Related Experiment Videos
How I Do It: Understanding and Leveraging Generative AI for Chest Radiology Reporting
Eun Kyoung Hong1,2, Seungho Lee3, Mizuki Nishino1
1Department of Radiology, Brigham and Women's Hospital, 75 Francis St, Boston MA 02215.
Radiology
|August 4, 2026
Summary
Generative artificial intelligence (AI) offers new ways to help radiologists with chest radiograph reporting. This approach uses AI as a draft assistant to improve workflow and consistency, while addressing potential issues.
Area of Science:
- Radiology
- Artificial Intelligence
- Medical Imaging
Background:
- Generative artificial intelligence (AI) is transforming medical practice, especially in diagnostic interpretation and workflow.
- Chest radiography is a prime application for generative AI due to high clinical volume, complexity, and data availability.
Purpose of the Study:
- To provide a practical, experience-based guide for understanding and implementing generative AI in chest radiograph reporting.
- To outline key concepts, architecture, and a clinical framework for AI as a draft assistant in radiology.
Main Methods:
- Describing generative AI concepts and model architecture.
- Presenting a clinical framework for AI integration as a draft assistant.
- Illustrating utility and limitations with representative examples.
Main Results:
- Generative AI can accelerate radiologist workflow and enhance reporting consistency.
- AI tools support trainee education and can mitigate issues like 'hallucinations'.
- Implementation strategies, interface design, and safety checkpoints are crucial for human-in-the-loop review.
Conclusions:
- Generative AI offers significant potential for thoracic radiology practice, using chest radiography as a model.
- Future directions include context-aware modeling, explainability, and multi-institutional validation.
- Radiologists need essential knowledge to safely and effectively use generative AI tools.