Morphological Completion Of Mandibular Defects Based On Non-Rigid Registration

This study aims to achieve three-dimensional (3D) morphological completion for mandibular defects based on the proposed non-rigid registration algorithm (PA-NICP_DS). This study was approved by the Bioethics Committee of the institution. Based on Cone Beam Computed Tomography data from patients at the institution, 3D reconstruction was used to obtain 20 morphologically complete mandibular 3D point cloud data. According to Urken's mandibular defect classification, 6 types of mandibular defects (including B, RB, BS, CRB, RBS and BSB) were simulated and constructed for each data, and 120 mandibular defect data were obtained. Based on the PA-NICP_DS algorithm, the 3D mandibular template constructed in previous studies was deformed and registered to each mandibular defect data, resulting in deformed template that maintained complete 3D morphology to achieve morphological completion for the mandibular defects. The 3D deviation (Root Mean Square, RMS value) between the completed mandibular morphology and the original mandibular morphology was calculated to evaluate the effectiveness of the completion. The 3D deviation (RMS value) for different types of mandibular defect completion is presented as follows. B: 1.31±0.31 mm; RB: 1.56±0.48 mm; BS: 1.80±0.52 mm; CRB: 2.06±0.73 mm; RBS: 2.22±0.75 mm; BSB: 3.19±1.07 mm. Statistical analysis revealed a statistically significant difference between the groups. The PA-NICP_DS algorithm can achieve 3D morphological completion for mandibular defects, with better outcomes for small-scale defects compared to large-scale defects in the mandible.