Research Article

Radiomics Analysis of MR Imaging with Gd-EOB-DTPA for Preoperative Prediction of Microvascular Invasion in Hepatocellular Carcinoma: Investigation and Comparison of Different Hepatobiliary Phase Delay Times

Table 1

The most predictive features between MVI-positive and MVI-negative selected by LASSO regression.

PhaseRadiomics parameterWhich category belongs toRegression coefficient

HBP 5 minCompactness2Shape0.30551482
MassShape0.03586056
VoxelSizeShape0.08370105
TextureStrengthNeighborIntensityDifference25-0.4756972
7.7EnergyGrayLevelCooccurenceMatrix3-0.22185
0.7HomogeneityGrayLevelCooccurenceMatrix30.17812152
4.1InformationMeasureCorr1GrayLevelCooccurenceMatrix30.10846544
1.4InverseDiffMomentNormGrayLevelCooccurenceMatrix30.0502487
X10.4InverseDiffNormGrayLevelCooccurenceMatrix3-0.10842471

HBP 10 min.333ShortRunHighGrayLevelEmphaGrayLevelRunLengthMatrix25-0.025523039
0ShortRunHighGrayLevelEmpha -GrayLevelRunLengthMatrix25-0.037606409
NumberOfObjectsShape-0.150641355
SurfaceAreaSurfaceArea0.04154656
1.7ContrastGrayLevelCooccurenceMatrix3-0.327361605
.333.7Dissimilarity7Dissimilarity-0.159344876
7.7EnergyGrayLevelCooccurenceMatrix3-0.247759079
6.7MaxProbabilityGrayLevelCooccurenceMatrix3-0.301320203

HBP 15 minMedianAbsoluteDeviationGradientOrientHistogram0.012259179
5PercentileGradientOrientHistogram-0.064760131
MassShape0.052851135
SphericalDisproportionShape-0.067929609
4.7AutoCorrelationGrayLevelCooccurenceMatrix3-0.182481531
1.7ContrastGrayLevelCooccurenceMatrix3-0.049683698
9.4ContrastGrayLevelCooccurenceMatrix3-0.300911891
6.1DifferenceEntropyGrayLevelCooccurenceMatrix3-0.105303216
4.7DissimilarityGrayLevelCooccurenceMatrix3-0.205762541
8.4InverseDiffNormGrayLevelCooccurenceMatrix3-0.002903426
1.1InverseVarianceGrayLevelCooccurenceMatrix30.046277632
11.4InverseVarianceGrayLevelCooccurenceMatrix3-0.001397914
12.4InverseVarianceGrayLevelCooccurenceMatrix3-0.188965726
8.4MaxProbabilityGrayLevelCooccurenceMatrix3-0.251615445

HBP, hepatobiliary phases.