Anomaly Localization and Classification of Jaw Cysts and Tumours Using Anatomy-Conditioned Diffusion on Panoramic Radiographs
4 affiliations
- 1Interdisciplinary Program in Bioengineering, Graduate School of Engineering, Seoul National University, Seoul, South Korea
- 2Department of Oral and Maxillofacial Radiology and Dental Research Institute, School of Dentistry, Seoul National University, Seoul, South Korea
- 3Department of Applied Bioengineering, Graduate School of Convergence Science and Technology, Seoul National University, Seoul, South Korea
- 4Department of Medical Informatics, School of Medicine, Kyungpook National University, Daegu, South Korea; AI-Driven Medical Innovation Center, Kyungpook National University Hospital, Daegu, South Korea
Abstract
Objectives: Accurate localization and classification of jaw cysts and tumours on panoramic radiographs are important for timely referral and treatment planning but remain challenging due to overlapping anatomy, heterogeneous lesion appearances, and dental restorations. This study aimed to develop a 2-stage diagnostic pipeline consisting of DentalDiff+, an anatomy-conditioned diffusion module for normal-only anomaly localization, and a separate heat map-based supervised classifier for multi-class classification of jaw cysts and tumours. Methods: The diffusion backbone and Anatomy-preserving Condition Network (ACN) within DentalDiff+ were trained using only normal panoramic radiographs (n = 300) to learn lesion-free anatomical priors. During inference, the model reconstructed the input radiograph and generated anomaly heat maps from reconstruction-based deviations. An ACN was introduced to preserve patient-specific normal structures and reduce artifact-driven false activations. The generated heat maps were then concatenated with the original radiographs and used in a separate supervised classifier for normal cases and 4 lesion categories: dentigerous cysts, periapical cysts, odontogenic keratocysts, and ameloblastomas. Results: DentalDiff+ achieved the best overall performance among the evaluated baselines, with a detection IoU of 0.65 ± 0.06 and AP of 0.70 ± 0.13. It also showed improved reconstruction fidelity, with a PSNR of 23.05 ± 2.90, SSIM of 0.75 ± 0.05, and FID of 54.16 ± 2.12. In the heat map-based classification stage, DentalDiff+ achieved an average accuracy of 0.93 ± 0.01, specificity of 0.95 ± 0.01, sensitivity of 0.82 ± 0.03, and multi-class AUC of 0.94 ± 0.01. Conclusions: DentalDiff+ combines normal-only diffusion-based anomaly localization with heat map-guided multi-class classification for jaw cysts and tumours on panoramic radiographs. By incorporating anatomy-conditioned denoising through ACN, the proposed method improved reconstruction fidelity and generated clinically meaningful anomaly cues while supporting lesion subtype classification. These findings suggest the potential of anatomy-conditioned diffusion for decision-support workflows, although larger multi-institutional and prospective validation is required before clinical deployment.
Related
- Evaluation of the Position of Impacted Maxillary Canines on Panoramic Radiographs Using Deep Learning with CBCT as the Reference Standard
- Deep Learning for Root Canal Filling Quality on Panoramic Radiographs with Detection and Grading Measured Separately
- AI-Based Automated Segmentation of the Mandibular Cortex and Mental Foramen for Radiomorphometric Measurement on Panoramic Radiographs: A Feasibility Study
- SBRF-Det: Scale-Boundary Representative Fusion for Multi-Finding Detection in Panoramic Dental Radiographs.
- Deep learning-based prediction of maxillary canine impaction on panoramic radiographs using YOLO-OBB derived DenSE-Net
BibTeX
@article{kim2026anomaly,
title = {{Anomaly Localization and Classification of Jaw Cysts and Tumours Using Anatomy-Conditioned Diffusion on Panoramic Radiographs}},
author = {Sujeong Kim and Taehyung Kim and Dahee Kim and Ji Yong Han and Heejin Yun and Jo-Eun Kim and Kyung-Hoe Huh and Min-Suk Heo and Sam-Sun Lee and Su Yang and Won-Jin Yi},
journal = {International Dental Journal},
year = {2026},
month = oct,
doi = {10.1016/j.identj.2026.111107},
url = {https://doi.org/10.1016/j.identj.2026.111107},
note = {CC BY},
}
RIS
TY - JOUR
TI - Anomaly Localization and Classification of Jaw Cysts and Tumours Using Anatomy-Conditioned Diffusion on Panoramic Radiographs
AU - Sujeong Kim
AU - Taehyung Kim
AU - Dahee Kim
AU - Ji Yong Han
AU - Heejin Yun
AU - Jo-Eun Kim
AU - Kyung-Hoe Huh
AU - Min-Suk Heo
AU - Sam-Sun Lee
AU - Su Yang
AU - Won-Jin Yi
PY - 2026
DA - 2026/10/31
JO - International Dental Journal
DO - 10.1016/j.identj.2026.111107
UR - https://doi.org/10.1016/j.identj.2026.111107
AB - Objectives: Accurate localization and classification of jaw cysts and tumours on panoramic radiographs are important for timely referral and treatment planning but remain challenging due to overlapping anatomy, heterogeneous lesion appearances, and dental restorations. This study aimed to develop a 2-stage diagnostic pipeline consisting of DentalDiff+, an anatomy-conditioned diffusion module for normal-only anomaly localization, and a separate heat map-based supervised classifier for multi-class classification of jaw cysts and tumours. Methods: The diffusion backbone and Anatomy-preserving Condition Network (ACN) within DentalDiff+ were trained using only normal panoramic radiographs (n = 300) to learn lesion-free anatomical priors. During inference, the model reconstructed the input radiograph and generated anomaly heat maps from reconstruction-based deviations. An ACN was introduced to preserve patient-specific normal structures and reduce artifact-driven false activations. The generated heat maps were then concatenated with the original radiographs and used in a separate supervised classifier for normal cases and 4 lesion categories: dentigerous cysts, periapical cysts, odontogenic keratocysts, and ameloblastomas. Results: DentalDiff+ achieved the best overall performance among the evaluated baselines, with a detection IoU of 0.65 ± 0.06 and AP of 0.70 ± 0.13. It also showed improved reconstruction fidelity, with a PSNR of 23.05 ± 2.90, SSIM of 0.75 ± 0.05, and FID of 54.16 ± 2.12. In the heat map-based classification stage, DentalDiff+ achieved an average accuracy of 0.93 ± 0.01, specificity of 0.95 ± 0.01, sensitivity of 0.82 ± 0.03, and multi-class AUC of 0.94 ± 0.01. Conclusions: DentalDiff+ combines normal-only diffusion-based anomaly localization with heat map-guided multi-class classification for jaw cysts and tumours on panoramic radiographs. By incorporating anatomy-conditioned denoising through ACN, the proposed method improved reconstruction fidelity and generated clinically meaningful anomaly cues while supporting lesion subtype classification. These findings suggest the potential of anatomy-conditioned diffusion for decision-support workflows, although larger multi-institutional and prospective validation is required before clinical deployment.
SN - 0020-6539
SN - 1875-595X
ER -