The validity of in vivo tooth volume determinations from cone-beam computed tomography

Abstract

Objective: To determine the accuracy of volumetric analysis of teeth in vivo using cone-beam computed tomography (CBCT).

Materials and methods: The physical volume (Vw) of 24 bicuspids extracted for orthodontic purposes (16 were imaged with the I-CAT and 8 with the CB MercuRay) were determined using the water displacement technique. Corresponding pretreatment CBCT image data were uploaded into Amira 4.0 for segmentation and radiographic volume (Va). All measurements were performed twice by two observers. The statistical difference between Vw and Va was assessed using a paired t-test. The intraobserver and interobserver reliability were determined by calculating Pearson correlation coefficients and intraclass correlation coefficients.

Results: The overall mean Vw of teeth specimens was 0.553 +/- 0.082 cm3, while the overall mean Va was 0.548 +/- 0.079 cm3 (0.529 +/- 0.078 cm3 for observer 1 and 0.567 +/- 0.085 cm3 for observer 2). There were statistically significant differences between Va and Vw (P < .05). Between observer 1 and observer 2, Va measurements were statistically significantly different (P < .05). The interobserver and intraobserver correlation coefficient for Vw was high. Lastly, surface smoothing reduced the volume by 3% to 12%.

Conclusions: In vivo determination of tooth volumes from CBCT data is feasible. The measurements slightly deviate from the physical volumes within -4% to 7%. Smoothing operations reduce volume measurements. Currently, no requirements for accuracy of volumetric determinations of tooth volume have been established.

Figures

Figure 1.

Teeth segmentation in Amira

Figure 2.

Virtual segmented teeth compared with physical teeth. (A) Physical four bicuspids, two roots for upper bicuspids. (B) Four bicuspids segmented from cone-beam computed tomography data sets. (C) Lower bicuspid with developing apex. (D) Apex can be seen clearly in segmented tooth

Figure 3.

Virtual teeth before and after smoothing. Upper bicuspid virtual model (A) before and (B) after smoothing. Lower bicuspid (C) before and (D) after smoothing