Not define text encoder or image encoder

[keepFaith-sdy]

As the attached image shows, the Prior class does not define image encoder or text encoder?

[QtacierP]

Yes. We have pre-defined encoders using omegaconf (you can find the definition in yaml files)

[keepFaith-sdy]

As I run the pretrain command you provided in readme.md, it throws out the error as the image showed.

Besides, in the data operation, those reports include findings but do not provide a final impression, and you decided to exclude them from the data set. Could you kindly tell me the reasons for ruling out the X-ray reports under this circumstances? Are there any specific concerns, such as implications for data completeness or quality? It does not mention in the paper or supplementary file, thanks!

[QtacierP]

Thank you for your attention to our work, and I apologize for the delayed reply.

1. I guess the issue may be related to some typos in the command, particularly with “+experiments/pre_train=train_prior.” If there’s a typo, the code may fail to detect the structural YAML file, preventing the creation of a pre-trained model instance.

2. Thank you for your thorough review. Since local alignment is done at the sentence level, reports with very few sentences are discarded. We mention this in Section 4.1 of the paper: "Additionally, we discard short reports containing fewer than four sentences, resulting in 182,475 image-report pairs." As findings sections are typically quite short, reports that only include findings are also excluded.