Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Radiologist Copilot: An Agentic Assistant with Orchestrated Tools for Radiology Reporting with Quality Control

Yongrui Yu, Zhongzhen Huang, Linjie Mu, Shaoting Zhang, Xiaofan Zhang

TL;DR

The paper tackles the challenge of automated radiology reporting for 3D medical imaging by integrating a quality control loop into the reporting pipeline. It introduces Radiologist Copilot, an agentic AI system that autonomously analyzes images, generates reports, and enforces quality standards through a coordinated set of tools and an LLM-based reasoning backbone. Empirical results on liver radiology reporting demonstrate substantial improvements over state-of-the-art methods in both natural language generation and clinical accuracy, with comprehensive ablations validating the importance of region analysis planning, strategic template selection, and quality control. The approach is training-free, modular, and adaptable to different anatomical regions and VLM backbones, offering a practical path toward workflow-efficient, reliable radiology reporting.

Abstract

Radiology reporting is an essential yet time-consuming and error-prone task for radiologists in clinical examinations, especially for volumetric medical images. Rigorous quality control is also critical but tedious, ensuring that the final report meets clinical standards. Existing automated approaches, including radiology report generation methods and medical vision-language models, focus mainly on the report generation phase and neglect the crucial quality control procedure, limiting their capability to provide comprehensive support to radiologists. We propose Radiologist Copilot, an agentic AI assistant equipped with orchestrated tools designed for automated radiology reporting with quality control. Leveraging large language models as the reasoning backbone, the agentic system autonomously selects tools, plans, and executes actions, emulating the behavior of radiologists throughout the holistic radiology reporting process. The orchestrated tools include region localization, think with image paradigm directed region analysis planning, strategic template selection for report generation, quality assessment and feedback-driven adaptive refinement for quality control. Therefore, Radiologist Copilot facilitates accurate, complete, and efficient radiology reporting, assisting radiologists and improving clinical efficiency. Experimental results demonstrate that Radiologist Copilot significantly surpasses other state-of-the-art methods in radiology reporting. The source code will be released upon acceptance.

Radiologist Copilot: An Agentic Assistant with Orchestrated Tools for Radiology Reporting with Quality Control

TL;DR

The paper tackles the challenge of automated radiology reporting for 3D medical imaging by integrating a quality control loop into the reporting pipeline. It introduces Radiologist Copilot, an agentic AI system that autonomously analyzes images, generates reports, and enforces quality standards through a coordinated set of tools and an LLM-based reasoning backbone. Empirical results on liver radiology reporting demonstrate substantial improvements over state-of-the-art methods in both natural language generation and clinical accuracy, with comprehensive ablations validating the importance of region analysis planning, strategic template selection, and quality control. The approach is training-free, modular, and adaptable to different anatomical regions and VLM backbones, offering a practical path toward workflow-efficient, reliable radiology reporting.

Abstract

Radiology reporting is an essential yet time-consuming and error-prone task for radiologists in clinical examinations, especially for volumetric medical images. Rigorous quality control is also critical but tedious, ensuring that the final report meets clinical standards. Existing automated approaches, including radiology report generation methods and medical vision-language models, focus mainly on the report generation phase and neglect the crucial quality control procedure, limiting their capability to provide comprehensive support to radiologists. We propose Radiologist Copilot, an agentic AI assistant equipped with orchestrated tools designed for automated radiology reporting with quality control. Leveraging large language models as the reasoning backbone, the agentic system autonomously selects tools, plans, and executes actions, emulating the behavior of radiologists throughout the holistic radiology reporting process. The orchestrated tools include region localization, think with image paradigm directed region analysis planning, strategic template selection for report generation, quality assessment and feedback-driven adaptive refinement for quality control. Therefore, Radiologist Copilot facilitates accurate, complete, and efficient radiology reporting, assisting radiologists and improving clinical efficiency. Experimental results demonstrate that Radiologist Copilot significantly surpasses other state-of-the-art methods in radiology reporting. The source code will be released upon acceptance.

Paper Structure

This paper contains 22 sections, 5 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Methodology
  4. Radiologist Copilot
  5. Agentic Assistant
  6. Action Planner.
  7. Action Executor.
  8. Orchestrated Tools
  9. Segmentator.
  10. Analyzer.
  11. Report Generator.
  12. Quality Controller.
  13. Experiments and Results
  14. Experimental Setup
  15. Datasets.
  16. ...and 7 more sections

Figures (5)

  • Figure 1: An overview of the proposed Radiologist Copilot with an agent framework and orchestrated tools for radiology reporting with quality control.
  • Figure 2: An illustrative workflow of the Radiologist Copilot.
  • Figure 3: Agent-level evaluation of the Radiologist Copilot using LLM-as-a-Judge.
  • Figure 4: Examples of the Radiologist Copilot workflow.
  • Figure 5: (a) Case study of Hulu-Med and Radiologist Copilot. (b) The validation of the Quality Controller Tool.