Construction and evaluation of a deep learning-based intelligent diagnosis model for temporomandibular joint osteoarthritis imaging
10.3969/j.issn.1001-3733.2025.04.014
- VernacularTitle:基于深度学习的颞下颌关节骨关节炎影像智能诊断模型构建与评估
- Author:
Dandan WU
1
;
Pei WANG
;
Yang JING
;
Zhen JIA
;
Jian YANG
Author Information
1. 710049,西安交通大学生命科学与技术学院
- Publication Type:Journal Article
- Keywords:
Temporomandibular joint osteoarthritis;
CBCT;
Auxiliary diagnosis;
Yolov5
- From:
Journal of Practical Stomatology
2025;41(4):519-524
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To develop an automatic diagnostic model for temporomandibular joint osteoarthritis(TMJOA)imaging based on deep learning technology,and to assist clinical diagnosis and improve the efficiency and accuracy of TMJOA diagnosis.Methods:CBCT data of 220 patients were collected,and 2 052 sagittal images were exported.Regions of interest were delineated according to the imaging analysis criteria for temporomandibular joint disorders,and the images were classified into TMJOA-free,TMJOA-uncer-tain and TMJOA.The data were randomly divided into a training set and a validation set according to 8∶2 ratio,and the training set data were used to train a TMJOA detection model based on three lightweight YOLOV5 deep learning frameworks,and the models' performance was evaluated on the validation set.Results:The Yolov5N model demonstrated the best performance,achieving a de-tection accuracy,recall,and precision of 92.5％,90.1％and 85.7％on the validation set,respectively.Conclusion:The auto-matic detection model for TMJOA imaging developed in this study can effectively identify arthritic lesions.Artificial intelligence tools are expected to become a powerful auxiliary tool for the clinical diagnosis of TMJOA.
