Cone beam computed tomography for diagnosis of bisphosphonate-related osteonecrosis of the jaw: evaluation of quantitative and qualitative image parameters

• DOI: 10.1007/s00256-014-1951-1
• Publication Date: 2014-07-03T23:20:51Z
• Authors: Roman Guggenberger; Emrah Koral; Wolfgang Zemann; Christine Jacobsen; Gustav Andreisek; Philipp Metzler
• Keywords: Region of interest; Kappa; Cut-off

To assess the diagnostic performance of quantitative and qualitative image parameters in cone-beam computed tomography (CBCT) for diagnosis of bisphosphonate-related osteonecrosis of the jaw (BRONJ).A BRONJ (22 patients, mean age 70.0 years) group was age and gender matched to a healthy control group (22 patients, mean age 68.0 years). On CBCT images two independent readers performed quantitative bone density value (BDV) measurements with region and volume-of-interest (ROI and VOI) based approaches and qualitative scoring of BRONJ-associated necrosis, sclerosis and periosteal thickening (1 = not present to 5 = definitely present). Intraoperative and clinical findings served as standard of reference. Interreader agreements and diagnostic performance were assessed by intraclass correlation coefficients (ICC), kappa-statistics and receiver-operating characteristic (ROC) analysis.Twenty-three regions in 22 patients were affected by BRONJ. ICC values for mean BDV VOI and mean BDV ROI were 0.864 and 0.968, respectively (p < 0.001). The area under the curve (AUC) for mean BDV VOI and mean BDV ROI was 0.58/0.83 with a sensitivity of 57/83% and specificity of 61/77% for diagnosis of BRONJ, respectively. Kappa values for presence of necrosis, sclerosis and periosteal thickening were 0.575, 0.617 and 0.885, respectively. AUC values for qualitative parameters ranged between 0.90-0.96 with sensitivity of 96% and specificities between 79-96% at respective cutoff scores.BRONJ can be effectively diagnosed with CBCT. Qualitative image parameters yield a higher diagnostic performance than quantitative parameters, and ROI-based attenuation measurements were more accurate than VOI-based measurements.