Skip to main content
. Author manuscript; available in PMC: 2024 Sep 1.
Published in final edited form as: Proc IEEE Int Symp Biomed Imaging. 2023 Sep 1;2023:10.1109/isbi53787.2023.10230623. doi: 10.1109/isbi53787.2023.10230623

Fig. 1.

Fig. 1.

Four training strategies compared. Stage 1 is the training pipeline with the unlabeled dataset. The model learns features from raw radiologist reports, which serve as a natural source of supervision. Stage 2 is the training pipeline with the labeled dataset. SC: supervised classifier, SMC: supervised multi-label classifier, SSC: self-supervised classifier, and SSCM: self-supervised classifier using multi-labeled data.