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?
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?