Shape Correspondence validation for 3D quantification of mandibular asymmetry

Objectives: Our aim is to determine if 3D shape correspondence quantification of mandibular asymmetry precisely diagnoses right and left differences in surgical patients. This was achieved by virtually simulating asymmetries and measuring the differences using UNC SPHARM-PDM toolbox. Methods: Cone-Beam CT data was acquired before treatment from 20 orthognathic surgery patients with different degrees and types of facial asymmetry. Segmentations of the maxillofacial hard tissues were created and mirroring for virtual surgical planning was performed using the CranioMaxilloFacial Application software. Mirrored surgical models were superimposed on the actual patient models using two different approaches: (1) mirroring on the midsagittal plane; and (2) mirroring on an arbitrary plane, followed by registration on the cranial base of the original image. Additional mandibular asymmetry in the 3 planes of space was simulated by known amounts. For both mirroring methods, SPHARM-PDM was used to localize and quantify virtually simulated asymmetries. Results: For all 6 degrees of freedom (rotation and translation), there was very high probability that the difference between measured and simulated asymmetry was less than 0.5mm. For mirroring on the midsagittal plane, the probabilities ranged from 0.99 to 1, and for mirroring on an arbitrary plane with cranial base registration, probabilities ranged from 0.84 to 1. Conclusion: Our methodology accurately quantifies mandibular asymmetries. Our work demonstrated that Midsagittal plane mirroring had similar performance to registration on the cranial base after mirroring on an arbitrary plane. Registration on the cranial base could be applied in difficult trauma situations or when key landmarks are unreliable or absent. Future studies will allow surgeons to apply this methodology to precisely plan corrective surgeries for mandibular asymmetry patients.

Supported by NIDCR DE017727, DE018962, DE005215 and NCRR UL1RR025847