Related Experiment Video
Updated: Jul 4, 2025
04:23
A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images
Published on: April 21, 2023
1.8K
Bidirectional Encoder Representations from Transformers in Radiology: A Systematic Review of Natural Language
Larisa Gorenstein1, Eli Konen2, Michael Green1
1Department of Diagnostic Imaging, Sheba Medical Center, Ramat-Gan, Israel; Faculty of Medicine, Tel Aviv University, Tel Aviv, Israel.
Journal of the American College of Radiology : JACR
|February 1, 2024
Summary
Bidirectional Encoder Representations from Transformers (BERT) models are increasingly used in radiology for report classification and information extraction. Future applications may enhance diagnostic precision and patient care.
Area of Science:
- Natural Language Processing
- Artificial Intelligence in Medicine
- Radiology Informatics
Background:
- Bidirectional Encoder Representations from Transformers (BERT) has significantly advanced natural language processing since 2018.
- BERT's contextual understanding has led to innovative applications, particularly in the field of radiology.
- This study systematically reviewed BERT's influence and applications within the radiologic domain.
Purpose of the Study:
- To assess the impact and applications of BERT-based models in radiology.
- To identify trends in the use of BERT for natural language processing tasks in medical imaging.
Main Methods:
- A systematic review was conducted following PRISMA guidelines.
- Literature searches were performed on PubMed for studies published between January 1, 2018, and February 12, 2023.
- Keywords included BERT, natural language processing, transformer architecture, and radiology-specific terms.
Main Results:
- 30 out of 597 studies met the inclusion criteria, with a focus on retrospective analyses.
- The majority of studies (14) were published in 2022.
- Primary applications involved classification and information extraction from radiology reports, predominantly using X-ray data. Automatic CT protocol assignment and chest X-ray interpretation were noted.
Conclusions:
- BERT is primarily utilized in radiology for report classification.
- Emerging applications include protocol assignment and report generation.
- Advancements in BERT technology are expected to drive further innovation, improving diagnostic accuracy, report efficiency, and patient care in radiology.
