Accuracy of anatomical landmark identification using different CBCT- and MSCT-based 3D images: an in vitro study

The aim of this study was to evaluate the reproducibility of anatomical landmarks and the accuracy of different cone-beam CTs (CBCTs/DVTs) and a multislice spiral CT (MSCT) scanner.

A human, fresh-frozen cadaver head was scanned with four CBCTs (Accuitomo 3D, 3D eXam, Pax Reve 3D, Pax Zenith 3D) and one MSCT (SOMATOM Sensation 64) scanner. The three-dimensional (3D) reconstruction of the volume data sets and location of the anthropometric landmarks (n=11), together with linear (n=5) and angular (n=1) measurements were carried out by three examiners using the program VoXim® 6.1. The measurements were taken twice at a 14-day interval. Descriptive analyses were made and the standard deviations were used to compare differences in the accuracy of landmark identification.

The descriptive statistics showed distinct differences in the reference points in the three axes of the coordinate system. Because of anatomical and morphological factors, the pogonion and gnathion reference points displayed higher standard deviations when set on the transverse plane (SD(CBCT) Pog: 0.66-1.57 mm; SD(MSCT) Pog: 0.14-1.09 mm; SD(CBCT) Gn: 1.05-1.77 mm; SD(MSCT) Gn: 0.20-0.85 mm), thus showing less accuracy. However, standard deviations on the sagittal and vertical planes were smaller. Genion, anterior nasal spine and infradentale had very low standard deviations on all three planes. The distance (Mfl-Mfr) and angle (Krl-Krr-Ge) revealed significantly smaller standard deviations in the MSCT (SD(CBCT) Krl-Krr-Ge: 0.51-0.75 mm; SD(MSCT) Krl-Krr-Ge: 0.22 mm).

The CBCT devices evaluated in this study are suitable for taking exact 3D measurements of anatomical structures and meet all requirements for 3D cephalometric analysis.