Assessment of mandibular landmark specification: correspondence between two-dimensional radiography and three-dimensional computed tomography.

Abstract

Objectives: This study aimed to establish accurate correspondence between landmarks on lateral (LA) and posteroanterior (PA) cephalograms and to demonstrate the locational relationship of these landmarks in two-dimensional (2D) and three-dimensional (3D) images.

Methods: Twenty patients with skeletal Class III jaw deformity, without significant facial asymmetry, were recruited for this study. LA and PA cephalograms, taken simultaneously just before the orthognathic surgery, were utilized. Twelve mandibular landmarks were identified on each cephalogram, and their 3D coordinates values, as well as the residual differences between LA and PA cephalograms were calculated. To clarify the geometric correspondence between the LA and PA landmarks, the epipolar line method was introduced, which aimed to reduce residuals. In addition, 3D coordinate values of landmarks were compared between cephalograms and CT images in five patients with minimal asymmetry, by measuring the distance between corresponding landmarks as an error metric.

Results: The use of epipolar lines significantly reduced the residuals for 3D coordinates across all landmarks compared to those without epipolar lines. In comparison of 3D coordinate values between cephalograms and CT images, the average error across all landmarks was approximately 0.64 mm, which was lower than reported in previous study.

Conclusion: Incorporating epipolar lines during landmark extraction from LA and PA cephalograms significantly improved the accuracy and reliability of landmark correspondence between the two cephalograms. The 3D coordinate values derived from cephalograms closely matched those obtained from CT images, demonstrating high accuracy although they were independently derived.