IADR Abstract Archives
Artifact Reduction Algorithm May Improve CBCT Diagnosis of TAD-Root Contact
Objectives: Cone beam computed tomography (CBCT) has been used to diagnose contact between temporary anchorage device (TAD) and adjacent roots, a major risk for root damage and TAD instability. However, recent evidence suggests that such diagnosis may have a high false positive rate due to metal artifacts. This study examined whether this inaccuracy can be improved by applying an artifact reduction algorithm (ARA).
Methods: Unembalmed pig mandible cadaver specimens (N=36) were used. With soft tissue kept intact, a TAD (Vector 1.4x8 mm) was placed below the lingual furcation of the first permanent molar, a location where root separation is similar to typical human TAD sites. CBCT scans were then taken at 0.4mm and 0.2 mm voxel levels with and without ARA application. Specimens containing the TAD sites were further separated and subjected to micro-CT scans (27-micron voxel size) after careful removal of the TAD. The acquired CBCT and microCT data were analyzed by two independent, blinded raters to diagnose TAD-root contact.
Results: Overall, intra-rater and inter-rater reliability was nearly perfect for micro-CT (Kappa test, κ=0.94-1), moderate to strong for CBCT scans (κ =0.61-0.94). ARA improved reliability of 0.4mm but not 0.2mm voxel CBCT scans. Using micro-CT as the gold standard, the prevalence of positive TAD-root contact diagnosis in this study was 30.5%, while CBCT diagnostic sensitivity and specificity ranges were 91% - 100% and 76% - 92%, respectively. ARA significantly decreased the false positive rate of one rater from 24% to 8% (McNemar test, p < 0.05) for 0.4mm voxel CBCT scans, corresponding to an improvement in positive predictive value from 64.5% to 84.7%.
Conclusions: Diagnosis of TAD-root contact with 0.4mm voxel size CBCT is reliable, but contains a certain degree of false positive inaccuracy, which may be improved by applying ARA during CBCT scan.
Methods: Unembalmed pig mandible cadaver specimens (N=36) were used. With soft tissue kept intact, a TAD (Vector 1.4x8 mm) was placed below the lingual furcation of the first permanent molar, a location where root separation is similar to typical human TAD sites. CBCT scans were then taken at 0.4mm and 0.2 mm voxel levels with and without ARA application. Specimens containing the TAD sites were further separated and subjected to micro-CT scans (27-micron voxel size) after careful removal of the TAD. The acquired CBCT and microCT data were analyzed by two independent, blinded raters to diagnose TAD-root contact.
Results: Overall, intra-rater and inter-rater reliability was nearly perfect for micro-CT (Kappa test, κ=0.94-1), moderate to strong for CBCT scans (κ =0.61-0.94). ARA improved reliability of 0.4mm but not 0.2mm voxel CBCT scans. Using micro-CT as the gold standard, the prevalence of positive TAD-root contact diagnosis in this study was 30.5%, while CBCT diagnostic sensitivity and specificity ranges were 91% - 100% and 76% - 92%, respectively. ARA significantly decreased the false positive rate of one rater from 24% to 8% (McNemar test, p < 0.05) for 0.4mm voxel CBCT scans, corresponding to an improvement in positive predictive value from 64.5% to 84.7%.
Conclusions: Diagnosis of TAD-root contact with 0.4mm voxel size CBCT is reliable, but contains a certain degree of false positive inaccuracy, which may be improved by applying ARA during CBCT scan.