ConTEXTual Net: A Multimodal Vision-Language Model for Segmentation of Pneumothorax

ConTEXTual net is a multimodal vision-langauge model for medical imaging. This github is still under construction for open source use.

Data

Information on the CANDID-PTX dataset can be found in their paper found here: https://pubs.rsna.org/doi/10.1148/ryai.2021210136 And the data can be accessed after signing a data use agreement and an online ethics course: https://pubs.rsna.org/doi/10.1148/ryai.2021210136 Once you have downloaded the data unzip all of the files and move them to the same folder with the file stucture below.

candid_ptx
|   Pneumothorax_reports.csv
|   chest_tube.csv
|   acute_rib_fracture.csv
|
|___chest_radiographs
    |   *images

Usage