Quantifying Explainers of Graph Neural Networks in Computational Pathology
This work addresses the problem of quantitatively evaluating GNN explainers in computational pathology, which is crucial for facilitating the clinical adoption of deep learning methods for pathologists.
This paper tackles the challenge of evaluating explainers for Graph Neural Networks (GNNs) in computational pathology, which traditionally lack biological context. They propose a set of novel quantitative metrics based on class separability statistics using pathologically measurable concepts to characterize graph explainers. The metrics were used to evaluate three types of graph explainers on Cell-Graph representations for Breast Cancer Subtyping using the BRACS dataset.
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.