Objectives
To evaluate the correlation of measures of periodontal defects in 3D models segmented by an AI-driven tool with the actual defects in dry skulls and mandibles and to verify the influence of arch, presence of metal artifact (dental fillings/metal posts), type of defect and dental implant artifact on the measures.
Material and Methods
45 periodontal defects were measured with a digital caliper and periodontal probe in three human dried skulls and five mandibles. These skulls and mandibles were scanned with a Cone-Beam Computed Tomography (CBCT) device and their digital files followed automated segmentation by an AI-driven tool (Patient Creator, Relu BV, Leuven, Belgium). The same periodontal defects were measured on the digital model generated by the AI-tool. Correlations of the measuring methods were assessed by means of Intraclass Correlation Coefficient and the influence of arch, presence of artifact and type of defects on the differences were assessed by Student’s t-test.
Results
The intraclass correlations ranged from moderate to excellent values. None of the studied factors (arch, dental fillings/metal posts and type of defect) played a role on the differences between actual and digital defects (P > 0.05). Three-wall defects presented significant influence of dental implant artifact on the measures of height (P = 0.002).
Conclusions
3D models generated by the AI-driven tool presented periodontal defects with linear measures ranging from moderate to excellent correlations with the actual measures.