Integrating Clinical Knowledge into Concept Bottleneck Models

TL;DR

This paper proposes integrating clinical knowledge into concept bottleneck models to improve their interpretability and robustness, especially on out-of-domain medical images, by aligning model concepts with clinician priorities.

Contribution

The paper introduces a method to incorporate clinical knowledge into CBMs, enhancing their generalization and interpretability in medical imaging tasks.

Findings

Improved classification accuracy on unseen datasets.

Enhanced alignment with clinical decision-making.

Increased robustness to domain shifts.

Abstract

Concept bottleneck models (CBMs), which predict human-interpretable concepts (e.g., nucleus shapes in cell images) before predicting the final output (e.g., cell type), provide insights into the decision-making processes of the model. However, training CBMs solely in a data-driven manner can introduce undesirable biases, which may compromise prediction performance, especially when the trained models are evaluated on out-of-domain images (e.g., those acquired using different devices). To mitigate this challenge, we propose integrating clinical knowledge to refine CBMs, better aligning them with clinicians' decision-making processes. Specifically, we guide the model to prioritize the concepts that clinicians also prioritize. We validate our approach on two datasets of medical images: white blood cell and skin images. Empirical validation demonstrates that incorporating medical guidance enhances the model's classification performance on unseen datasets with varying preparation methods, thereby increasing its real-world applicability.