers

Luoxd1996 commented 2 years ago

Hi, Again, your concern has out of the scope of this project, and this project has not a duty to provide the relevant works. Actually, the literature review is your own business, but I also provide a reference[1] for you. Finally, thanks for your attention. [1] Shuai Chen et al. Multi-Task Attention-Based Semi-Supervised Learning for Medical Image Segmentation, In MICCAI2019. Best, Xiangde.