BioX-CPath: Biologically-driven Explainable Diagnostics for Multistain IHC Computational Pathology

2025-03-26

AI Summary

To address the limited biological interpretability of deep learning models in multi-stain immunohistochemistry (IHC) whole-slide image classification, this paper proposes a Stain-Aware Graph Neural Network (SA-GNN) that jointly encodes spatial topology and multi-stain semantic features to generate pathologically meaningful patient embeddings. Its core innovation is the Stain-Aware Attention Pooling (SAAP) module, which quantitatively aligns model decisions with underlying pathological mechanisms via stain-specific attention scores, entropy regularization, and inter-stain interaction scoring—thereby co-optimizing diagnostic accuracy and biological interpretability. Evaluated on a multi-stain dataset comprising rheumatoid arthritis and Sjögren’s syndrome cases, SA-GNN achieves state-of-the-art classification performance while delivering clinically verifiable, biologically grounded explanations. The implementation code and documentation are publicly released.

Technology Category

• Computer Vision: Medical and Biological Imaging
• Machine Learning: Multi-class/Multi-label Learning & Extreme Classification
• Knowledge Representation and Reasoning: Diagnosis and Abductive Reasoning

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Search and Retrieval-Augmented AI: Web query analysis, representation and understanding
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

The development of biologically interpretable and explainable models remains a key challenge in computational pathology, particularly for multistain immunohistochemistry (IHC) analysis. We present BioX-CPath, an explainable graph neural network architecture for whole slide image (WSI) classification that leverages both spatial and semantic features across multiple stains. At its core, BioX-CPath introduces a novel Stain-Aware Attention Pooling (SAAP) module that generates biologically meaningful, stain-aware patient embeddings. Our approach achieves state-of-the-art performance on both Rheumatoid Arthritis and Sjogren's Disease multistain datasets. Beyond performance metrics, BioX-CPath provides interpretable insights through stain attention scores, entropy measures, and stain interaction scores, that permit measuring model alignment with known pathological mechanisms. This biological grounding, combined with strong classification performance, makes BioX-CPath particularly suitable for clinical applications where interpretability is key. Source code and documentation can be found at: https://github.com/AmayaGS/BioX-CPath.

Problem

• Develop interpretable models for multistain IHC pathology analysis
• Classify whole slide images using spatial and semantic features
• Provide biologically meaningful insights for clinical diagnostics

Innovation

• Graph neural network for multistain pathology analysis
• Stain-Aware Attention Pooling for patient embeddings
• Interpretable insights via stain attention and interaction