Objective: Develop an automated algorithm that accurately aligns upper and lower 3D digital jaw images with minimal operator input.
Methods: Digital images of maxillary and mandibular tooth surfaces and alignment records are generally acquired independently; thus, the resulting 3D databases are stored with their own coordinate systems. Automatic alignment is done in two stages. First, images of the jaws and alignment record are brought into close alignment in the same coordinate system. A plane in generalized 3D coordinates and a 2D ellipse equation within the plane characterize each of the three databases. Plane and ellipse equations are obtained analytically using least-square regressions to fit the equations to the database. The jaw planes are aligned to the alignment record plane, then the best-fit 2D ellipses are translated and rotated to align. Final alignment uses an optimizing algorithm to match corresponding anatomy of the jaw and alignment records.
Results: Preliminary testing shows that alignment of the databases can be done without operator input. Evaluation of algorithm-generated occlusal contacts found good subjective agreement with contacts generated in a previous study that aligned databases manually using clinical records. Using a 2.0GHz CPU computer, 3D jaw images with 40,000 points, and an alignment record with 20,000 points, the algorithm takes less than 30 seconds to complete alignment and generate occlusal contacts.
Conclusions: Reproducing occlusal contacts with one button click is achieved by geometric characterizations of 3D jaw and alignment record databases. Future work will compare automatically generated contacts to clinical contacts to determine the accuracy of the algorithm.