obtained on dental plaster casts. Both intra- and inter-
operator reliabilities were assessed.
2. Material and methods
Data from six adult Caucasian subjects with full
dentition, no implant surgery, dental fillings, prostheses
or caries that could affect the morphology of teeth
were obtained. The absence of implants and metal
fillings was selected as inclusion criterion to reduce
the presence of metal artefacts that can alter the
measurement process.
All patients were retrospectively selected from a clinical
database and underwent CBCT examination for clinical
reasons uncorrelated with this study. Their plaster casts
poured from alginate impressions, cast in gypsum and
conventionally trimmed, were collected as well. They
reproduced the full arches with no surface damage. The
casts were imaged by a laser scan (iSeries, Dental Wings,
Montreal, Canada), and their 3D digital models obtained [1].
The work described was carried out in accordance with
The Code of Ethics of the World Medical Association
(Declaration of Helsinki). Informed consent was
obtained from all patients, and their privacy rights
observed. Considering the retrospective nature of the
study, no ethical approval was required. No clinical
information was retrieved from the database.
Twelve dental distances (Fig. 1) were measured on dental
plaster casts using a digital calliper; on digital 3D CBCT
images using inVivoDental software (Anatomage, San
Jose, CA); and on laser scanned surfaces using Mirror®
Vectra Software (Canfield Scientific, Fairfield, NJ).
Two different operators performed all measurements
twice. A previous calibration session was performed:
each operator made the whole set of measurements on
a dental plaster cast and on its digital reproduction, as
well as on the CBCT images of a patient not included in
the study. The results were discussed until a consensus
about landmark location was obtained.
Intra- and inter-operator reliability was assessed by
Bland-Altman analysis, and for each comparison both
the reproducibility coefficient and the bias (difference
between measurements divided by the mean value)
were calculated [7, 18].
The mean values were computed separately for tooth
and measurement (mesiodistal and vestibulopalatal or
vestibulolingual crown diameters). The three different
techniques were compared by Bland-Altman analysis
and Kruskal-Wallis test, with the Wilcoxon test for post-
hoc comparisons.
For all tests, the statistical significance level was set to
p < 0.01, with the Bonferroni correction for post-hoc
comparisons.
3. Results
The intra- and inter-operator biases ranged between 0
and 0.34 mm, and only 3/72 biases were equal to larger
than |0.3| mm (Table 1). These biases were observed for
the vestibulopalatal diameters of teeth 24 and 26 (intra-
operator analysis), and the vestibulopalatal diameter of
tooth 26 (inter-operator analysis). Reproducibility ranged
between 72 and 99%, the worst coefficients were found
for CBCT measurements (18/24 were lower than