Abstract:
Manual landmarking is used in several manual and semi-automated prediction guidelines for
approximation of the nose. The manual placement of landmarks may, however, render the analysis
less repeatable due to observer subjectivity and, consequently, have an impact on the accuracy of the
human facial approximation. In order to address this subjectivity and thereby improve facial
approximations, we are developing an automated three-dimensional (3D) method based on an
automatic dense landmarking procedure using non-rigid surface registration. The aim of this study was
to validate the automatic landmarking method by comparing the intra-observer errors (INTRA-OE) and
inter-observer errors (INTER-OE) between automatic and manual landmarking.
Cone beam computed tomography (CBCT) scans of adult South Africans were selected from the Oral
and Dental Hospital, University of Pretoria, South Africa. In this study, the validation of the automatic
landmarking was performed on 20 3D surfaces. INTRA-OE and INTER-OE were analyzed by registering 41
craniometric landmarks from 10 hard-tissue surfaces and 21 capulometric landmarks from 10 soft-tissue
surfaces of the same individuals. Absolute precision of the landmark positioning (both on the samples as
well as the template) was assessed by calculating the measurement error (ME) for each landmark over
different observers. Systematic error (bias) and relative random error (precision) was further quantified
through repeated measures ANOVA (ANOVA-RM).
The analysis showed that the random component of the ME in landmark positioning between the
automatic observations were on average on par with the manual observations, except for the soft-tissue
landmarks where automatic landmarking showed lower ME compared to manual landmarking. No bias
was observed within the craniometric landmarking methods, but some bias was observed for
capulometric landmarking.
In conclusion, this research provides a
first validation of the precision and accuracy of the automatic
placement of landmarks on 3D hard- and soft-tissue surfaces and demonstrates its utilization as a
convenient prerequisite for geometric morphometrics based shape analysis of the nasal complex.
