Linear accuracy and reliability of volume data sets acquired by two CBCT-devices and an MSCT using virtual models: a comparative in-vitro study.

OBJECTIVE: To discriminate clinically relevant aberrance, the accuracy of linear measurements in three-dimensional (3D) reconstructed datasets was investigated.

MATERIALS AND METHODS: Three partly edentulous human skulls were examined. Landmarks were defined prior to acquisition. Two CBCT-scanners and a Quad-slice CT-scanner were used. Actual distances were physically measured with calipers and defined as a reference. Subsequently, from digital DICOM datasets, 3D virtual models were generated using maximum intensity projections (MIPs). Linear measurements were performed by semi-automated image analysis. Virtual and analogue linear measurements were compared using repeated measurements in a mixed model (p ≤ 0.05).

RESULTS: No significant difference was found among all of the digital measurements when compared to one another, whereas a significant difference was found in matched-pairs analysis between CBCT and calipers (p = 0.032). All digitally acquired data resulted in lower mean values compared to the measurements via calipers. A high level of inter-observer reliability was obtained in the digital measurements (inter-rater correlation = 0.988-0.993).

CONCLUSIONS: The reconstructed datasets led to highly consistent values among linear measurements. Yielding sub-millimeter precision, these modalities are assumed to reflect reality in a clinically irrelevant altered manner. During data acquisition and evaluation, a maximum of precision must be achieved.