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

arXiv cs.AI / 4/17/2026

📰 NewsTools & Practical UsageModels & Research

共有:

Key Points

RadAgent is a tool-using AI agent designed to generate chest CT reports through stepwise, interpretable reasoning rather than only producing final outputs.
The system provides clinicians with an inspectable decision and tool-interaction trace so they can verify and refine how each finding was derived.
Experiments on report generation show RadAgent outperforms the 3D vision-language model baseline (CT-Chat), improving macro-F1 by 6.0 points and micro-F1 by 5.4 points.
RadAgent also shows stronger robustness to adversarial conditions (up 24.7 points) and adds a faithfulness metric of 37.0%, which the baseline lacks.
Overall, the approach aims to make AI-assisted radiology more transparent and reliable by explicitly structuring tool-augmented iterative interpretation.

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.