RESUMO
OBJECTIVES: To investigate the accuracy of cone beam CT (CBCT) images obtained with and without artefact reduction (AR) in detecting simulated buccal peri-implant and buccal periodontal defects. METHODS: 42 implants inserted into edentulous mandibles, and 38 teeth present in dry mandibles were used. Simulated buccal peri-implant defects (n = 22) and buccal periodontal defects (n = 22) were prepared. 20 implants and 18 teeth without simulated defects were the control group. Images of the mandibles were obtained using a Planmeca ProMax(®) 3D Max CBCT unit (Planmeca Oy, Helsinki, Finland). Image reconstructions were prepared without and with low, medium and high AR modes. Images were viewed randomly by six observers twice for the presence of defects. Kappa coefficient was calculated. F2_LD_F1 design for non-parametric analysis of longitudinal data was used. Area under curves (AUCs) were calculated for each observer. Significance level was taken as α = 0.05. RESULTS: Intraobserver kappa ranged from 0.140 to 0.792 for peri-implant and from 0.189 to 1.0 for periodontal defects. All factors were statistically significant (p < 0.001), except for image mode and implant brand. Pairwise interactions were found between periodontal defects and peri-implant defects (p < 0.001), observers (p < 0.001), observer and image mode (p < 0.001), defect model and observer (p < 0.001) and defect model, image mode and observer (p = 0.04). AUC values ranged from 0.39 to 0.52 for peri-implant and from 0.45 to 0.71 for periodontal defects. Higher AUC values were found for periodontal defects than for peri-implant defects. CONCLUSIONS: Buccal peri-implant defects were more difficult to detect than buccal periodontal defects. No difference was found among CBCT images obtained with and without AR modes.