RadAgent: A tool-using AI agent for stepwise interpretation of chest computed tomography
Summary
RadAgent is a tool-using AI agent that generates chest CT reports through interpretable step-by-step reasoning, improving clinical accuracy by 36.4% relative and achieving 37% faithfulness—a capability absent in existing 3D vision-language models. The system provides fully inspectable reasoning traces allowing clinicians to validate and refine diagnostic outputs.
View Cached Full Text
Cached at: 04/20/26, 08:27 AM
Paper page - RadAgent: A tool-using AI agent for stepwise interpretation of chest computed tomography
Source: https://huggingface.co/papers/2604.15231
Abstract
RadAgent, a tool-using AI agent, enhances chest CT report generation through interpretable step-by-step reasoning traces that improve clinical accuracy, robustness, and faithfulness compared to existing 3D vision-language models.
Vision-language models (https://huggingface.co/papers?q=Vision-language%20models) (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 (https://huggingface.co/papers?q=reasoning%20trace) for them to inspect, validate, or refine. To address this, we introduce RadAgent, a tool-using AI agent (https://huggingface.co/papers?q=tool-using%20AI%20agent) that generates CT reports (https://huggingface.co/papers?q=CT%20reports) 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 (https://huggingface.co/papers?q=Clinical%20accuracy) improves by 6.0 points (36.4% relative) in macro-F1 and 5.4 points (19.6% relative) in micro-F1. Robustness (https://huggingface.co/papers?q=Robustness) under adversarial conditions improves by 24.7 points (41.9% relative). Furthermore, RadAgent achieves 37.0% in faithfulness (https://huggingface.co/papers?q=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 (https://huggingface.co/papers?q=reasoning%20trace), RadAgent brings us closer toward transparent and reliable AI for radiology.
View arXiv page (https://arxiv.org/abs/2604.15231) View PDF (https://arxiv.org/pdf/2604.15231) Project page (https://rad-agent.github.io/) Add to collection (https://huggingface.co/login?next=%2Fpapers%2F2604.15231)
Get this paper in your agent:
hf papers read 2604.15231
Don’t have the latest CLI? curl -LsSf https://hf.co/cli/install.sh | bash
Models citing this paper1
RadAgent/radagent-qwen3-14b-lora Text Generation • Updated 3 days ago • 66 • 3 (https://huggingface.co/RadAgent/radagent-qwen3-14b-lora)
Datasets citing this paper0
No dataset linking this paper
Cite arxiv.org/abs/2604.15231 in a dataset README.md to link it from this page.
Spaces citing this paper0
No Space linking this paper
Cite arxiv.org/abs/2604.15231 in a Space README.md to link it from this page.
Collections including this paper0
No Collection including this paper
Add this paper to a collection (https://huggingface.co/new-collection) to link it from this page.
Similar Articles
A specialized reasoning large language model for accelerating rare disease diagnosis: a randomized AI physician assistance trial
This paper presents RaDaR, a 32B open-source reasoning LLM trained on public and synthetic rare disease cases, which outperforms larger models like DeepSeek-R1 in diagnosis benchmarks and improves physician accuracy by 21.44 percentage points in a randomized trial.
AgentX - AI Agent evaluation framework
AgentX is an AI agent evaluation framework that helps pinpoint issues and fix them with one click.
AgentRail
AgentRail is a local control plane for AI coding agents, providing management and orchestration for AI-assisted coding workflows.
Skill-Augmented AI Agents for Medical Research Analysis: An Exploratory Multi-Model Human Evaluation in an NSCLC Transcriptomic Biomarker Task
This exploratory study evaluates whether augmenting AI agents with a medical research skill package improves the quality of transcriptomic research analysis outputs compared to native AI, using a multi-model human evaluation in an NSCLC biomarker task. Results show a directional but statistically non-significant improvement, highlighting the need for larger, more robust evaluations.
Configurable Clinical Information Extraction with Agentic RAG: What Works, What Breaks, and Why
ACIE, an agentic RAG system for clinical information extraction, achieves 96.5% acceptance rate in nuclear-medicine physicians' judgments across 7,326 instances, addressing challenges of heterogeneous patient contexts and missing metadata.