Applying arterial enhancement fraction (AEF) texture features to predict the tumor response in hepatocellular carcinoma (HCC) treated with Transarterial chemoembolization (TACE)

Table 3 Coefficients of each AEF texture features in two prediction models by multi-variate logistic regression analyses

Model	AUC	Acc.	Sen.	Spe.	Items	Coef.	z	p
A	0.941	0.911	1.000	0.826	Intercept	−7.1555	−0.0000	1.000
					MinIntensity	−8.5860	−1.1469	0.251
					MaxIntensity	−17.1622	−1.1696	0.242
					MedianIntensity	1.9920	0.7683	0.442
					Uniformity	12.5851	1.1778	0.239
					Inertia	4.2988	1.2518	0.211
					ClusterProminence	10.9129	1.2056	0.228
					RunLengthNonuniformity	0.5123	0.3110	0.756
					ShortRunHighGreyLevelEmphasis	−4.6584	−1.0423	0.297
					LongRunHighGreyLevelEmphasis	−12.0604	−0.0000	1.000
B	0.824	0.711	0.581	1.000	Intercept	3.1287	0.0230	0.982
					Kurtosis	−0.5309	− 0.8292	0.407
					ClusterProminence	1.8890	1.7340	0.083
					HighGreyLevelRunEmphasis	0.5523	1.1414	0.254
					LongRunHighGreyLevelEmphasis	5.2633	0.0152	0.988

Note:Model A was applied for the prediction of “Improved” outcome; Model B was applied for the prediction of “Un-worsened” outcome. AUC Area Under Curve; Acc. Accuracy; Sen. Sensitivity; Spe. Specificity

ISSN: 1470-7330