Reliability of computer-generated cephalometrics.

The use of computer-generated cephalometric analyses has gained popularity in both research and clinical applications. This study was conducted to determine the reliability of the computerized cephalometric methodology. A customized cephalometric analysis of 40 cephalometric radiographs was performed using 22 cephalometric landmarks to evaluate 40 measurements. Reproducibility studies were performed for each step of the computerized cephalometry used. The computer-generated measurements were also compared with those obtained through the conventional hand-measured method. The measurements obtained through the computerized methodology were reproducible for most parameters studied, although point B was found to be unreliable in the vertical plane. The combined errors from video imaging, digitization of the image, and the software were not methodologically significant. No statistically significant errors were found for the repeated measurements of the retrieved digital images. However, the absolute values of all parameters related to the horizontal plane from the computerized method were 0.7 to 1.0 mm greater than those from the conventional hand-measured method, indicating that some horizontal magnification occurred. This information is useful for the interpretation of data obtained from computerized cephalometry.