Related Experiment Video
Updated: Jun 20, 2026
Three-Dimensional Cephalometric Landmark Annotation Demonstration on Human Cone Beam Computed Tomography Scans
Published on: September 8, 2023
Precision of cephalometric analysis via fully and semiautomatic evaluation of digital lateral cephalographs
T Sommer1, R Ciesielski, J Erbersdobler
1Department of Orthodontics, Christian-Albrechts-University of Kiel, Kiel, Germany. sommer@kfo-zmk.uni-kiel.de
Objectives:
The aim of this study was to evaluate the precision of the cephalometric analysis using the orthodontic software Orthometric((R)) in two modes: fully automatic without any help by the user and semiautomatic with manual determination of all cephalometric landmarks.
Methods:
From 100 indirectly digitally produced cephalographs, 72 were chosen randomly. On each one of these, three methods were applied on 12 decision-relevant cephalometric angles according to Hasund/Segner: (1) hand drawing on a transparent sheet over the cephalographic film; (2) software semiautomatically with manual correction on the screen; and (3) fully automatically. The hand-based method (1) is well established as a quality standard and reference. The angle measurement results of this method were compared with those of the semiautomatic and the fully automatic method by calculation of the mean absolute differences. Additionally, five randomly chosen cephalographs were evaluated ten times using all three methods for statistical evaluation of reproducibility.
Results:
At tenfold evaluation of five cephalographs, the standard deviations of the angles were between 0.05 degrees and 0.8 degrees . These results were similar using all three methods. All three methods had similar reproducibility. At evaluation of 72 cephalographs, the mean absolute angle differences between the hand-based (1) and semiautomatic (2) methods were below 2 degrees and so within the allowed tolerance limit, except for the nasolabial angle. In contrast to this, all mean absolute angle differences between the (1) hand-based and (3) fully automatic methods exceeded 2 degrees , and with this the tolerance limit allowed by literature.
Conclusions:
The fully automatic determination of the cephalometric landmarks has led to relevant errors up to now, so that it is mandatory to check all automatically set landmarks and to correct them if required. In the case of the semiautomatic method, all relevant angles can be determined with the same reliability as hand drawing.
