RadAgents: Multimodal Agentic Reasoning for Chest X-ray Interpretation with Radiologist-like Workflows

2025-09-24

AI Summary

Current chest X-ray (CXR) interpretation methods suffer from three key limitations: weak clinical interpretability, insufficient multimodal (vision–text) evidence fusion, and inconsistent tool outputs lacking dynamic validation mechanisms. To address these, we propose a clinical-prior-guided multi-agent framework that emulates radiologists’ diagnostic workflow, enabling vision-grounded collaborative reasoning. Specialized agents perform image analysis, report generation, consistency verification, and external knowledge retrieval; cross-modal evidence integration is achieved via multimodal fusion and vision–language alignment. We further introduce a dynamic conflict-resolution mechanism and retrieval-augmented contextual verification to enhance robustness and reliability. Experiments demonstrate significant improvements in diagnostic accuracy (+4.2%) and report consistency (+18.7%), yielding structured reports that are more transparent, clinically aligned, and compliant with established guidelines.

Technology Category

• Computer Vision: Multi-modal Vision
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Knowledge Representation and Reasoning: Diagnosis and Abductive Reasoning

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Agentic systems offer a potential path to solve complex clinical tasks through collaboration among specialized agents, augmented by tool use and external knowledge bases. Nevertheless, for chest X-ray (CXR) interpretation, prevailing methods remain limited: (i) reasoning is frequently neither clinically interpretable nor aligned with guidelines, reflecting mere aggregation of tool outputs; (ii) multimodal evidence is insufficiently fused, yielding text-only rationales that are not visually grounded; and (iii) systems rarely detect or resolve cross-tool inconsistencies and provide no principled verification mechanisms. To bridge the above gaps, we present RadAgents, a multi-agent framework for CXR interpretation that couples clinical priors with task-aware multimodal reasoning. In addition, we integrate grounding and multimodal retrieval-augmentation to verify and resolve context conflicts, resulting in outputs that are more reliable, transparent, and consistent with clinical practice.

Problem

• Improve clinical interpretability and guideline alignment in CXR interpretation
• Address insufficient multimodal evidence fusion in radiology workflows
• Detect and resolve cross-tool inconsistencies with verification mechanisms

Innovation

• Multi-agent framework with clinical priors
• Multimodal reasoning with grounding retrieval
• Conflict resolution via verification mechanisms