Application of Mask R-CNN for automatic recognition of teeth and caries in cone-beam computerized tomography
Neural networks pre-trained with COCO transfer learning exhibit superior annotation accuracy compared to those pre-trained with ImageNet. This suggests that COCO's diverse and richly annotated images offer more relevant features for detecting dental structures and carious lesions. Furthermore, employing ResNet-50 as the backbone architecture enhances the detection of teeth and carious regions, achieving significant improvements with just 200 training epochs, potentially increasing the efficiency of clinical image interpretation.
Authors
- Yujie Ma
- Peixuan Liu
- Maged Ali Al‐Aroomi (ORCID: https://orcid.org/0000-0003-2442-8733)
- Wenjie Ren (ORCID: https://orcid.org/0000-0003-1889-1385)
- Canhua Jiang (ORCID: https://orcid.org/0000-0002-8133-1981)
- Ye Liang (ORCID: https://orcid.org/0000-0002-0824-0178)
- Qing Wu (ORCID: https://orcid.org/0000-0002-7273-6944)
- Yunfei Zheng
Institutions
- Central South University (CN)
- Xiangya Hospital Central South University (CN)
Publication Details
- Journal
- BMC Oral Health
- Published
- 2025-06-06
- DOI
- https://doi.org/10.1186/s12903-025-06293-8
- Citations
- 3
- Primary Topic
- Dental Radiography and Imaging
- Type
- article
- Field-Weighted Citation Impact
- 5.37
Funders
- National Natural Science Foundation of China
- Central South University