迷失在翻译中：语言如何重新调整跨物种病理学的视野

Foundation models are increasingly utilized in computational pathology, yet their performance under cross-cancer and cross-species transfer remains unclear. This study investigates the effects of fine-tuning CPath-CLIP on cancer detection across same-cancer, cross-cancer, and cross-species conditions using whole-slide image patches from canine and human histopathology. Performance was evaluated using the area under the receiver operating characteristic curve (AUC). Few-shot fine-tuning enhanced same-cancer performance from 64.9% to 72.6% AUC and cross-cancer performance from 56.84% to 66.31% AUC. Cross-species evaluation indicated that while tissue matching allows for meaningful transfer, performance still falls short of state-of-the-art benchmarks (H-optimus-0: 84.97% AUC).