Counterfactual Intervention in Attention Multiple Instance Learning For Digital Pathology

Imane Chraki, Pierre Marza, Stergios Christodoulidis, Maria Vakalopoulou
Proceedings of The 9th International Conference on Medical Imaging with Deep Learning, PMLR 315:3336-3354, 2026.

Abstract

Attention-based Multiple Instance Learning (MIL) has become a prominent framework for analysing whole-slide images (WSI). These models have been shown to achieve good performance on classification tasks, while also offering an inherent proxy for interpretability through attention weights. In this work, we first question the validity of using attention for the interpretability of MIL models. Subsequently, we propose Counterfactual Intervention in Attention for MIL (), a causal extension of attention-based MIL that explicitly measures and optimizes the contribution of attention to slide-level predictions. Across four histopathology classification benchmarks (BRCA, NSCLC, LUAD, Camelyon16) and two feature encoders (Resnet50, UNI), we investigate how the interpretability of attention relates to the representation space, and the downstream performance. We then show that achieves performance comparable to strong MIL baselines while providing a more causally meaningful attention vector for explaining the model’s outcome. Qualitative perturbation experiments show that dropping the top-attended patches leads to a larger confidence degradation in compared to baseline ABMIL, highlighting the potential of causal supervision for reliable and interpretable WSI-based prediction.

Cite this Paper

BibTeX
@InProceedings{pmlr-v315-chraki26a, title = {Counterfactual Intervention in Attention Multiple Instance Learning For Digital Pathology}, author = {Chraki, Imane and Marza, Pierre and Christodoulidis, Stergios and Vakalopoulou, Maria}, booktitle = {Proceedings of The 9th International Conference on Medical Imaging with Deep Learning}, pages = {3336--3354}, year = {2026}, editor = {Huo, Yuankai and Gao, Mingchen and Kuo, Chang-Fu and Jin, Yueming and Deng, Ruining}, volume = {315}, series = {Proceedings of Machine Learning Research}, month = {08--10 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v315/main/assets/chraki26a/chraki26a.pdf}, url = {https://proceedings.mlr.press/v315/chraki26a.html}, abstract = {Attention-based Multiple Instance Learning (MIL) has become a prominent framework for analysing whole-slide images (WSI). These models have been shown to achieve good performance on classification tasks, while also offering an inherent proxy for interpretability through attention weights. In this work, we first question the validity of using attention for the interpretability of MIL models. Subsequently, we propose Counterfactual Intervention in Attention for MIL (), a causal extension of attention-based MIL that explicitly measures and optimizes the contribution of attention to slide-level predictions. Across four histopathology classification benchmarks (BRCA, NSCLC, LUAD, Camelyon16) and two feature encoders (Resnet50, UNI), we investigate how the interpretability of attention relates to the representation space, and the downstream performance. We then show that achieves performance comparable to strong MIL baselines while providing a more causally meaningful attention vector for explaining the model’s outcome. Qualitative perturbation experiments show that dropping the top-attended patches leads to a larger confidence degradation in compared to baseline ABMIL, highlighting the potential of causal supervision for reliable and interpretable WSI-based prediction.} }
Endnote
%0 Conference Paper %T Counterfactual Intervention in Attention Multiple Instance Learning For Digital Pathology %A Imane Chraki %A Pierre Marza %A Stergios Christodoulidis %A Maria Vakalopoulou %B Proceedings of The 9th International Conference on Medical Imaging with Deep Learning %C Proceedings of Machine Learning Research %D 2026 %E Yuankai Huo %E Mingchen Gao %E Chang-Fu Kuo %E Yueming Jin %E Ruining Deng %F pmlr-v315-chraki26a %I PMLR %P 3336--3354 %U https://proceedings.mlr.press/v315/chraki26a.html %V 315 %X Attention-based Multiple Instance Learning (MIL) has become a prominent framework for analysing whole-slide images (WSI). These models have been shown to achieve good performance on classification tasks, while also offering an inherent proxy for interpretability through attention weights. In this work, we first question the validity of using attention for the interpretability of MIL models. Subsequently, we propose Counterfactual Intervention in Attention for MIL (), a causal extension of attention-based MIL that explicitly measures and optimizes the contribution of attention to slide-level predictions. Across four histopathology classification benchmarks (BRCA, NSCLC, LUAD, Camelyon16) and two feature encoders (Resnet50, UNI), we investigate how the interpretability of attention relates to the representation space, and the downstream performance. We then show that achieves performance comparable to strong MIL baselines while providing a more causally meaningful attention vector for explaining the model’s outcome. Qualitative perturbation experiments show that dropping the top-attended patches leads to a larger confidence degradation in compared to baseline ABMIL, highlighting the potential of causal supervision for reliable and interpretable WSI-based prediction.
APA
Chraki, I., Marza, P., Christodoulidis, S. & Vakalopoulou, M.. (2026). Counterfactual Intervention in Attention Multiple Instance Learning For Digital Pathology. Proceedings of The 9th International Conference on Medical Imaging with Deep Learning, in Proceedings of Machine Learning Research 315:3336-3354 Available from https://proceedings.mlr.press/v315/chraki26a.html.

Related Material