Additive MIL: Intrinsically Interpretable Multiple Instance Learning for Pathology
This work addresses the need for interpretable AI in clinical pathology to build trust and debug failures, though it is incremental as it builds on existing MIL methods.
The authors tackled the problem of interpretability in Multiple Instance Learning (MIL) models for pathology by proposing Additive MIL, which enables exact computation and visualization of region contributions, maintaining similar predictive performance and aligning with pathologist-used regions.
Multiple Instance Learning (MIL) has been widely applied in pathology towards solving critical problems such as automating cancer diagnosis and grading, predicting patient prognosis, and therapy response. Deploying these models in a clinical setting requires careful inspection of these black boxes during development and deployment to identify failures and maintain physician trust. In this work, we propose a simple formulation of MIL models, which enables interpretability while maintaining similar predictive performance. Our Additive MIL models enable spatial credit assignment such that the contribution of each region in the image can be exactly computed and visualized. We show that our spatial credit assignment coincides with regions used by pathologists during diagnosis and improves upon classical attention heatmaps from attention MIL models. We show that any existing MIL model can be made additive with a simple change in function composition. We also show how these models can debug model failures, identify spurious features, and highlight class-wise regions of interest, enabling their use in high-stakes environments such as clinical decision-making.