Figure 2

(A) Receiver operating characteristic (ROC) curve of primary tumor SUVmax (tSUVmax) value for predicting bone invasion of patients with pathologic positive results. The optimal cutoff point based on the Youden index is 9.2; the area under the curve (AUC) is 84.5% (95% confidence interval 0.801–0.881) (p < 0.0001). (B) Pairwise comparison of ROC curves of depth of tumor invasion (DOI) and tSUVmax for predicting bone invasion. The optimal DOI cutoff point for predicting bone invasion is 9 mm. AUC for DOI is 84.8% (95% confidence interval 0.805–0.885) and that for tSUVmax is 84.5% (95% confidence interval 0.801–0.881) (p = 0.8936).