Image-based automated measurement of orthodontic tooth displacement using bracket detection

Abstract

Accurate measurement of tooth displacement is essential for orthodontic treatment planning and outcome evaluation. In routine practice, however, commonly used methods such as manual caliper measurement are constrained by limited resolution, marked operator dependence, and low efficiency when repeated measurements are required. This study proposes a fully automated image-based measurement framework in which orthodontic brackets are detected as anatomical reference markers using a YOLOv11-based detection model. The detected brackets are then automatically sorted, classified into maxillary and mandibular groups, and used for scale-normalized distance calculation, enabling high-precision displacement measurement. The method can support real-time longitudinal monitoring during orthodontic treatment and retrospective quantitative analysis of archived clinical images. Evaluation on an intraoral clinical dataset from six orthodontic patients showed that the model achieved an average mAP@0.5 of 95.4% and a precision of 97.9% in patient-level cross-validation, with an average processing time of 40 ms per image. The system measurement resolution reached 0.01 mm. In 140 clinical gauge-block validation measurements, the mean absolute error was 0.028 mm, and all errors were within 0.05 mm. These results indicate that the proposed system provides a rapid, accurate, and repeatable solution for clinical assessment of orthodontic tooth displacement.