Accuracy of landmark identification on postero-anterior cephalograms

The aim of this study was to examine accuracy in landmark identification on digital postero-anterior cephalograms, (PAC) assessing the intra-observer and the inter-observer repeatability in determining these landmarks.

Twenty postero-anterior digital cephalometric radiographs were randomly selected from data files of patients aged between 11 to 15 years and were used in this study. Thirty-four commonly used PAC landmarks were included in this investigation and each observer performed the determination of all landmarks on every PAC cephalograms twice (T1 and T2). Mean positions located by the three observers in the first recording session both on the x and y axes, were calculated and defined as the "best estimate" for every landmarks. Systematic differences between the two records at T1 and T2 of the same observer were evaluated using the paired t-test with P<0.05. In order to test the hypothesis of equal mean of the vectors of differences to the "best estimate" for the three raters, a one-way ANOVA test was performed.

Most landmarks showed a good reproducibility between the two recording sessions, except for some landmarks located in the zygomatic arch, mandible and dentition. As regards inter-observer reliability, there were statistically significant differences (P<0.05) in both x and y axis.

Each landmark has its own margin of error and there is a clear difference in the localization accuracy between the x and y coordinates. Since condylar points (CDr-CDl), coronoid points (CORr-CORl) and mental foramen points (MFr-MFl) showed to have a questionable localization accuracy, PAC cannot be strictly recommended for accurate evaluation of any asymmetry. Some dental landmarks seem to be poorly reproducible; therefore the usefulness of postero-anterior cephalometric analysis to evaluate dental discrepancies or maxillo-mandibular molar relationship is questioned.