Objective: To test the feasibility of automated lateral cephalometric radiograph (LCR) superimposition using an image fitting algorithm.
Materials and methods: Using radiopaque markers, we identified seven cephalometric landmarks on three dry skulls, took digital LCRs on each in several rotated positions, and used a custom software program (XRay3D) to automatically superimpose each rotated image on the initial image using an anterior cranial base reference. We measured superimposition error at each landmark and adjusted image brightness levels to simulate potential fitting error due to exposure variation.
Results: The greatest mean error for 24 image rotation trials of less than 10 degrees was less than 0.5 mm. Rotations of 10 degrees or more were not reliably superimposed. Errors of 0.2-1.6 mm occurred for +/-10% brightness but increased exponentially with further brightness alteration.
Conclusion: Automated superimposition of LCRs, using this fitting technique, has great potential when rotation is less than 10 degrees and brightness variation is less than 10%.