Automatic Numbering and Pathological Recognition of Pediatric Teeth Using CNN and Attention Mechanisms

Hongzhou Zhu; Yuhao Qiu; Renjie Hu; Ang Li; Shengji Zhu; Lei Wang

ICASSP 2025accepted0 citations

Automatic Numbering and Pathological Recognition of Pediatric Teeth Using CNN and Attention Mechanisms

Hongzhou Zhu, Yuhao Qiu, Renjie Hu, Ang Li, Shengji Zhu, Lei Wang

Abstract

Preliminary progress has been made in using deep learning networks for tooth segmentation and numbering, as well as pathological identification in dental panoramic images. However, The publicly available datasets specifically for children’s teeth are very scarce. To address this issue, this paper proposes a fully public database of 849 children’s panoramic radiographs. We also introduce two models based on CNN and attention mechanisms: DCD-Net (Dental Classification and Detection Net) and DPD-Net (Dental Pathology Detection Net). The former, when combined with our category refinement model, can automatically segment and number children’s teeth, achieving an mAP@0.5 of 96.4% while significantly reducing the required training images. The latter detects dental pathologies with an mAP@0.5 of 80%.

BibTeX

@inproceedings{icassp2025_automaticnumberi,
  title = {Automatic Numbering and Pathological Recognition of Pediatric Teeth Using CNN and Attention Mechanisms},
  author = {Hongzhou Zhu and Yuhao Qiu and Renjie Hu and Ang Li and Shengji Zhu and Lei Wang},
  booktitle = {ICASSP 2025},
  year = {2025}
}