A New Method for Quantifying Craniofacial Shape Changes

Objectives: The aims of this study was to introduce and validate a non-rigid deformation method to quantify corresponding shape changes for highly complex structures in the craniofacial complex.
Methods: Three patients with temporomandibular joint disorder and a history of Maxillo-Mandibular Advancement (MMA) surgery with at least 4 longitudinal CBCTs (pre-surgery = T1, post-surgery = T2, 1 year follow up = T3, 2-3 year follow up = T4) were selected for the study. The CBCTs were registered on the anterior cranial fossa and 3D models of the registered mandibles were built. Shape correspondences were estimated using group-wise diffeomorphic shape registration. Accuracy of the group-wise diffeomorphic method was calculated by measuring the closest point errors between the calculated diffeomorphic surfaces and the actual 3D surfaces for the mandible. Corresponding shape analysis results were computed using ModelToModelDistance to evaluate 1) the amount of surgical movement, 2) the amount of condylar displacement/ remodeling and amount of bone healing at the osteotomy site after orthognathic surgery.
Results: The mean error between the 3D mandibular models generated from the group-wise diffeomorphic shape registration method to the actual 3D mesh was 0.01mm +/- 0.18mm. 95% confidence interval histograms show error distribution between -0.3mm to 0.2mm (Figure 1). Colormap shows the degree of condylar displacement at different timeThe condyles were displaced distally from T1 to T2 and mesially relapsed from T2 to T3. Condylar displacement was minimal (0.92mm) from T3 to T4 (Figure 2).
Conclusions: Group-wise diffeomorphic shape registration is a new and accurate tool to measure corresponding shape changes in the mandible. Unlike other shape correspondence analysis, it can be accurately used for multiple times points.