Fig. 2

The performance of mSEPT9, AFP and PIVKA-II and their combination in HCC detection in the training cohort. Panel A: the ROC curves and scatter plots for mSEPT9(upper lane) and AFP(middle lane) and PIVKA-II(lower lane). The dashed lines indicate the optimal sensitive and specificity based on the calculation of Youden’s index. AUC is labeled on each panel. The dashed lines in scatter plots indicate the optimized threshold for positive/negative interpretation. The bars in the scatter plots represent mean with 95% confidence interval. Panel B: The ROC curve for combined test and comparisons of detection sensitivity and specificity. The dashed lines ROC curves indicate the optimal sensitive and specificity based on the calculation of Youden’s index. The values of sensitivity and specificity in panel B were from a series of single calculations based on interpretation of sample test results of the corresponding groups. Results of statistics of sensitivity and specificity calculated from ROC curves are shown in Table 2. The asterisks indicate the significance from Chi-square test when compared with the combined group. *:P < 0.05; ***:P < 0.001