MIMIC-CXR Pretrain Dataset Pre-Processing about cxr-clip

[kihyunU]

If any study is missing "findings", the text column should be [impression]. Sorry for insufficient description.

[SinuoWang]

If any study is missing "findings", the text column should be [impression]. Sorry for insufficient description.

Thank you for your clarifications! I am also curious whether other text cleaning steps are required, such as removing the "\n" in the text, etc. I know this is kinda trivial, but I just wanna double check.

Thanks in advance :)

[kihyunU]

Yes we removed "\n". uppercase letter and lowercase letter don't matter because it will be processed by tokenizer.

[GravitySaika]

Hi! I'm wondering how long does it takes in the augmentation stage, I used 1 A40 GPU to handle it, but is shows that it needs 13h to finish. Is that normal?

[kihyunU]

I think it is normal. This process includes sequential two models with beam search. So it will take some time.

[Orange2107]

Hi @kihyunU
It is a meaningful work.
I would like to ask:
The "text" in each subject is like: [[Findings1,Impression1],[Findings2,Impression2][Findings3,Impression3]...[Findings,Impression]].
Is this correct?
Thank you