Evaluation of reconstructed images with different voxel sizes of acquisition in the diagnosis of simulated external root resorption using cone beam computed tomography

• Published in: International endodontic journal 2012-03, Vol.45 (3), p.234-239
• Main Authors: Neves, F. S., Vasconcelos, T. V., Vaz, S. L. A., Freitas, D. Q., Haiter-Neto, F.
• Format: Article
• Language: English
• Subjects: Bicuspid - diagnostic imaging; cone beam computed tomography; Cone-Beam Computed Tomography - methods; Dentistry; external root resorption; Humans; Image Processing, Computer-Assisted - methods; noise; Observer Variation; Predictive Value of Tests; Radiographic Image Enhancement - methods; Root Resorption - classification; Root Resorption - diagnostic imaging; Sensitivity and Specificity; Tooth Apex - diagnostic imaging; Tooth Cervix - diagnostic imaging; Tooth Root - diagnostic imaging; voxel size
• ISSN: 0143-2885; 1365-2591
• DOI: 10.1111/j.1365-2591.2011.01966.x

Neves FS, Vasconcelos TV, Vaz SLA, Freitas DQ, Haiter‐Neto F. Evaluation of reconstructed images with different voxel sizes of acquisition in the diagnosis of simulated external root resorption using cone beam computed tomography. International Endodontic Journal, 45, 234–239, 2012. Aim  To evaluate the diagnosis of simulated external root resorption in cone beam computed tomography (CBCT) images acquired with different voxel sizes, then reconstructed with the same voxel size. Methodology  Buccal, mesial and distal root surfaces of twenty single‐rooted premolar teeth were evaluated for the presence or absence of simulated resorption defects. Images were obtained using a Classic i‐CAT CBCT unit with two different voxel sizes (0.25 and 0.30 mm). Then all images were reconstructed to 0.25 mm voxel size. The McNemar’s test was applied to evaluate the agreement between the two protocols (different voxel sizes/reconstruction parameters) with the gold standard (defects created in the root). Accuracy, sensitivity, specificity, positive and negative predictive values were calculated for each voxel and defect sizes. Results  Better sensitivity and accuracy were obtained with 0.25 (0.25 mm) voxel size. Specificity, positive and negative predictive values were highest with the 0.25 (0.30 mm) voxel size. McNemar’s test results revealed significant differences between the protocols with the gold standard (P