Consistency and precision of landmark identification in three-dimensional cone beam computed tomography scans.

The purpose of this study was to quantify the consistency and precision of locating three-dimensional (3D) anatomic landmarks. The hypotheses tested are that these landmarks have characteristic and variable error patterns associated with their type and location. The consistency and precision of nine orthodontists identifying 32 landmarks of 19 patients were quantified. The cone beam computed tomography (CBCT) data were acquired using a Hitachi CB MercuRay system. Prior to the study, all examiners were calibrated with respect to the definitions of the landmarks and on the use of the software program (Dolphin) for identifying the landmarks. In addition, a reference guide was provided that had the definitions and sample images of the landmarks. Data were collected in spreadsheets as x, y, and z co-ordinates and statistically analysed to determine the mean and standard deviation (SD). The mean location for a given landmark on a given patient served as the reference point. The mean of the distances from the reference point was used as the consistency, while the SD of this mean was used as a measure of precision. The error in the x, y, and z planes was calculated in order to determine the specific characteristics of each landmark. The consistency in landmark location and precision did not differ significantly among the nine examiners. Sella turcica was the most consistently (0.50 mm) and most precisely (0.23 mm) identified anatomic landmark. The most inconsistent landmark was porion-right (2.72 mm) and the most imprecise landmark was orbitale-right (1.81 mm). Due to the lack of even distribution of the errors, careful use of these landmarks for analysis purposes is needed.