METHODS: 3D craniofacial surface models were reconstructed from pre-surgery Cone-beam CT images of 20 patients. Virtual jaw surgeries were performed with interactive osteotomies and repositioning of bony segments, and saved as simulated post-surgery models using the Cranio-Maxillo Facial application. The known rotational and translational parameters of the virtual surgery served as the gold standard measurement. SPHARM-based shape correspondence and rigid Procrustes alignment were then used to compare pre-surgery and the simulated post-surgery models. The six degrees of freedom (DOF) of the simulated surgical displacements were calculated for each bony segment: Right and left ramus, chin, maxilla and mandibular body. Statistical analysis included: (1) probability of sample mean difference (d, difference between six DOF of the ground truth and SPHARM Procrustes alignment) to be less than 0.5 mm for translation or 5o degrees for rotation; (2) 95% conﬁdence interval; and (3) 95% prediction interval for d. RESULTS: Measurements of all surgical displacements revealed very high probability (ranging p=0.99-1) that the sample hypothesized measurement difference (0mm translation or 00 of rotation differences) was less than 0.5mm and 50 respectively, with exception of the chin (p=0.75) in the Y-plane translations. For this clinical application, measurements within the spatial resolution of the image (0.5mm) in translations and within 50 in rotation were considered accurate. All the 95% CI and 95% PI intervals contained the hypothesized means.
CONCLUSIONS: This pilot study reveals the suitability of the SPHARM-PDM UNC shape correspondence toolbox to accurately measure surgical displacements. 3D shape analysis has the potential to complement current procedures to assess treatment outcomes and stability of jaw surgery.
SUPPORT: NIDCR DE017727, DE018962, DE005215 and NCRR UL1RR025847