Buccal, mesial, and distal root surfaces of 20 single-rooted premolar teeth were evaluated for the presence or absence of simulated defects. Images were obtained using a Classic i-CAT CBCT device (Imaging Sciences International, Inc, Hatfield, PA) and analyzed by 3 oral radiologists. Evaluations were performed with and without CBCT filters. Accuracy, sensitivity, specificity, and positive and negative predictive values were determined. The McNemar test verified the disagreement between all images versus the gold standard and original images versus images with filters; P values < .05 were considered statistically significant. Inter- and intraobserver agreement was calculated by the kappa test.
The means of intra- and interobserver agreement ranged from good to excellent. Sharpen 3×3 had the highest sensitivity (0.64), accuracy (0.62), and negative predictive value (0.33). Specificity was the highest (0.56) for S9, Smooth, Smooth 3×3, and Angio Sharpen High filters. Positive predictive values were the highest (0.81) for the Smooth 3×3 and Angio Sharpen High filters. The McNemar test showed statistically significant differences from the gold standard for all images (P < .05) as well as for original images versus images with the Shadow filter (P = .01). The other filters did not show statistically significant differences from the original images.
The influence of enhancement CBCT filters on ERR diagnosis is small. The Sharpen 3×3 filter gave the best results and the Shadow filter the worst results.