Table 2.
Radiomics features selected in ANOVA and LASSO regression analysis.
| Radiomics features | Coefficient |
|---|---|
| lbp-2D_firstorder_Range_ scan | 0.007418999 |
| lbp-2D_firstorder_10Percentile_scan | 0.004015608 |
| Wavelet-HHL_glrlm_RunLengthNonUniformityNormalized_scan wavelet-HHH_ngtdm_Contrast_scan |
-0.07654446 -0.00770341 |
| wavelet-HLH_glszm_GrayLevelNonUniformityNormalized_scan | -0.01873799 |
| wavelet-HHL_firstorder_10Percentile_scan | 0.015804638 |
| wavelet-HLH_glszm_SmallAreaHighGrayLevelEmphasis_scan | -0.02602846 |
| wavelet-HHH_glszm_HighGrayLevelZoneEmphasis_scan | 0.001441019 |
| wavelet-HHL_firstorder_Mean_scan | 0.003108156 |
| original_shape_LeastAxisLength_AP | 0.006855814 |
| original_shape_Elongation_AP | 0.00434467 |
| original_gldm_LargeDependenceEmphasis_AP | 0.017969827 |
| original_glrlm_RunLengthNonUniformityNormalized_AP | -0.0047654 |
| logarithm_glrlm_RunLengthNonUniformityNormalized_AP | -0.00065297 |
Fourteen radiomics features with non-zero coefficients in one-way analysis of variance (ANOVA) and the least absolute shrinkage and selection operator (LASSO) logistic regression model were selected.