Using pre-operative radiomics to predict microvascular invasion of hepatocellular carcinoma based on Gd-EOB-DTPA enhanced MRI

Table 4 The univariate logistic regression analysis of each variable with OR in the validation set of the radiomics model

Sequence	Feature	OR	95% CI	P value
TP	Wavelet.GLDM.LLH_DV	1.39	(1.01–1.91)	0.041
PP	GLCM_IDN	1.54	(1.11–2.14)	0.008
T1WI	Exponential.GLDM_DV	1.19	(1.01–1.40)	0.028
DWI	Wavelet.FirstOrder.LLL_Range	1.09	(0.46–2.58)	0.851
DWI	Wavelet.GLSZM.HLL_SZN	2.25	(0.83–6.07)	0.107
DWI	SquareRoot.GLSZM_SZN	9.12	(1.82–15.78)	0.004
DWI	Logarithm.GLSZM_SAHGLE	1.04	(1.02–1.06)	0.030
T2WI	Wavelet.FirstOrder.HLL_Maximum	2.70	(1.16–6.28)	0.012

DV measures the variance in dependence size in the image; ID normalizes the difference between the neighboring intensity values; Range represents the range of gray values in the ROI; SZN measures the variability of size zone volumes in the image; SAHGLE measures the proportion in the image of the joint distribution of smaller size zones with higher gray-level values; Maximum is the maximum gray level intensity within the ROI
OR odds ratio, CI confidence interval, TP transitional phase, PP portal venous phase, T1WI T1-weighted imaging, DWI,diffusion-weighted imaging, T2WI T2-weighted imaging, GLDM grey-level dependence matrix, GLCM grey-level co-occurrence matrix, GLSZM grey-level size-zone matrix, DV dependence variance, IDN inverse difference normalized, SZN size zone non-uniformity, SAHGLE small area high gray level emphasis

ISSN: 1471-2342