The accuracy of CBCT for the detection and diagnosis of vertical root fractures in vivo

Aim To compare the in vivo accuracy of CBCT for the detection of fracture lines versus the diagnosis of vertical root fractures (VRFs) according to characteristic patterns of associated bone resorption. Methodology Eighty-eight patients with symptoms typical of VRFs in root filled teeth, who underwent a CBCT examination and later had the teeth extracted, were divided into two groups: the fracture group (n = 65) and the control group (n = 23). Five blinded observers assessed the CBCT images in two sessions. During the first session, they were asked to state the diagnosis according to the CBCT and clinical data. During the second session after 2 weeks, they assessed only axial slices and were asked to detect a fracture line. The mean CBCT specificity, sensitivity, accuracy values and area under the receiver operating characteristic (AUROC) curve were calculated and compared using the Wilcoxon signed-rank test. Results The average sensitivity of CBCT for the diagnosis of VRFs was 0.84 +/- 0.2. The accuracy and AUC values were 0.81 +/- 0.08 and 0.84 +/- 0.17, respectively. The sensitivity, accuracy and AUC values for the detection of VRFs were significantly lower: 0.17 +/- 0.24 (P = 0.042), 0.54 +/- 0.07 (P = 0.043), and 0.52 +/- 0.09 (P = 0.043), respectively. The specificity of CBCT for the detection and diagnosis of VRFs did not differ significantly (P = 0.50). Conclusion Cone-beam computed tomography was helpful in VRF diagnosis even when it was not possible to visualize the fracture line.