RadImageNet and ImageNet as Datasets for Transfer Learning in the Assessment of Dental Radiographs: A Comparative Study.

Autor: Okazaki S; Department of Medical Systems Engineering, Graduate School of Biomedical and Health Sciences, Hiroshima University, Hiroshima, Japan.; Project Research Center for Integrating Digital Dentistry, Hiroshima University, Hiroshima, Japan., Mine Y; Department of Medical Systems Engineering, Graduate School of Biomedical and Health Sciences, Hiroshima University, Hiroshima, Japan. mine@hiroshima-u.ac.jp.; Project Research Center for Integrating Digital Dentistry, Hiroshima University, Hiroshima, Japan. mine@hiroshima-u.ac.jp., Yoshimi Y; Department of Orthodontics and Craniofacial Developmental Biology, Graduate School of Biomedical and Health Sciences, Hiroshima University, Hiroshima, Japan., Iwamoto Y; Department of Pediatric Dentistry, Graduate School of Biomedical and Health Sciences, Hiroshima University, Hiroshima, Japan., Ito S; Department of Orthodontics and Craniofacial Developmental Biology, Graduate School of Biomedical and Health Sciences, Hiroshima University, Hiroshima, Japan., Peng TY; School of Dentistry, College of Oral Medicine, Taipei Medical University, Taipei, Taiwan., Nishimura T; Department of Pediatric Dentistry, Graduate School of Biomedical and Health Sciences, Hiroshima University, Hiroshima, Japan., Suehiro T; Department of Genomic Oncology and Oral Medicine, Graduate School of Biomedical and Health Science, Hiroshima University, Hiroshima, Japan., Koizumi Y; Department of Orthodontics and Craniofacial Developmental Biology, Graduate School of Biomedical and Health Sciences, Hiroshima University, Hiroshima, Japan., Nomura R; Department of Pediatric Dentistry, Graduate School of Biomedical and Health Sciences, Hiroshima University, Hiroshima, Japan., Tanimoto K; Department of Orthodontics and Craniofacial Developmental Biology, Graduate School of Biomedical and Health Sciences, Hiroshima University, Hiroshima, Japan., Kakimoto N; Department of Oral and Maxillofacial Radiology, Graduate School of Biomedical and Health Sciences, Hiroshima University, Hiroshima, Japan., Murayama T; Department of Medical Systems Engineering, Graduate School of Biomedical and Health Sciences, Hiroshima University, Hiroshima, Japan. murayatk@hiroshima-u.ac.jp.; Project Research Center for Integrating Digital Dentistry, Hiroshima University, Hiroshima, Japan. murayatk@hiroshima-u.ac.jp.
Jazyk: angličtina
Zdroj: Journal of imaging informatics in medicine [J Imaging Inform Med] 2024 Jul 24. Date of Electronic Publication: 2024 Jul 24.
DOI: 10.1007/s10278-024-01204-9
Abstrakt: Transfer learning (TL) is an alternative approach to the full training of deep learning (DL) models from scratch and can transfer knowledge gained from large-scale data to solve different problems. ImageNet, which is a publicly available large-scale dataset, is a commonly used dataset for TL-based image analysis; many studies have applied pre-trained models from ImageNet to clinical prediction tasks and have reported promising results. However, some have questioned the effectiveness of using ImageNet, which consists solely of natural images, for medical image analysis. The aim of this study was to evaluate whether pre-trained models using RadImageNet, which is a large-scale medical image dataset, could achieve superior performance in classification tasks in dental imaging modalities compared with ImageNet pre-trained models. To evaluate the classification performance of RadImageNet and ImageNet pre-trained models for TL, two dental imaging datasets were used. The tasks were (1) classifying the presence or absence of supernumerary teeth from a dataset of panoramic radiographs and (2) classifying sex from a dataset of lateral cephalometric radiographs. Performance was evaluated by comparing the area under the curve (AUC). On the panoramic radiograph dataset, the RadImageNet models gave average AUCs of 0.68 ± 0.15 (p < 0.01), and the ImageNet models had values of 0.74 ± 0.19. In contrast, on the lateral cephalometric dataset, the RadImageNet models demonstrated average AUCs of 0.76 ± 0.09, and the ImageNet models achieved values of 0.75 ± 0.17. The difference in performance between RadImageNet and ImageNet models in TL depends on the dental image dataset used.
(© 2024. The Author(s).)
Databáze: MEDLINE
Externí odkaz: