Ontology-Based Concept Distillation for Radiology Report Retrieval and Labeling

2025-08-27

AI Summary

Existing radiology report retrieval methods rely on high-dimensional text embeddings, suffering from poor interpretability, high computational overhead, and misalignment with medical knowledge structures. To address these limitations, this paper proposes a UMLS ontology–based concept distillation framework. First, it jointly leverages RadGraph-XL and SapBERT to extract and normalize clinical entities into standardized UMLS concepts. Second, it introduces a task-adaptive, enhanced weighted Tversky index that explicitly models synonymy, negation, and hierarchical semantic relations, enabling interpretable, semantics-aware similarity computation. Experiments on MIMIC-CXR demonstrate that our method significantly outperforms mainstream embedding-based retrieval models—particularly improving recall for long-tail disease categories—and generates high-quality, ontology-aligned disease annotations.

Technology Category

• Knowledge Representation and Reasoning: Diagnosis and Abductive Reasoning
• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval
• Computer Vision: Image and Video Retrieval

Application Category

• 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: Web query analysis, representation and understanding
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Retrieval-augmented learning based on radiology reports has emerged as a promising direction to improve performance on long-tail medical imaging tasks, such as rare disease detection in chest X-rays. Most existing methods rely on comparing high-dimensional text embeddings from models like CLIP or CXR-BERT, which are often difficult to interpret, computationally expensive, and not well-aligned with the structured nature of medical knowledge. We propose a novel, ontology-driven alternative for comparing radiology report texts based on clinically grounded concepts from the Unified Medical Language System (UMLS). Our method extracts standardised medical entities from free-text reports using an enhanced pipeline built on RadGraph-XL and SapBERT. These entities are linked to UMLS concepts (CUIs), enabling a transparent, interpretable set-based representation of each report. We then define a task-adaptive similarity measure based on a modified and weighted version of the Tversky Index that accounts for synonymy, negation, and hierarchical relationships between medical entities. This allows efficient and semantically meaningful similarity comparisons between reports. We demonstrate that our approach outperforms state-of-the-art embedding-based retrieval methods in a radiograph classification task on MIMIC-CXR, particularly in long-tail settings. Additionally, we use our pipeline to generate ontology-backed disease labels for MIMIC-CXR, offering a valuable new resource for downstream learning tasks. Our work provides more explainable, reliable, and task-specific retrieval strategies in clinical AI systems, especially when interpretability and domain knowledge integration are essential. Our code is available at https://github.com/Felix-012/ontology-concept-distillation

Problem

• Improves rare disease detection in chest X-rays
• Replaces high-dimensional embeddings with ontology-driven comparisons
• Enhances interpretability and efficiency in radiology report retrieval

Innovation

• Ontology-driven UMLS concept comparison for reports
• Enhanced entity extraction using RadGraph-XL and SapBERT
• Modified Tversky Index for semantic similarity measurement