Differentiation of renal angiomyolipoma without visible fat from small clear cell renal cell carcinoma by using specific region of interest on contrast-enhanced CT: a new combination of quantitative tools

Table 5 Diagnostic performance of predictive models based on each scanning phase for differentiating ccRCC and AML.wovf

Model parameter	Model_PCP	Model_CMP	Model_NP	Model_EP
Sensitivity	0.932 (69/74)	0.959 (71/74)	0.946 (70/74)	0.905 (67/74)
Specificity	0.645 (20/31)	0.871 (27/31)	0.774 (24/31)	0.710 (22/31)
Positive predictive value	0.863 (69/80)	0.960 (72/75)	0.909 (70/77)	0.882 (67/76)
Negative predictive value	0.800 (20/25)	0.900 (27/30)	0.857 (24/28)	0.759 (22/29)
Accuracy	0.848 (89/105)	0.933 (98/105)	0.895 (94/105)	0.848 (89/105)
AUC	0.898	0.986	0.935	0.902
AUC (95% CI)
Lower bound	0.828	0.970	0.881	0.834
Upper bound	0.968	1.000	0.989	0.970

ccRCC: clear cell renal cell carcinoma, AML.wovf: angiomyolipoma without visible fat
AUC: area under curve, CI: confidence interval
Model_PCP is a combined model of gender, pseudocapsule sign, angular interface and AVT_PCP (1)
Model_CMP is a combined model of gender, cystic degeneration, RER_CMP (2) and SHR_CMP. Model_NP is a combined model of gender, pseudocapsule sign, RER_NP (2) and HDT_NP
Model_EP is a combined model of gender, angular interface, NEV_EP (2) and HDT_EP
Values are ratios of the numerator and denominator in parentheses

ISSN: 1470-7330