mlsanityData to 14 Oct 2026

Current Pathology Foundation Models are unrobust to Medical Center Differences

arXiv 2501.18055 ↗First posted 29 Jan 2025cs.LG cs.AIcited by 39

Not on the map. Why?

Abstract (arXiv)

Pathology Foundation Models (FMs) hold great promise for healthcare. Before they can be used in clinical practice, it is essential to ensure they are robust to variations between medical centers. We measure whether pathology FMs focus on biological features like tissue and cancer type, or on the well known confounding medical center signatures introduced by staining procedure and other differences. We introduce the Robustness Index. This novel robustness metric reflects to what degree biological features dominate confounding features. Ten current publicly available pathology FMs are evaluated. We find that all current pathology foundation models evaluated represent the medical center to a strong degree. Significant differences in the robustness index are observed. Only one model so far has a robustness index greater than one, meaning biological features dominate confounding features, but only slightly. A quantitative approach to measure the influence of medical center differences on FM-based prediction performance is described. We analyze the impact of unrobustness on classification performance of downstream models, and find that cancer-type classification errors are not random, but specifically attributable to same-center confounders: images of other classes from the same medical center. We visualize FM embedding spaces, and find these are more strongly organized by medical centers than by biological factors. As a consequence, the medical center of origin is predicted more accurately than the tissue source and cancer type. The robustness index introduced here is provided with the aim of advancing progress towards clinical adoption of robust and reliable pathology FMs.

Key facts. Cited by 39 papers in the map (data through 14 Oct 2026). First posted on arXiv 29 Jan 2025.

Builds on

0 of 0 references.

No references here yet.

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 20; 2026: 19.

39 papers.

Sort:

By area:

  1. 2025A Survey of Pathology Foundation Model: Progress and Future Directionscited by 16
  2. 2025Foundation Models in Medical Imaging: A Review and Outlookcited by 8
  3. 2025Quantized-Tinyllava: a new multimodal foundation model enables efficient split learningcited by 8
  4. 2025MeDi: Metadata-Guided Diffusion Models for Mitigating Biases in Tumor Classificationcited by 7
  5. 2025Pathology Foundation Models are Scanner Sensitive: Benchmark and Mitigation with Contrastive ScanGen Losscited by 7
  6. 2026Atlas 2 -- Foundation models for clinical deploymentcited by 7
  7. 2025HASD: Hierarchical Adaption for pathology Slide-level Domain-shiftcited by 6
  8. 2025Beyond the Failures: Rethinking Foundation Models in Pathologycited by 6
  9. 2025Benchmarking histopathology foundation models in a multi-center dataset for skin cancer subtypingcited by 3
  10. 2025ModalTune: Fine-Tuning Slide-Level Foundation Models with Multi-Modal Information for Multi-task Learning in Digital Pathologycited by 2
  11. 2025G2L:From Giga-Scale to Cancer-Specific Large-Scale Pathology Foundation Models via Knowledge Distillationcited by 2
  12. 2026Computational Pathology in the Era of Emerging Foundation and Agentic AI -- International Expert Perspectives on Clinical Integration and Translational Readinesscited by 2
27 more
  1. 2025MVHybrid: Improving Spatial Transcriptomics Prediction with Hybrid State Space-Vision Transformer Backbone in Pathology Vision Foundation Modelscited by 1
  2. 2025AdvDINO: Domain-Adversarial Self-Supervised Representation Learning for Spatial Proteomicscited by 1
  3. 2026A protocol for evaluating robustness to H&E staining variation in computational pathology modelscited by 1
  4. 2026Auditing Data Leakage in Whole-Slide Image Multimodal Benchmarkscited by 1
  5. 2025Open-source framework for detecting bias and overfitting for large pathology imagescited by 0
  6. 2025AdaFusion: Prompt-Guided Inference with Adaptive Fusion of Pathology Foundation Modelscited by 0
  7. 2025From Explainable to Explained AI: Ideas for Falsifying and Quantifying Explanationscited by 0
  8. 2025Comparing Computational Pathology Foundation Models using Representational Similarity Analysiscited by 0
  9. 2025Zero-shot segmentation of skin tumors in whole-slide images with vision-language foundation modelscited by 0
  10. 2025Simple Agents Outperform Experts in Biomedical Imaging Workflow Optimizationcited by 0
  11. 2025DA-SSL: self-supervised domain adaptor to leverage foundational models in turbt histopathology slidescited by 0
  12. 2025PANDA-PLUS-Bench: A Clinical Benchmark for Evaluating Robustness of AI Foundation Models in Prostate Cancer Diagnosiscited by 0
  13. 2026Do Pathology Foundation Models Encode Disease Progression? A Pseudotime Analysis of Visual Representationscited by 0
  14. 2026Histopathology Image Normalization via Latent Manifold Compactioncited by 0
  15. 2026When Are Multimodal Predictions Biologically Supported? A Diagnostic Evaluation Frameworkcited by 0
  16. 2026Vision-Language Models as Zero-Annotation Oracles in Histopathologycited by 0
  17. 2026Semantic-Anchored Evidential Fusion for Domain-Robust Whole-Slide Survival Analysiscited by 0
  18. 2026Mitigating Batch Effects in Histopathology via Language-Mediated Robust Embedding Generationcited by 0
  19. 2026The Good, the Bad, and the Brittle: Benchmarking Robustness and Generalisation of Histopathology Foundation Modelscited by 0
  20. 2026Medical foundation models converge less under label supervisioncited by 0
  21. 2026Robustifying pathology foundation models via fine-tuningcited by 0
  22. 2026A Distributional Robustness Margin For Pathology Foundation Modelscited by 0
  23. 2026An end-to-end-trained vision-language model for native-language prostate pathology report generationcited by 0
  24. 2026EXPOSE: Explainable and Domain-Robust Embeddings from Pathology Vision Foundation Models using Sparse Autoencoderscited by 0
  25. 2026Multimodal LLMs Outperform Pathology Foundation Models in Cross-Domain Histological Similaritycited by 0
  26. 2026HERO: Histology Encoder for Robust Representation in Oncologycited by 0
  27. 2026Dataset-Specific Encoding of Image Content Underlies Cross-Dataset Failure in Brain MRI Foundation Modelscited by 0