Some problems abut dataset

[fyb99]

Hi, thanks for your solid research and well-structured code. I have a question. I find that this work use MIMIC 2.0 as the main data while your previous work MGCA use MIMICA-CXR-JPG for pretraining. Since MIMIC 2.0 has over 4.6TB data volumes and requires a very long time to download, does your method support training on MIMIC-CSR-JPG?

[fuying-wang]

Hi,

Yes. We are using the MIMIC-CXR-JPG v2.0 for pretraining.

[fyb99]

Thanks for your reply!. Since the link in the github page point to the DCM-based MIMIC 2.0, I want to double check do you use this dataset for your pretraining (https://physionet.org/content/mimic-cxr-jpg/2.1.0/)

[fuying-wang]

Yes. Please use it.