Back to Search Start Over

Exploring the performance and explainability of fine-tuned BERT models for neuroradiology protocol assignment.

Authors :
Talebi, Salmonn
Tong, Elizabeth
Li, Anna
Yamin, Ghiam
Zaharchuk, Greg
Mofrad, Mohammad R. K.
Source :
BMC Medical Informatics & Decision Making; 2/7/2024, Vol. 24 Issue 1, p1-12, 12p
Publication Year :
2024

Abstract

Background: Deep learning has demonstrated significant advancements across various domains. However, its implementation in specialized areas, such as medical settings, remains approached with caution. In these high-stake environments, understanding the model's decision-making process is critical. This study assesses the performance of different pretrained Bidirectional Encoder Representations from Transformers (BERT) models and delves into understanding its decision-making within the context of medical image protocol assignment. Methods: Four different pre-trained BERT models (BERT, BioBERT, ClinicalBERT, RoBERTa) were fine-tuned for the medical image protocol classification task. Word importance was measured by attributing the classification output to every word using a gradient-based method. Subsequently, a trained radiologist reviewed the resulting word importance scores to assess the model's decision-making process relative to human reasoning. Results: The BERT model came close to human performance on our test set. The BERT model successfully identified relevant words indicative of the target protocol. Analysis of important words in misclassifications revealed potential systematic errors in the model. Conclusions: The BERT model shows promise in medical image protocol assignment by reaching near human level performance and identifying key words effectively. The detection of systematic errors paves the way for further refinements to enhance its safety and utility in clinical settings. [ABSTRACT FROM AUTHOR]

Details

Language :
English
ISSN :
14726947
Volume :
24
Issue :
1
Database :
Complementary Index
Journal :
BMC Medical Informatics & Decision Making
Publication Type :
Academic Journal
Accession number :
175304713
Full Text :
https://doi.org/10.1186/s12911-024-02444-z