Fig. 3
From: Comprehensive Benchmark Dataset for Pathological Lymph Node Metastasis in Breast Cancer Sections
Performance comparison between Camelyon-17-Origin23 and Camelyon-17-Refine across different MIL models and feature encoders. (a–c) AUC comparison using three feature encoders: (a) PLIP3, (b) UNI4, and (c) Gigapath5. (d–f) F1-score comparison using three feature encoders: (d) PLIP, (e) UNI, and (f) Gigapath.
