Predicting ulcer in H&E images of inflammatory bowel disease using domain-knowledge-driven graph neural network

2025-04-13

AI Summary

Weakly supervised localization of ulcer regions in hematoxylin-and-eosin (H&E) whole-slide images of inflammatory bowel disease (IBD) remains challenging due to the absence of pixel-level annotations and the neglect of spatial contextual cues in conventional multiple-instance learning (MIL) frameworks. Method: We propose DomainGCN, a domain-knowledge-guided graph convolutional network that explicitly encodes pathological priors—epithelial integrity, lymphocytic infiltration, and tissue fragmentation—as an interpretable graph structure, where nodes represent tissue regions and edges encode topological and semantic relationships; graph convolution then performs ulcer discrimination by jointly reasoning over histological semantics and spatial topology. Contribution/Results: DomainGCN achieves biologically interpretable and robust predictions by unifying domain knowledge with geometric deep learning. On a clinical IBD dataset, it outperforms state-of-the-art weakly supervised methods, improving ulcer detection AUC by over 5%, thereby validating the efficacy of the “pathology-knowledge graph construction + graph neural reasoning” paradigm.

Technology Category

• Machine Learning: Graph-based Machine Learning
• Knowledge Representation and Reasoning: Diagnosis and Abductive Reasoning
• Data Mining & Knowledge Management: Graph Mining, Social Network Analysis & Community

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Semantics and Knowledge: Scalable techniques for the creation, curation, publication, maintenance, and consumption of large, Web-based, structured, reusable, knowledge graphs and ontologies
• Search and Retrieval-Augmented AI: Vertical and domain-specific search

Abstract

Inflammatory bowel disease (IBD) involves chronic inflammation of the digestive tract, with treatment options often burdened by adverse effects. Identifying biomarkers for personalized treatment is crucial. While immune cells play a key role in IBD, accurately identifying ulcer regions in whole slide images (WSIs) is essential for characterizing these cells and exploring potential therapeutics. Multiple instance learning (MIL) approaches have advanced WSI analysis but they lack spatial context awareness. In this work, we propose a weakly-supervised model called DomainGCN that employs a graph convolution neural network (GCN) and incorporates domain-specific knowledge of ulcer features, specifically, the presence of epithelium, lymphocytes, and debris for WSI-level ulcer prediction in IBD. We demonstrate that DomainGCN outperforms various state-of-the-art (SOTA) MIL methods and show the added value of domain knowledge.

Problem

• Predicting ulcers in IBD H&E images using GCN
• Incorporating domain knowledge for ulcer feature identification
• Improving WSI analysis with spatial context awareness

Innovation

• Domain-knowledge-driven graph neural network
• Incorporates ulcer features like epithelium
• Weakly-supervised model for WSI analysis