Abstract
The aim of this study was to assess the accuracy of 3D rendering of the mandibular condylar region obtained from different semi-automatic segmentation methodology. A total of 10 Cone beam computed tomography (CBCT) were selected to perform semi-automatic segmentation of the condyles by using three free-source software (Invesalius, version 3.0.0, Centro de Tecnologia da Informação Renato Archer, Campinas, SP, Brazil; ITK-Snap, version2.2.0; Slicer 3D, version 4.10.2) and one commercially available software Dolphin 3D (Dolphin Imaging, version 11.0, Chatsworth, CA, USA). The same models were also manually segmented (Mimics, version 17.01, Materialise, Leuven, Belgium) and set as ground truth. The accuracy of semi-automatic segmentation was evaluated by (1) comparing the volume of each semi-automatic 3D rendered condylar model with that obtained with manual segmentation, (2) deviation analysis of each 3D rendered mandibular models with those obtained from manual segmentation. No significant differences were found in the volumetric dimensions of the condylar models among the tested software (p > 0.05). However, the color-coded map showed underestimation of the condylar models obtained with ITK-Snap and Slicer 3D, and overestimation with Dolphin 3D and Invesalius. Excellent reliability was found for both intra-observer and inter-observer readings. Despite the excellent reliability, the present findings suggest that data of condylar morphology obtained with semi-automatic segmentation should be taken with caution when an accurate definition of condylar boundaries is required.
Highlights
Morphological and dimensional changes of the mandibular condyles can alter craniofacial growth, sometimes favoring the development of skeletal malocclusions, or can be a specific risk factor for temporomandibular disorders (TMD) [1,2,3]
Examination), the analysis suggested that 16 condyles were required to reach the 80% power to detect a mean difference of 21.07 mm3 in the volumetric assessment of the condylar region between manual segmentation and semi-automatic segmentation (ITK-Snap software), with a confidence level of 95%
No statistically significant differences were found in the volumetric dimensions of the rendered condylar models obtained with different software, with manual segmentation (Mimics/ground truth model) showing a median value of 1631 mm3 and semiautomatic segmentation showing median values ranging from 1602 (ITK-snap) to 1662 mm3 (Invesalius) (Table 2)
Summary
Morphological and dimensional changes of the mandibular condyles can alter craniofacial growth, sometimes favoring the development of skeletal malocclusions, or can be a specific risk factor for temporomandibular disorders (TMD) [1,2,3]. Public Health 2020, 17, 4789; doi:10.3390/ijerph17134789 www.mdpi.com/journal/ijerph
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
