# A study on the precision of voxel- and surface-based mandibular superimposition

## TL;DR

This study compares two techniques for aligning 3D images of the lower jaw to assess their accuracy in tracking dental and bone changes during orthodontic treatment.

## Contribution

The study evaluates and compares the precision of voxel- and surface-based mandibular superimposition techniques using the mandibular ramus as a reference.

## Key findings

- Surface-based superimposition showed lower measurement errors (< 0.10 mm) compared to voxel-based (< 0.20 mm).
- Statistically significant differences were found between voxel- and surface-based methods for multiple measurements.
- Both techniques showed acceptable reliability with surface errors under 0.20 mm.

## Abstract

To apply the more accurate technique for mandibular superimposition and provide a valuable reference for the assessment of mandibular tooth movement and condylar remodeling before and after orthodontic treatment.

This retrospective study involved 38 adult patients who underwent two cone beam computer tomography (CBCT) scans at different stages of treatment at Fujian Provincial People’s Hospital between September, 2020 and December, 2022. The software Dolphin was used for mandible segmentation, enabling voxel‐based mandibular superimposition with the mandibular ramus as the reference region. The Geomagic Wrap software was employed to process surface‐based mandibular superimposition with the mandibular ramus as a reference. Additionally, the voxel and surface‐based methods were compared for precision, with the mandibular ramus being the reference.

After voxel‐based mandibular superimposition using the mandibular ramus as a reference, with all measurement errors (< 0.20 mm). In contrast, the results of surface‐based mandibular superimposition with the same reference, and the measurement errors were all less than 0.10 mm. The Wilcoxon signed‐rank test revealed statistically significant differences between AS1 and BS1, AS2 and BS2, AS3 and BS3, and AS4 and BS4 (all r< 0.05). Moreover, the absolute mean distances of AS1-AS4 were all greater than those of BS1-BS4.

All mandibular superimposition procedures, including the voxel‐ and surface‐based ones using the mandibular ramus as a reference, have acceptable surface errors (< 0.20 mm), indicating the good reliability of these techniques. Under the specified conditions, surface‐based mandibular superimposition appears to yield a higher degree of precision compared with the voxel‐based technique.