Skip to main content
. 2024 Dec 3;8:277. doi: 10.1038/s41698-024-00770-z

Fig. 3. Patch-level quantitative analysis on novel cohorts.

Fig. 3

a List of eight deep learning models trained on different cohort combinations. b–h Performance of the eight models in novel cohorts where none of the test immunostain types has ever been utilized during the training phase by any of the models. b Claudin 18.2, c DeLta-Like 3 (DLL3), d Fibroblast Growth Factor Receptor 2 (FGFR2), e Human Epidermal growth factor Receptor 3 (HER3), f Mesenchymal-Epithelial Transition factor (MET), g MUCin-16 (MUC16), h TROPhoblast cell-surface antigen 2 (TROP2). mF1, mean F1 score. (****p < 0.0001; ***p < 0.001; **p < 0.01; *p < 0.05; ns, not significant).