Table 3.
ROC analysis of the features selected from mRMR.
| Features | AUC (95% CI) | Sensitivity (95% CI) | Specificity (95% CI) |
|---|---|---|---|
| Radiomics signature | 0.979 (0.873–1.000) | 90.00 (68.3–98.8) | 100.00 (82.4–100.0) |
| wavelet.LLL_glcm_MaximumProbability | 0.903 (0.764–0.974) | 90.00 (68.3–98.8) | 84.21 (60.4–96.6) |
| wavelet.LLH_glcm_Idmn | 0.792 (0.632–0.905) | 60.00 (36.1–80.9) | 94.74 (74.0–99.9) |
| wavelet.LHH_gldm_LargeDependenceLowGrayLevelEmphasis | 0.839 (0.687–0.937) | 70.00 (45.7–88.1) | 94.74 (74.0–99.9) |
| original_shape_Sphericity | 0.718 (0.552–0.850) | 85.00 (62.1–96.8) | 57.89 (33.5–79.7) |
| wavelet.HHH_gldm_DependenceNonUniformityNormalized | 0.703 (0.535–0.838) | 65.00 (40.8–84.6) | 84.21 (60.4–96.6) |
| wavelet.LHL_glcm_Idn | 0.758 (0.594–0.880) | 95.00 (75.1–99.9) | 52.63 (28.9–75.6) |
| wavelet.LLH_gldm_DependenceEntropy | 0.711 (0.543–0.844) | 55.00 (31.5–76.9) | 84.21 (60.4–96.6) |
| wavelet.LLH_glcm_MCC | 0.679 (0.510–0.819) | 55.00 (31.5–76.9) | 84.21 (60.4–96.6) |
| wavelet.LHL_glrlm_LongRunHighGrayLevelEmphasis | 0.737 (0.571–0.865) | 75.00 (50.9–91.3) | 73.68 (48.8–90.9) |
| wavelet.LHL_firstorder_Skewness | 0.647 (0.478–0.793) | 100.00 (83.2–100.0) | 36.84 (16.3–61.6) |