Comparison of deep learning models to detect crossbites on 2D intraoral photographs

Beatrice Noeldeke; Stratos Vassis; Mohammedreza Sefidroodi; Ruben Pauwels; Peter Stoustrup

doi:10.1186/s13005-024-00448-8

Comparison of deep learning models to detect crossbites on 2D intraoral photographs

Beatrice Noeldeke, Stratos Vassis^*, Mohammedreza Sefidroodi, Ruben Pauwels, Peter Stoustrup

^*Corresponding author for this work

Research output: Contribution to journal/Conference contribution in journal/Contribution to newspaper › Journal article › Research › peer-review

Abstract

Background: To support dentists with limited experience, this study trained and compared six convolutional neural networks to detect crossbites and classify non-crossbite, frontal, and lateral crossbites using 2D intraoral photographs. Methods: Based on 676 photographs from 311 orthodontic patients, six convolutional neural network models were trained and compared to classify (1) non-crossbite vs. crossbite and (2) non-crossbite vs. lateral crossbite vs. frontal crossbite. The trained models comprised DenseNet, EfficientNet, MobileNet, ResNet18, ResNet50, and Xception. Findings: Among the models, Xception showed the highest accuracy (98.57%) in the test dataset for classifying non-crossbite vs. crossbite images. When additionally distinguishing between lateral and frontal crossbites, average accuracy decreased with the DenseNet architecture achieving the highest accuracy among the models with 91.43% in the test dataset. Conclusions: Convolutional neural networks show high potential in processing clinical photographs and detecting crossbites. This study provides initial insights into how deep learning models can be used for orthodontic diagnosis of malocclusions based on intraoral 2D photographs.

Original language	English
Article number	45
Journal	Head & Face Medicine
Volume	20
DOIs	https://doi.org/10.1186/s13005-024-00448-8
Publication status	Published - Sept 2024

Keywords

Artificial intelligence
Crossbite
Deep learning
Neural networks
Orthodontic diagnosis

Access to Document

10.1186/s13005-024-00448-8Licence: CC BY-NC-ND

Cite this

@article{38b1e6f51b1c46a0b787fc1fa45573a6,

title = "Comparison of deep learning models to detect crossbites on 2D intraoral photographs",

abstract = "Background: To support dentists with limited experience, this study trained and compared six convolutional neural networks to detect crossbites and classify non-crossbite, frontal, and lateral crossbites using 2D intraoral photographs. Methods: Based on 676 photographs from 311 orthodontic patients, six convolutional neural network models were trained and compared to classify (1) non-crossbite vs. crossbite and (2) non-crossbite vs. lateral crossbite vs. frontal crossbite. The trained models comprised DenseNet, EfficientNet, MobileNet, ResNet18, ResNet50, and Xception. Findings: Among the models, Xception showed the highest accuracy (98.57%) in the test dataset for classifying non-crossbite vs. crossbite images. When additionally distinguishing between lateral and frontal crossbites, average accuracy decreased with the DenseNet architecture achieving the highest accuracy among the models with 91.43% in the test dataset. Conclusions: Convolutional neural networks show high potential in processing clinical photographs and detecting crossbites. This study provides initial insights into how deep learning models can be used for orthodontic diagnosis of malocclusions based on intraoral 2D photographs.",

keywords = "Artificial intelligence, Crossbite, Deep learning, Neural networks, Orthodontic diagnosis",

author = "Beatrice Noeldeke and Stratos Vassis and Mohammedreza Sefidroodi and Ruben Pauwels and Peter Stoustrup",

note = "Publisher Copyright: {\textcopyright} The Author(s) 2024.",

year = "2024",

month = sep,

doi = "10.1186/s13005-024-00448-8",

language = "English",

volume = "20",

journal = "Head & Face Medicine",

issn = "1746-160X",

publisher = "BioMed Central Ltd.",

}

TY - JOUR

T1 - Comparison of deep learning models to detect crossbites on 2D intraoral photographs

AU - Noeldeke, Beatrice

AU - Vassis, Stratos

AU - Sefidroodi, Mohammedreza

AU - Pauwels, Ruben

AU - Stoustrup, Peter

N1 - Publisher Copyright: © The Author(s) 2024.

PY - 2024/9

Y1 - 2024/9

N2 - Background: To support dentists with limited experience, this study trained and compared six convolutional neural networks to detect crossbites and classify non-crossbite, frontal, and lateral crossbites using 2D intraoral photographs. Methods: Based on 676 photographs from 311 orthodontic patients, six convolutional neural network models were trained and compared to classify (1) non-crossbite vs. crossbite and (2) non-crossbite vs. lateral crossbite vs. frontal crossbite. The trained models comprised DenseNet, EfficientNet, MobileNet, ResNet18, ResNet50, and Xception. Findings: Among the models, Xception showed the highest accuracy (98.57%) in the test dataset for classifying non-crossbite vs. crossbite images. When additionally distinguishing between lateral and frontal crossbites, average accuracy decreased with the DenseNet architecture achieving the highest accuracy among the models with 91.43% in the test dataset. Conclusions: Convolutional neural networks show high potential in processing clinical photographs and detecting crossbites. This study provides initial insights into how deep learning models can be used for orthodontic diagnosis of malocclusions based on intraoral 2D photographs.

AB - Background: To support dentists with limited experience, this study trained and compared six convolutional neural networks to detect crossbites and classify non-crossbite, frontal, and lateral crossbites using 2D intraoral photographs. Methods: Based on 676 photographs from 311 orthodontic patients, six convolutional neural network models were trained and compared to classify (1) non-crossbite vs. crossbite and (2) non-crossbite vs. lateral crossbite vs. frontal crossbite. The trained models comprised DenseNet, EfficientNet, MobileNet, ResNet18, ResNet50, and Xception. Findings: Among the models, Xception showed the highest accuracy (98.57%) in the test dataset for classifying non-crossbite vs. crossbite images. When additionally distinguishing between lateral and frontal crossbites, average accuracy decreased with the DenseNet architecture achieving the highest accuracy among the models with 91.43% in the test dataset. Conclusions: Convolutional neural networks show high potential in processing clinical photographs and detecting crossbites. This study provides initial insights into how deep learning models can be used for orthodontic diagnosis of malocclusions based on intraoral 2D photographs.

KW - Artificial intelligence

KW - Crossbite

KW - Deep learning

KW - Neural networks

KW - Orthodontic diagnosis

UR - http://www.scopus.com/inward/record.url?scp=85202951295&partnerID=8YFLogxK

U2 - 10.1186/s13005-024-00448-8

DO - 10.1186/s13005-024-00448-8

M3 - Journal article

C2 - 39223562

AN - SCOPUS:85202951295

SN - 1746-160X

VL - 20

JO - Head & Face Medicine

JF - Head & Face Medicine

M1 - 45

ER -

Comparison of deep learning models to detect crossbites on 2D intraoral photographs

Abstract

Keywords

Access to Document

Other files and links

Fingerprint

Cite this