Comparative linear accuracy and reliability of cone beam CT derived 2-dimensional and 3-dimensional images constructed using an orthodontic volumetric rendering program

The purpose of this project was to compare the accuracy and reliability of linear measurements made on 2D projections and 3D reconstructions using Dolphin 3D software (Chatsworth, CA) as compared to direct measurements made on human skulls.;Our results show that for 2D mid sagittal measurements that Simulated LC images are accurate and similar to those from PSP images (except for Ba-Na), and for bilateral measurements simulated LC measurements were similar to PSP but less accurate, underestimating dimensions by between 4.7% to 17%. For 3D volumetric renderings, 2/3rd of CBCT measurements are statistically different from actual measurements, however this possibly is not clinically relevant.;The linear dimensions between 6 bilateral and 8 mid-sagittal anatomical landmarks on 23 dentate dry human skulls were measured three times by multiple observers using a digital caliper to provide twenty orthodontic linear measurements. The skulls were stabilized and imaged via PSP digital cephalometry as well as CBCT. The PSP cephalograms were imported into Dolphin (Chatsworth, CA, USA) and the 3D volumetric data set was imported into Dolphin 3D (Version 2.3, Chatsworth, CA, USA). Using Dolphin 3D, planar cephalograms as well as 3D volumetric surface reconstructions were (3D CBCT) generated. The linear measurements between landmarks of each three modalities were then computed by a single observer three times. For 2D measurements, a one way ANOVA for each measurement dimension was calculated as well as a post hoc Scheffe multiple comparison test with the anatomic distance as the control group. 3D measurements were compared to anatomic truth using Student's t test (P≤50.05). The intraclass correlation coefficient (ICC) and absolute linear and percentage error were determined as indices of intraobserver reliability.