Evaluation of cone-beam computed tomography in the diagnosis of simulated small osseous defects in the mandibular condyle

Introduction In this study, we investigated the impact of defect size and scan voxel size on the accuracy of cone-beam computed tomography (CBCT) diagnoses of simulated condylar defects and assessed the value of orthodontic CBCT images typically scanned at lower settings (0.4-mm voxel size and full-size field of view) in diagnosing condylar erosion defects. Methods Cylindrical holes simulating condylar defects with varied diameters (≤2, 2-3, and >3 mm) and depths (≤2 and >2 mm) were created in 22 fresh pig mandibular condyles, with defect number and size per condyle and quadrant randomly determined. With the soft tissues repositioned, 2 CBCT scans (voxel sizes, 0.4 and 0.2 mm) of the pig heads were obtained from an i-CAT unit (Imaging Science International, Hatfield, Pa). Reconstructed CBCT data were analyzed independently by 2 calibrated, blinded raters using Dolphin-3D (Dolphin Imaging and Management Solutions, Chatsworth, Calif) for defect identification and localization and defect diameter and depth measurements, which were compared with physical diagnoses obtained from polyvinyl siloxane impressions. Results Identification and localization of simulated defects demonstrated moderate interrater reliability and excellent specificity and sensitivity, except for extremely small defects (both diameter and depth ≤2 mm) viewed with 0.4-mm scans, which had a significantly lower sensitivity (67.3%). Geometric measurements of simulated defects demonstrated good but not excellent interrater reliability and submillimeter inaccuracy for all defects. Receiver operating characteristic analyses demonstrated that the overall accuracy of diagnosing simulated condylar defects based on CBCT geometric measurements was fair and good for the 0.4-mm and 0.2-mm voxel-size scans, respectively. With the prevalence of condylar erosion defects in the patients considered, the positive predictive values of diagnoses based on 0.5-mm size (diameter or depth) cutoff points were near 15% and 50% for asymptomatic and symptomatic temporomandibular joints, respectively; the negative predictive values were near 95% and 90%, respectively. Conclusions When using orthodontic CBCT images for diagnosing condylar osseous defects, extremely small (