通过微调增强病理学基础模型的鲁棒性
Robustifying pathology foundation models via fine-tuning
浏览论文内容
中文总结 AI 辅助
研究针对病理学基础模型对采集因素敏感问题，开发新颖微调方法，应用于十个不同模型提升鲁棒性与下游性能，平均鲁棒性指数提高23%，综合性能提高43%，并公开两个模型的微调版本。
中文摘要 AI 辅助
病理学基础模型（FMs）能生成强大的切片级表示，但对扫描仪和染色变异性敏感，影响跨实验室部署。我们开发了一种新颖的微调方法，可提高病理学FMs对采集因素的鲁棒性。应用于十个不同的FMs，该微调策略持续提升每个模型的鲁棒性及下游性能，无权衡现象。平均而言，PathoROB鲁棒性指数提高23%，在Patho - Bench、HEST和THUNDER上综合性能提高43%，个别模型鲁棒性增益达72%，性能增益达76%。我们还公开了Phikon - v2（Phaet）和Midnight - 12k（Mascaret）的微调版本。
英文摘要
Pathology foundation models (FMs) produce powerful tile-level representations which remain sensitive to scanner and staining variability, undermining deployment across laboratories. We develop a novel fine-tuning recipe that improves the robustness of pathology FMs to acquisition factors. Applied to ten different FMs, our fine-tuning strategy consistently improves robustness for every model as well as downstream performance, with no observed trade-off. On average, it raises the PathoROB robustness index by 23% (from 0.72 to 0.87) and increases the overall cross-benchmark performance by 43% on Patho-Bench, HEST and THUNDER combined, with individual gains reaching up to 72% in robustness (Phikon-v2) and 76% in performance (Midnight-12k). We publicly release the fine-tuned versions of Phikon-v2 (Phaet) and Midnight-12k (Mascaret) at https://huggingface.co/wearewaiv/models.
发表机构
- Waiv
- Institute of Pathology, Technical University of Munich（慕尼黑工业大学病理学研究所）
- School of Computation, Information and Technology, Technical University of Munich（慕尼黑工业大学计算、信息与技术学院）
机构由 AI 辅助整理，请以论文原文为准。
