Deep learning-assisted cone-beam computed tomographic analysis of condylar changes after mandibular setback surgery

Area of Science:

• Orthodontics and Dentofacial Orthopedics
• Medical Imaging
• Artificial Intelligence in Medicine

Background:

• Skeletal Class III malocclusion often requires surgical correction.
• Assessing condylar changes post-surgery is crucial for treatment outcomes.
• Current methods for condylar analysis can be time-consuming and operator-dependent.

Purpose of the Study:

• To evaluate condylar changes using a fully automated deep learning-based cone-beam computed tomography (CBCT) workflow.
• To quantify volumetric and surface-based condylar changes after mandibular setback surgery.
• To investigate the influence of mandibular rotation on condylar remodeling.

Main Methods:

• A fully automated pipeline integrating nnU-Net segmentation, rigid surface registration, and surface cropping was developed.
• Preoperative and postoperative CBCT scans from 50 skeletal Class III patients were analyzed.
• Condylar changes were quantified using volumetric, linear, and surface-based metrics.

Main Results:

• High segmentation accuracy was achieved (mandible Dice: 0.98, condyle Dice: 0.99).
• Mandibular rotation significantly affected condylar volume changes bilaterally (p=0.039) and surface metrics (p=0.036).
• Condylar volume changes correlated negatively with preoperative volume (r=-0.44 to -0.77, p<0.001).

Conclusions:

• Condylar remodeling after mandibular setback surgery is rotation-dependent and regionally heterogeneous.
• The automated CBCT workflow provides reproducible, operator-independent quantification of condylar changes.
• This standardized framework facilitates robust postoperative assessment of condylar remodeling.