Quantifying Explainers of Graph Neural Networks in Computational Pathology

TL;DR

This paper introduces a set of quantitative metrics to evaluate graph explainers in deep learning models for digital pathology, making explanations more interpretable for clinicians, especially in breast cancer subtyping.

Contribution

It proposes novel metrics based on class separability for assessing graph explainers, and applies them to evaluate explainers in cell-graph models for breast cancer subtyping.

Findings

Metrics effectively differentiate explainers' performance.

Graph explainers provide insights aligned with pathological concepts.

Evaluation framework is adaptable to other domains.

Abstract

Explainability of deep learning methods is imperative to facilitate their clinical adoption in digital pathology. However, popular deep learning methods and explainability techniques (explainers) based on pixel-wise processing disregard biological entities' notion, thus complicating comprehension by pathologists. In this work, we address this by adopting biological entity-based graph processing and graph explainers enabling explanations accessible to pathologists. In this context, a major challenge becomes to discern meaningful explainers, particularly in a standardized and quantifiable fashion. To this end, we propose herein a set of novel quantitative metrics based on statistics of class separability using pathologically measurable concepts to characterize graph explainers. We employ the proposed metrics to evaluate three types of graph explainers, namely the layer-wise relevance propagation, gradient-based saliency, and graph pruning approaches, to explain Cell-Graph representations for Breast Cancer Subtyping. The proposed metrics are also applicable in other domains by using domain-specific intuitive concepts. We validate the qualitative and quantitative findings on the BRACS dataset, a large cohort of breast cancer RoIs, by expert pathologists.