1.
经多变量逻辑回归筛选后用于构建影像组学模型的特征
Radiomics features used to construct the radiomics models through multivariate logistic regression model
| Phase | Radiomics features | Coef. | P > |z| | [0.025 | 0.975] |
| non-enhanced | |||||
| wavelet.HLL-gldm-DependenceVariance | 0.8074 | 0.0040 | 0.2581 | 1.3568 | |
| log.sigma.5.0.mm.3D-glrlm-RunLengthNonUniformity | -1.2137 | 0.0019 | -1.9795 | -0.4479 | |
| Original-firstorder-Median | -0.8111 | 0.0058 | -1.3874 | -0.2348 | |
| wavelet.LLL-gldm-SmallDependenceEmphasis | 0.6424 | 0.0136 | 0.1322 | 1.1525 | |
| Cutoff Value | -2.0002 | 0.0000 | -2.6435 | -1.3568 | |
| Arterial | |||||
| wavelet.LLL-gldm-LargeDependenceLowGrayLevelEmphasis | 0.5279 | 0.0100 | 0.1264 | 0.9293 | |
| log.sigma.4.0.mm.3D-firstorder-90Percentile | 0.7139 | 0.0013 | 0.2796 | 1.1482 | |
| wavelet.LLL-firstorder-Median | -0.7103 | 0.0085 | -1.2394 | -0.1811 | |
| Cutoff Value | -1.5967 | 0.0000 | -2.0817 | -1.1118 | |
| Venous phases | |||||
| log.sigma.5.0.mm.3D-firstorder-90Percentile | 0.7291 | 0.0004 | 0.3237 | 1.1345 | |
| Original-gldm-LargeDependenceLowGrayLevelEmphasis | 0.3672 | 0.0496 | 0.0006 | 0.7337 | |
| Cutoff Value | -1.4608 | 0.0000 | -1.8998 | -1.0218 | |
| Combined | |||||
| NP_wavelet.HLL_gldm_DependenceVariance | 1.1184 | 0.0003 | 0.5157 | 1.7211 | |
| NP_wavelet.LLL_firstorder_Median | -1.0359 | 0.0006 | -1.6269 | -0.4448 | |
| VP_original_gldm_LargeDependenceLowGrayLevelEmphasis | 0.6775 | 0.0054 | 0.2004 | 1.1545 | |
| NP_wavelet.HLL_glcm_SumSquares | 0.6300 | 0.0126 | 0.1348 | 1.1252 | |
| NP_log.sigma.5.0.mm.3D_glrlm_RunLengthNonUniformity | -1.0210 | 0.0050 | -1.7342 | -0.3077 | |
| Cutoff Value | -1.5967 | 0.0000 | -2.0817 | -1.1118 |