xMIL: Insightful Explanations for Multiple Instance Learning in Histopathology

Citation: J. Hense*, M. Jamshidi Idaji*, O. Eberle, T. Schnake, J. Dippel, L. Ciernik, O. Buchstab, A. Mock, F. Klauschen, K.-R. Müller, 2024, “xMIL: Insightful Explanations for Multiple Instance Learning in Histopathology”, In Advances in Neural Information Processing Systems, vol. 37, pp. 8300–8328. DOI: 10.52202/079017-0266, *=equal contribution

• The paper is accessible through NeurIPS 2024.

• You can download all the codes via Github.

Summary: In this study, we revisit MIL through the lens of explainable AI (XAI) and introduce xMIL, a refined framework with more general assumptions. We demonstrate how to obtain improved MIL explanations using layer-wise relevance propagation (LRP) and conduct extensive evaluation experiments on three toy settings and four real-world histopathology datasets.