Domain shift analysis of deep learning models for tooth detection in pediatric panoramic radiographs
Clinical Snapshot
PICO Framework
| P — Population | Pediatric patients with early mixed dentition requiring panoramic radiographs |
| I — Intervention | YOLOv8 and YOLOv10 deep learning models for automated tooth detection |
| C — Comparator | Cross-institutional validation comparing internal dataset (without bite blocks) versus external open-source dataset (with bite blocks) |
| O — Outcomes | Mean average precision (mAP), per-class average precision, precision-recall curves, confusion matrix analysis, and domain-specific error patterns |
Bottom Line
This diagnostic accuracy study demonstrates that deep learning models (YOLOv8/v10) can achieve high performance for automated tooth detection in pediatric panoramic radiographs, with mean average precision exceeding 90% across different institutional settings. However, significant domain shift effects were observed when models trained on one dataset were applied to external data, particularly affecting detection of rare tooth classes and creating specific error patterns (e.g., bite blocks misidentified as primary central incisors). While the technology shows promise for streamlining pediatric dental radiographic interpretation, the domain dependency highlights the need for institution-specific validation and potentially continuous model updating. The lack of traditional diagnostic accuracy metrics and confidence intervals limits clinical interpretation, and further validation studies with larger sample sizes and diverse pediatric populations are needed before clinical implementation.
Key Findings
P Value: Not reported
Effect Size: mAP ranged from 0.901-0.958 across experiments and models
Primary Outcome: Mean average precision (mAP) for automated tooth detection
Nnt Or Sensitivity: YOLOv8 achieved mAP 0.958 (internal), 0.910 (external), 0.906 (combined); YOLOv10 achieved mAP 0.941, 0.906, 0.901 respectively
Confidence Interval: Not reported
Clinical Application
High technical feasibility with modern YOLO architectures, but requires institutional validation due to domain shift effects Relevant to Australian pediatric dental practice; would require TGA approval for clinical implementation and integration with existing radiographic systems used in public and private dental services Pediatric patients aged 6-12 years with early mixed dentition requiring panoramic radiographic assessment
Abstract
The aim of this study was to externally validate object detection models for comprehensive tooth detection on pediatric panoramic radiographs, quantify the impact of domain shift across institutions and imaging protocols, and compare YOLOv8 and YOLOv10. Two datasets were used: an internal set of 200 images of early mixed dentition without bite blocks, and an external, open-source set of 192 images acquired with bite blocks. Performance was assessed using mean average precision (mAP), per‑class AP, precision‑recall curves, and confusion matrix. Overall performance was high but domain dependent. mAP was 0.958/0.941 (YOLOv8/v10) in Experiment 1; 0.910/0.906 in Experiment 2; and 0.906/0.901 in Experiment 3. Error analysis revealed domain‑specific failure modes: bite blocks were occasionally mistaken for the primary central incisor. Deep learning enables accurate and comprehensive detection of teeth on pediatric panoramic radiographs. However, domain shift alters error patterns and impairs the detection of rare yet clinically important classes.
References
- 1.Nishimura, T., Mine, Y., Iwamoto, Y., Okazaki, S., Tabata, E., Takeda, S., Peng, T.-Y., Kakimoto, N., Nomura, R., & Murayama, T. (2026). Domain shift analysis of deep learning models for tooth detection in pediatric panoramic radiographs. Dental Materials Journal. https://doi.org/10.4012/dmj.2025-301
Related Research
JMIR medical education
Evaluating the Pediatric Behavior Guidance of Students Based on Actual Clinical Transcripts Scored by Faculty and Large Language Models: Pilot Comparative Study
14 June 2026
Calcified tissue international
Diagnostic Indices for Epidemiological Assessment of Molar Incisor Hypomineralization: A Systematic Review
11 May 2026
This content is for educational purposes for healthcare professionals only and does not constitute clinical advice. Clinical decisions should be based on individual patient assessment, current guidelines, and appropriate specialist consultation. Editorial Standards · Privacy Policy · Terms of Service