Graph-Augmented Topological Internalization with Dual-Stream Classifiers for Medical Report Generation

2026-05-04

AI Summary

This work addresses the limitations of existing medical report generation methods, which treat thoracic abnormalities as isolated categories and neglect disease co-occurrence patterns and anatomical topology, thereby struggling to accurately reason about complex or subtle pathologies. To overcome this, the authors propose a dual-stream classification framework: the main stream incorporates disease topological priors via graph convolutional networks to generate structurally constrained diagnostic prompts, while the auxiliary stream employs an asymmetric optimization strategy to dynamically adjust decision boundaries under class imbalance and introduces a diagnosis-guided spatial attention mechanism to align visual features with diagnostic semantics. Without relying on external knowledge retrieval, the method explicitly models inter-disease topological relationships, achieving strong clinical relevance and linguistic fluency on MIMIC-CXR and demonstrating robust zero-shot generalization on the IU X-Ray dataset.

Abstract

Automated medical report generation, MRG, holds substantial value for alleviating radiologist workload and enhancing diagnostic efficiency. However, mainstream approaches typically treat diverse chest abnormalities as isolated classification targets. This paradigm often overlooks inherent disease co-occurrences and struggles to translate medical topological structures into explicit data correlations, constraining the model's reasoning capacity on complex or subtle lesions. To address this, we propose a Graph-Augmented Dual-Stream Medical Report Generation with Topological Internalization, GDMRG. Our framework introduces a Topological Knowledge Internalization module, TKI, which leverages a Graph Convolutional Network, GCN, to generate an explicit parameterized weight matrix based on global disease co-occurrence priors. This facilitates efficient topological knowledge injection without relying on external retrieval mechanisms. Building upon this, we construct a dual-stream classification system: the main branch generates discrete diagnostic prompts under topological constraints, while the auxiliary branch employs an asymmetric optimization strategy to dynamically calibrate decision boundaries for highly imbalanced samples. Concurrently, to establish a logical closed loop between diagnosis and visual grounding, we design a diagnostic-driven Diagnosis-Guided Spatial Attention, DGSA, that utilizes high-dimensional clinical semantics to recalibrate the visual encoder, mitigating feature hallucinations. Comprehensive experiments on the MIMIC-CXR dataset demonstrate that GDMRG achieves competitive clinical efficacy, CE, while maintaining natural language fluency. Furthermore, our model exhibits robust zero-shot generalization on the IU X-Ray dataset. In summary, this work presents an integrated and interpretable paradigm for medical report generation.

Problem

• medical report generation
• disease co-occurrence
• topological structure
• chest abnormalities
• diagnostic reasoning

Innovation

• Graph Convolutional Network
• Topological Knowledge Internalization
• Dual-Stream Classifier
• Diagnosis-Guided Spatial Attention
• Medical Report Generation