Automatic identification of individuals using deep learning method on panoramic radiographs.
Automatic identification
Deep learning
Panoramic radiograph
Journal
Journal of dental sciences
ISSN: 2213-8862
Titre abrégé: J Dent Sci
Pays: Netherlands
ID NLM: 101293181
Informations de publication
Date de publication:
Apr 2023
Apr 2023
Historique:
received:
11
10
2022
revised:
18
10
2022
medline:
7
4
2023
entrez:
6
4
2023
pubmed:
7
4
2023
Statut:
ppublish
Résumé
The dentition shows individual characteristics and dental structures are stable with respect to postmortem decomposition, allowing the dentition to be used as an effective tool in forensic dentistry. We developed an automatic identification system using panoramic radiographs (PRs) with a deep learning method. In total, 4966 PRs from 1663 individuals with various changes in image characteristics due to various dental treatments were collected. In total, 3303 images were included in the data set used for model training. Vgg16, Vgg19, ResNet50, ResNet101, and EfficientNet models were applied for identification. The precision curves were evaluated. The matching precision rates of all models (Vgg16, Vgg19, ResNet50, ResNet101, and EfficientNet) were examined. Vgg16 was the best model, with a precision of around 80-90% on 200 epochs, using the Top-N metrics concept with 5-15 candidate labels. The model can successfully identify the individual even with low quantities of dental features in 5-10 s. This identification system with PRs using a deep learning method appears useful. This identification system could prove useful not only for unidentified bodies, but also for unidentified wandering elderly people. This project will be beneficial for police departments and government offices and support disaster responses.
Sections du résumé
Abstract Background/purpose
UNASSIGNED
The dentition shows individual characteristics and dental structures are stable with respect to postmortem decomposition, allowing the dentition to be used as an effective tool in forensic dentistry. We developed an automatic identification system using panoramic radiographs (PRs) with a deep learning method.
Materials and methods
UNASSIGNED
In total, 4966 PRs from 1663 individuals with various changes in image characteristics due to various dental treatments were collected. In total, 3303 images were included in the data set used for model training. Vgg16, Vgg19, ResNet50, ResNet101, and EfficientNet models were applied for identification. The precision curves were evaluated.
Results
UNASSIGNED
The matching precision rates of all models (Vgg16, Vgg19, ResNet50, ResNet101, and EfficientNet) were examined. Vgg16 was the best model, with a precision of around 80-90% on 200 epochs, using the Top-N metrics concept with 5-15 candidate labels. The model can successfully identify the individual even with low quantities of dental features in 5-10 s.
Conclusion
UNASSIGNED
This identification system with PRs using a deep learning method appears useful. This identification system could prove useful not only for unidentified bodies, but also for unidentified wandering elderly people. This project will be beneficial for police departments and government offices and support disaster responses.
Identifiants
pubmed: 37021248
doi: 10.1016/j.jds.2022.10.021
pii: S1991-7902(22)00273-2
pmc: PMC10068681
doi:
Types de publication
Journal Article
Langues
eng
Pagination
696-701Informations de copyright
© 2022 Association for Dental Sciences of the Republic of China. Publishing services by Elsevier.
Déclaration de conflit d'intérêts
The authors have no conflicts of interest relevant to this article.
Références
Oral Radiol. 2021 Jan;37(1):13-19
pubmed: 31893343
Imaging Sci Dent. 2021 Jun;51(2):187-193
pubmed: 34235064
J Forensic Sci. 2003 May;48(3):497-503
pubmed: 12762516
Sci Rep. 2020 Mar 2;10(1):3801
pubmed: 32123249
Dent Clin North Am. 2001 Apr;45(2):237-51
pubmed: 11370453
Dentomaxillofac Radiol. 2022 May 01;51(4):20210383
pubmed: 34826252
Aust Dent J. 2012 Mar;57 Suppl 1:24-32
pubmed: 22376094
Rofo. 2018 Dec;190(12):1152-1158
pubmed: 29913524
Sci Rep. 2020 Aug 11;10(1):13559
pubmed: 32782269
Sci Rep. 2020 Nov 6;10(1):19261
pubmed: 33159125
Comput Biol Med. 2017 Jan 1;80:24-29
pubmed: 27889430