Related Experiment Video
Updated: May 30, 2025
Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness
Published on: December 6, 2024
Open-Source Large Language Models in Radiology: A Review and Tutorial for Practical Research and Clinical Deployment
Cody H Savage1, Adway Kanhere1, Vishwa Parekh1
1From the University of Maryland Medical Intelligent Imaging (UM2ii) Center, Department of Diagnostic Radiology and Nuclear Medicine, University of Maryland School of Medicine, 22 S Greene St, Baltimore, MD 21201 (C.H.S., A.K., V.P., F.X.D.); Departments of Radiology, Medicine, and Biomedical Data Science, Stanford University, Palo Alto, Calif (C.P.L.); Department of Computer Science and Electrical Engineering, College of Engineering and Information Technology, University of Maryland, Baltimore County, Baltimore, Md (A.J.); Department of Computer Science, University of Maryland, College Park, College Park, Md (H.H.); and University of Maryland Institute for Health Computing, University of Maryland, North Bethesda, Md (H.H., F.X.D.).
This tutorial guides the implementation of open-source large language models (LLMs) in radiology, offering advantages over proprietary LLMs for medical professionals and researchers. It provides practical tools and techniques for easier adoption and customization.
Area of Science:
- Medical Informatics
- Artificial Intelligence in Healthcare
- Radiology
Background:
- Large language models (LLMs) offer significant potential for improving healthcare delivery and clinical workflows.
- However, the integration of LLMs into healthcare presents challenges related to accuracy, privacy, accessibility, and regulation.
- While proprietary LLMs have gained attention, open-source LLMs offer distinct advantages for medical institutions and researchers.
Purpose of the Study:
- To provide a comprehensive tutorial for implementing open-source LLMs in radiology.
- To highlight the benefits and drawbacks of open-source versus proprietary LLMs in a medical context.
- To offer practical guidance and code for utilizing open-source LLMs in radiology workflows.
Main Methods:
- The article presents a tutorial focusing on the implementation of open-source LLMs.
- It includes examples of tools for text generation and techniques for prompt engineering, retrieval-augmented generation, and fine-tuning.
- Implementation-ready code is provided to facilitate local setup and customization.
Main Results:
- Open-source LLMs present key advantages for medical institutions and researchers compared to proprietary models.
- The tutorial addresses common challenges in implementing open-source LLMs, such as infrastructure and customization.
- Discussion includes differentiating characteristics of popular open-source LLMs and recent advancements.
Conclusions:
- Open-source LLMs are a viable and advantageous alternative to proprietary models in radiology.
- This tutorial aims to lower the barrier to entry for using open-source LLMs in medical research and practice.
- Further adoption of open-source LLMs can be facilitated by accessible tools and community support.
