Factors Affecting Alveolar Bone Height Measurements from CBCT Images

Objectives: Cone beam computed tomography (CBCT) is routinely used to measure alveolar bone dimensions. This study investigated factors likely affecting the accuracy of such measurements. Methods: Marker holes, apical to maxillary and mandibular molar roots, and mesio-distal molar occlusal reference grooves were created in fresh pig heads (n=26 quadrants), followed by CBCT scans at 0.4 and 0.25 mm voxel sizes before and after soft tissue removal. Subsequently, bucco-lingual sections bisecting the marker holes were cut, from which physical alveolar bone height and thickness were measured.  Two blinded, calibrated raters, using Dolphin 3-D and OsiriX software, independently collected alveolar bone height measurements from CBCT images. Differences between CBCT and physical measurements (D_CBCT-Phy) were calculated. The mean D_CBCT-Phy and limit of agreement (LOA, ±1.96SD) for each factor were depicted by Bland-Altman methods. The influence of each factor was examined by repeated measures ANOVAs.  Result s:  The rater factor affected CBCT measurements in both jaws, with mean D_CBCT-Phy from a more-experienced rater significantly (p<0.05) closer to "0" than from a less-experienced rater.  Their LOA ranges were similar (Maxilla, 0.16±2.21mm vs.-0.48±2.53mm; Mandible, 0.01±1.38mm vs. -0.70±1.48mm)._. The software factor affected CBCT measurements in both jaws, with mean D_CBCT-Phy from Osirix significantly (p<0.05) closer to "0" than from Dolphin-3D despite similar LOA ranges (Maxilla, -0.09±2.38mmvs. -0.22±2.42mm; Mandible, -0.30±1.47mm vs. -0.39±1.38mm). The soft-tissue factor only affected the mandible significantly (p<0.05), with mean D_CBCT-Phy from soft-tissue-absent specimens closer to "0" than those from soft-tissue-present specimens (Maxilla, -0.13±2.82mm vs. -0.19±1.99mm; Mandible, -0.25±1.70mm vs. -0.43±1.22mm). The voxel-size factor had an insignificant effect.  Generally, mandibular CBCT measurements exhibited smaller LOA ranges than maxillary, possibly because mandibular alveolar bone was thicker. Conclusions:   Individual alveolar bone height measurements from CBCT images can deviate from physical truth by a clinically relevant amount (1.2-2.5mm). Factors such as rater, software, tissue presence/absence, and bone thickness can introduce systematic sub-millimeter errors.