A new artificial intelligence framework, known as DentAll, has been developed to improve the automated localization and segmentation of individual teeth in panoramic dental radiographs. By integrating advanced instance segmentation with contextual estimation of anatomically expected missing-tooth locations, the researchers behind DentAll aimed to provide a more reliable, computationally practical framework for structured dental analysis.
The DentAll system introduces a specialized architecture that utilizes Contextual Spatial Attention Blocks (S-CSAB). These blocks integrate dilated convolutions and region-of-interest-based positional encoding, designed to refine the delineation of tooth boundaries while maintaining anatomical coherence among adjacent teeth. Building upon these spatial features, the system incorporates a Missing-Tooth Localization Predictor (MTLP). This module estimates missing-tooth locations through patient-conditioned coordinate regression. By utilizing the identity of the target tooth and the geometry of the nearest same-arch neighboring teeth, the MTLP can localize missing teeth without requiring explicit, pre-existing missing-tooth annotations in the training data.
To evaluate the framework, the research team tested the system on a dataset of 1,325 panoramic radiographs. The study employed several performance metrics, including COCO-style mask average precision and controlled pseudo-missing localization experiments. Under an all-ground-truth evaluation protocol, the DentAll framework achieved a Dice Similarity Coefficient (DSC) of 88.21%, with a 92.74% score among successfully matched instances. The system further demonstrated an average precision at 50% intersection over union (AP50) of 97.34%, an AP75 of 87.20%, and a mean average precision (mAP50:95) of 68.58%. These high precision scores indicate the model’s effectiveness in boundary delineation.
A critical component of the study was the assessment of the MTLP module in identifying missing teeth. When comparing the performance of the contextual MTLP against an index-only formulation, the researchers found a substantial improvement in accuracy. The contextual MTLP increased the intersection over union (IoU) rate of ≥ 0.5 from 25.20% in the index-only approach to 72.58%. These findings suggest that incorporating patient-specific anatomical context significantly enhances the framework’s ability to localize missing teeth accurately. This approach represents a step forward in creating more robust, automated diagnostic tools for dental clinical workflows, offering a method that manages anatomical variability effectively while reducing the need for manual oversight.
Subject of Research: AI-assisted dental diagnostic imaging
Article Title: Context-aware tooth instance segmentation and missing-tooth localization in panoramic radiographs
Article References: Bajpai, S., Shafiq, H., & Lee, B. (2026). Context-aware tooth instance segmentation and missing-tooth localization in panoramic radiographs. Scientific Reports. https://doi.org/10.1038/s41598-026-78690-0
Image Credits: AI Generated
DOI: 10.1038/s41598-026-78690-0
Keywords: dentistry, panoramic radiographs, instance segmentation, missing-tooth localization, artificial intelligence, computer-aided diagnosis, dental imaging, medical informatics, spatial attention, coordinate regression, clinical workflow, anatomical context, radiology, imaging analysis, machine learning in dentistry
Cite Scienmag News
Blake Davidson. (November 19, 2026). AI Framework Improves Missing-Tooth Localization and Segmentation in Panoramic Radiographs. Scienmag. https://scienmag.com/ai-framework-improves-missing-tooth-localization-and-segmentation-in-panoramic-radiographs/
Blake Davidson. "AI Framework Improves Missing-Tooth Localization and Segmentation in Panoramic Radiographs." Scienmag, 19 November 2026, https://scienmag.com/ai-framework-improves-missing-tooth-localization-and-segmentation-in-panoramic-radiographs/. Accessed 19 November 2026.
Blake Davidson. "AI Framework Improves Missing-Tooth Localization and Segmentation in Panoramic Radiographs." Scienmag. November 19, 2026. https://scienmag.com/ai-framework-improves-missing-tooth-localization-and-segmentation-in-panoramic-radiographs/








