About WSIs Data Preparation

[future97]

Based on the tutorial "Tutorial-Processing WSIs for MIL" and its recommended parameters, I followed the workflow to extract patches and generate a MIL dataset using CLAM. Then, I used CTransPath to extract features. However, I noticed that the number of patches per sample in the ‘.pt’ files I generated is significantly higher than in the files you provided. Did you perform any sampling in your process? Additionally, could you please share the parameter settings you used during WSIs Data Preparation?

For example,
In my_data_dir:
slide_id: TCGA-3M-AB46-01Z-00-DX1.70F638A0-BDCB-4BDE-BBFE-6D78A1A08C5B wsi_bag.shape: torch.Size([18661, 768])
slide_id: TCGA-3M-AB47-01Z-00-DX1.D7D70922-A91F-4798-8D6E-DFD84D145A59 wsi_bag.shape: torch.Size([18106, 768])
slide_id: TCGA-B7-5816-01Z-00-DX1.1B05F96A-D5E2-4366-A098-A861313F3461 wsi_bag.shape: torch.Size([7944, 768])
slide_id: TCGA-BR-6452-01Z-00-DX1.aec8246e-9dce-436d-9faa-df8fa1f6e47c wsi_bag.shape: torch.Size([15916, 768])
slide_id: TCGA-BR-6453-01Z-00-DX1.2e52d64c-e34f-47a9-9e5e-8954b5ef18ca wsi_bag.shape: torch.Size([12366, 768])
In PIBD_data_dir:
slide_id: TCGA-3M-AB46-01Z-00-DX1.70F638A0-BDCB-4BDE-BBFE-6D78A1A08C5B wsi_bag.shape: torch.Size([4798, 768])
slide_id: TCGA-3M-AB47-01Z-00-DX1.D7D70922-A91F-4798-8D6E-DFD84D145A59 wsi_bag.shape: torch.Size([4570, 768])
slide_id: TCGA-B7-5816-01Z-00-DX1.1B05F96A-D5E2-4366-A098-A861313F3461 wsi_bag.shape: torch.Size([2045, 768])
slide_id: TCGA-BR-6452-01Z-00-DX1.aec8246e-9dce-436d-9faa-df8fa1f6e47c wsi_bag.shape: torch.Size([4027, 768])
slide_id: TCGA-BR-6453-01Z-00-DX1.2e52d64c-e34f-47a9-9e5e-8954b5ef18ca wsi_bag.shape: torch.Size([3131, 768])

For Segmenting-and-Patching-Slides, the parameter setting:

For Extracting-Patch-Features, the parameter setting:

I look forward to your reply and greatly appreciate your assistance.

[zylbuaa]

Thank you for reaching out and for catching that discrepancy. We extract features at level 1 of the WSI, corresponding to a magnification factor of 10x. We appreciate your keen observation and apologize for the confusion; there is a little typo in the Appendix Bag construction part. We'll correct it in the next version.

[future97]

In other words, I just need to set the MAG parameter to 10 using the tutorial "Tutorial-Processing WSIs for MIL", right? Thank you sincerely for your help and guidance!

[2687689464]

Hello, teacher, very sorry to disturb you, I have read our PROTOTYPICAL INFORMATION BOTTLENECKING AND
After the paper DISENTANGLING FOR MULTIMODAL CANCER SURVIVAL PREDICTION, I encountered some difficulties when running the relevant code, so I would like to consult the teacher. When the code was running, there was an error in the picture and the corresponding file could not be found, so I would like to consult the teacher
/home/zhany0x/Documents/data/ctranspath-pibd/stad/tiles-l1-s224/feats-l1-s224-CTransPath-sampler/pt_files
Whether this directory file is a file created by the code running itself, but I actually did not create success in the process of running the code, the above is my question, looking forward to the teacher's answer.