TLNM: Externally Validated Tooth Detection, Numbering and Segmentation from Smartphone Photographs Using Mask R-CNN

2026-08-06

AI Summary

This study addresses the challenge of oral health screening in resource-limited settings by proposing an end-to-end method for automatic tooth detection, numbering, and segmentation from smartphone photographs. Built upon a customized Mask R-CNN architecture, the approach innovatively integrates gray-world white balance color correction with anatomical structure constraints, enabling tooth-level anatomical mapping from non-clinical images for the first time. Guided by domain knowledge, the model’s design significantly enhances generalization and robustness on real-world user data. Experimental results demonstrate high performance and stability, achieving instance mask AP@50 scores of 0.818 and 0.901 on internal and external test sets, respectively, with a standard deviation of only 0.009 across ten training runs, confirming its strong generalization capability and reliability.

Abstract

Oral health issues affect billions globally, but the cost and limited access to professional dental care hinder preventive oral healthcare. Research relies on clinical-grade radiographs or intraoral camera images, unavailable for public self-screening. This study introduces a tooth localisation and numbering model for smartphone photographs. We developed a customised Mask Region-based Convolutional Neural Network (Mask R-CNN) pipeline trained on 1,272 annotated smartphone images. To address variability in patient-generated health data, the pipeline incorporates two domain-informed mechanisms: a masked gray-world white-balancing algorithm to mitigate artificial colour casts and an anatomically constrained detection layer to enforce structural validity and suppress false positives. Evaluation comprised four stages: internal held-out testing, independent external testing, a descriptive ablation study, and fold-based training stability analysis using the same internal test set. On the internal test set, the model achieved an instance-mask AP@50 of 0.818, class-aware PQ of 0.780, and operational F1 of 0.884. Training stability showed limited between-model variation: across ten runs, instance-mask AP@50 had a standard deviation of 0.009. On the external dataset, the model achieved an instance-mask AP@50 of 0.901, class-aware PQ of 0.832, and operational F1 of 0.928 despite differences in population, sensors, and acquisition protocols. The inference pipeline is available as an open-source, containerised API. These results demonstrate that consumer-grade smartphone imagery can support automated tooth-level anatomical mapping, offering a scalable, potentially low-cost foundation for remote screening and tele-dentistry in resource-constrained environments.

Problem

• tooth detection
• smartphone photography
• dental segmentation
• tele-dentistry
• oral healthcare accessibility

Innovation

• Mask R-CNN
• tooth segmentation
• smartphone photography
• white-balancing algorithm
• anatomically constrained detection