Automatic Cephalometric Landmark Localization on CBCT-Derived Digitally Reconstructed Radiographs for Skeletal Malocclusion Classification

2026-08-17

AI Summary

This study addresses the challenge of automating cephalometric analysis in cone-beam computed tomography (CBCT) by proposing CephViT, a novel model that pioneers the application of Vision Transformers to digitally reconstructed radiographs derived from CBCT data. Integrating coordinate normalization, this framework enables automated landmark localization and skeletal malocclusion classification, establishing an end-to-end pipeline from 3D imaging to 2D analysis. Experimental results demonstrate a mean landmark localization error of 1.28 mm with a 92% success rate and a malocclusion classification accuracy of 70%, achieving performance comparable to manual annotation. These findings validate the efficacy of Vision Transformers in medical image analysis, significantly enhancing clinical assessment efficiency and demonstrating substantial potential for scalable deployment in orthodontic diagnostics.

Abstract

Manual cephalometric landmark annotation is important for craniofacial assessment but is labor-intensive and difficult to scale. We introduce CephViT, a Vision Transformer-based model for automated 2D lateral cephalometric landmark localization, and evaluate its use in downstream skeletal malocclusion classification. CephViT was trained and benchmarked on a public lateral cephalogram dataset, achieving a mean radial error of 1.28 +/- 1.42 mm and a successful detection rate of 92.0% at 3.0 mm. Because the private evaluation cohort consisted of 3D CBCT scans, lateral cephalogram-like digitally reconstructed radiographs (DRRs) were generated from each volume and used as 2D inputs to the landmark localization model. Landmark coordinates were normalized into a common coordinate frame, and skeletal malocclusion classification was performed using landmarks shared between the reference and DRR-based pipelines. Classification performance using DRR-localized landmarks was comparable to that obtained using manually annotated reference landmarks, with accuracies of 70.0% and 68.3%, respectively. These results support the feasibility of automated cephalometric analysis on CBCT-derived DRRs for skeletal malocclusion assessment.

Problem

• Cephalometric Landmark Localization
• Skeletal Malocclusion Classification
• CBCT
• Digitally Reconstructed Radiographs
• Automated Cephalometric Analysis

Innovation

• Vision Transformer
• Cephalometric Landmark Localization
• Digitally Reconstructed Radiographs
• Skeletal Malocclusion Classification
• CBCT