One-step diagnosis of cervical vertebra maturation and sagittal skeletal type based on convolutional neu-ral network
10.3969/j.issn.1001-3733.2025.05.011
- VernacularTitle:基于卷积神经网络的颈椎成熟度及矢状骨面型的一步式诊断
- Author:
Li'na CHEN
1
;
Jukun SONG
;
Yuanqi JIANG
;
Hao LV
;
Min LI
Author Information
1. 550004 贵阳,贵州医科大学口腔医学院
- Publication Type:Journal Article
- Keywords:
Artificial intelligence;
Deep learning;
Convolutional neural network;
InceptionV3;
Cervical vertebra maturation;
Sagittal skeletal type;
Lateral cephalogram
- From:
Journal of Practical Stomatology
2025;41(5):643-650
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To compare the performance of 4 mainstream convolutional neural network models in one-step automatic di-agnosis of cervical vertebra maturation(CVM)and sagittal skeletal type on lateral cephalogram.Methods:600 lateral cephalograms were collected.2 orthodontists were included to determine the stage of CVM and semi-automatic measurement of sagittal skeletal type.After the data set was enhanced,80％was used as the training set and 20％as the test set.The same training method was used to compare the performance of convolutional neural network models of InceptionV3,ResNet50,ShuffleNetV2 and VGG16).Results:The accuracy of InceptionV3,ResNet50,ShuffleNetV2 and VGG16 in distinguishing CVM stage was 98.44％,89.85％,60.55％and 67.97％,and the accuracy of sagittal bone surface classification 96.56％,92.19％,80.00％and 88.13％,respec-tively.Conclusion:Among the 4 convolutional neural network models,InceptionV3 has the best overall learning performance for CVM stage and sagittal skeletal type classification.
