search

Stanford-AIMI / RRG24

Shared task on Large-Scale Radiology Report Generation @ BioNLP ACL'24
https://stanford-aimi.github.io/RRG24/
MIT License
9 stars 0 forks source link

mimic_cxr_sectioned.csv #3

Closed StevenSong closed 6 months ago

StevenSong commented 6 months ago

Thank you guys for putting together this challenge, along with the discharge challenge too!

Just wanted to confirm that mimic_cxr_sectioned.csv should be generated from the source note .txt files using the code from MIMIC-CXR-JPG code repo: https://doi.org/10.5281/zenodo.2591653?

Also a minor suggestion that it's not immediately clear that one should use MIMIC-CXR-JPG for the image dataset either, it's obvious enough after looking at the file structure and the make-interpret-mimic-cxr.py script but maybe others might benefit from these clarifications

jbdel commented 6 months ago

Correct! using this script: https://github.com/MIT-LCP/mimic-cxr/blob/master/txt/create_section_files.py