RadAgent: A tool-using AI agent for stepwise interpretation of chest computed tomography

2026-04-16

AI Summary

This work addresses the lack of interpretable reasoning in current AI systems for chest CT interpretation, which hinders clinicians’ ability to scrutinize and correct intermediate diagnostic judgments. The authors propose a tool-augmented AI agent that, for the first time, implements an explicit, traceable iterative reasoning process by structuring radiological diagnostic steps and introducing a faithfulness evaluation metric. Leveraging a vision-language model integrated with tool-calling mechanisms, the proposed method substantially outperforms the baseline CT-Chat, achieving a 6.0-point gain in macro F1 (+36.4%), a 5.4-point improvement in micro F1 (+19.6%), and a 24.7-point increase in adversarial robustness (+41.9%). Additionally, it attains a faithfulness score of 37.0%, significantly enhancing the accuracy, robustness, and transparency of generated radiology reports.

Technology Category

• Computer Vision: Interpretability, Explainability, and Transparency
• Philosophy and Ethics of AI: Safety, Robustness & Trustworthiness
• Natural Language Processing: Safety and Robustness

Application Category

• Search and Retrieval-Augmented AI: Agentic search
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Vision-language models (VLM) have markedly advanced AI-driven interpretation and reporting of complex medical imaging, such as computed tomography (CT). Yet, existing methods largely relegate clinicians to passive observers of final outputs, offering no interpretable reasoning trace for them to inspect, validate, or refine. To address this, we introduce RadAgent, a tool-using AI agent that generates CT reports through a stepwise and interpretable process. Each resulting report is accompanied by a fully inspectable trace of intermediate decisions and tool interactions, allowing clinicians to examine how the reported findings are derived. In our experiments, we observe that RadAgent improves Chest CT report generation over its 3D VLM counterpart, CT-Chat, across three dimensions. Clinical accuracy improves by 6.0 points (36.4% relative) in macro-F1 and 5.4 points (19.6% relative) in micro-F1. Robustness under adversarial conditions improves by 24.7 points (41.9% relative). Furthermore, RadAgent achieves 37.0% in faithfulness, a new capability entirely absent in its 3D VLM counterpart. By structuring the interpretation of chest CT as an explicit, tool-augmented and iterative reasoning trace, RadAgent brings us closer toward transparent and reliable AI for radiology.

Problem

• medical imaging interpretation
• interpretability
• AI transparency
• radiology reporting
• clinician-AI collaboration

Innovation

• tool-using agent
• interpretable reasoning
• chest CT interpretation
• vision-language model
• faithfulness