Agentic Systems Leverage FDA-Cleared Radiology Tools for Comprehensive Reporting
Original title: Can Agentic Systems Close the Gap in Radiology with Production AI Tools?
Agentic framework recovers under a third of actionable findings with cleared tools, improving substantially with unlimited catalogs.
The paper describes a radiology agentic framework that orchestrates modular tools to generate structured reports. It tests the system using tools from FDA-cleared radiology AI devices on a multi-modality report corpus. With a consolidated catalog of findings, the orchestrator recovers under a third of clinically actionable findings. Unlimited catalog and target-report-based planning improve recovery. Most missed findings are classes no cleared device offers; a third result from tools offered but not called.
- Orchestrator recovers under a third of findings with cleared catalog
- Unlimited catalog substantially improves recovery
- Most misses are unsupported finding classes
- Tool availability and orchestration are complementary constraints
From the original article
1 min read at arxiv.orgRadiology AI has developed along two paths: specialized models that perform well on narrow tasks but cover few clinical findings, and generalist models that address a broader range of tasks but remain prone to hallucination and inaccurate measurements. Combining the two through agentic systems offers a route towards comprehensive radiology reporting, but published systems are tied to one modality or body part.
By Aubin Charley · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.