Jove
Visualize
Contact Us

Related Experiment Video

Updated: May 28, 2026

A 3D Digital Model for the Diagnosis and Treatment of Pulmonary Nodules
10:26

A 3D Digital Model for the Diagnosis and Treatment of Pulmonary Nodules

Published on: May 19, 2023

Automated Radiology Report Labeling in Chest X-Ray Pathologies: Development and Evaluation of a Large Language Model

Abdullah Abdullah1, Seong Tae Kim1

  • 1Department of Computer Science and Engineering, Kyung Hee University, 1732 Deogyeong-daero, Giheung-gu, Yongin, Gyeonggi-do, 17104, Republic of Korea, 82 312013761.

JMIR Medical Informatics
|March 28, 2025
PubMed
Summary
This summary is machine-generated.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Interpretability of multimodal neural networks for prediction of visual acuity in patients with branch retinal vein occlusion.

Scientific reports·2026
Same author

A Hybrid Multimodal Emotion Recognition Framework for UX Evaluation Using Generalized Mixture Functions.

Sensors (Basel, Switzerland)·2023
Same journal

The Orphanet Nomenclature and Classification of Rare Diseases for Improved Patient Recognition and Data Interoperability: Qualitative and Quantitative Analysis.

JMIR medical informatics·2026
Same journal

Feasibility of Tailoring Artificial Intelligence-Assisted Ambient Scribes for Intensive Care Unit Rounds: Algorithm Development and Validation.

JMIR medical informatics·2026
Same journal

Determinants of Acute Kidney Injury After Endoscopic Retrograde Cholangiopancreatography in Patients With Liver Cirrhosis: Retrospective Observational Study.

JMIR medical informatics·2026
Same journal

Multimodule Human-Artificial Intelligence Collaboration Pipeline for Large Language Model-Assisted Thematic Analysis Across Digital Health Interview Studies: Comparative Evaluation Study.

JMIR medical informatics·2026
Same journal

Graph Network Feature Space Fusion for Predicting Irregularly Sampled Medical Time-Series Data: Deep Learning Model Development and Validation Study.

JMIR medical informatics·2026
Same journal

Intrasystem Repeatability of S-Detect for Breast Ultrasound Classification With Identical Static Images: Single-Center Retrospective Repeatability Study.

JMIR medical informatics·2026
See all related articles
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

A generative pretrained transformer (GPT)-based large language model (LLM) shows strong performance in labeling radiology reports, outperforming CheXpert and matching CheXbert. This indicates LLMs offer a promising alternative for structured data creation in medical imaging.

Area of Science:

  • Artificial Intelligence in Medical Imaging
  • Natural Language Processing for Clinical Data
  • Radiology Report Analysis

Background:

  • Automated labeling of unstructured radiology reports is vital for structured dataset creation.
  • Current methods like BERT-based models and manual annotations have scalability and performance limitations.

Purpose of the Study:

  • To evaluate a generative pretrained transformer (GPT)-based large language model (LLM) for radiology report labeling.
  • To compare the LLM's effectiveness against CheXbert and CheXpert using the MIMIC-CXR dataset.

Main Methods:

  • An LLM approach was fine-tuned on expert-labeled radiology reports.
  • Performance was assessed on 687 chest X-ray reports, comparing F1 scores for 14 thoracic pathologies.
  • Statistical significance was determined using paired t tests and Wilcoxon signed-rank tests.
Keywords:
BERTGPTLLMgenerative pre-trained transformerslabelinglarge language modelradiology reportthoracic pathologies

More Related Videos

Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness
03:14

Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness

Published on: December 6, 2024

Introduction of an Integrated Pathology Image Management, Artificial Intelligence, and Reporting System
05:33

Introduction of an Integrated Pathology Image Management, Artificial Intelligence, and Reporting System

Published on: July 11, 2025

Related Experiment Videos

Last Updated: May 28, 2026

A 3D Digital Model for the Diagnosis and Treatment of Pulmonary Nodules
10:26

A 3D Digital Model for the Diagnosis and Treatment of Pulmonary Nodules

Published on: May 19, 2023

Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness
03:14

Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness

Published on: December 6, 2024

Introduction of an Integrated Pathology Image Management, Artificial Intelligence, and Reporting System
05:33

Introduction of an Integrated Pathology Image Management, Artificial Intelligence, and Reporting System

Published on: July 11, 2025

Main Results:

  • The GPT-based LLM achieved an average F1 score of 0.9014, outperforming CheXpert (0.8864) and nearing CheXbert (0.9047).
  • For positive/negative certainty, the LLM scored 0.8708, exceeding CheXpert (0.8525).
  • The LLM showed superior performance on complex, longer descriptions due to its extended context length.

Conclusions:

  • GPT-based LLMs offer competitive and often superior performance in radiology report labeling compared to existing methods.
  • LLMs provide enhanced context understanding and reduce the need for feature engineering, making them suitable alternatives to BERT.
  • The extended context length of LLMs is advantageous for processing complex radiological information.