Extended Data Table 2 Automated report generation metrics

From: Collaboration between clinicians and vision–language models in radiology report generation

  1. (a) Comparison of automatic report generation metrics on the MIMIC-CXR dataset. The column ‘Sections’ indicates which sections of the radiology reports are generated by the respective models; ‘F’ indicates FINDINGS and `I` indicates IMPRESSIONS sections. Note that the metrics are retrieved from the corresponding publications. For all metrics, the higher (the bluer) the better, and the best results are shown in bold. (b) Automated report generation metrics on the IND1 dataset. We note that there are no published report generation metrics due to the private nature of the dataset. The disease classification accuracy (F1 scores) are also computed for two radiologists.
Back to article page