Table 3 The accuracy, area under the curve, F1 scores and AVE in feature engineering-based radiomic paths showing great prediction performances
Feature engineering-based radiomic path | ACC (95% CI) | AUC (95% CI) | F1 score (95% CI) | AVE |
|---|---|---|---|---|
CT–A–g–II | 0.907 (0.849, 0.966) | 0.807 (0.701, 0.913) | 0.842 (0.751, 0.934) | 0.852 |
CT–B–d–I | 0.826 (0.753, 0.898) | 0.917 (0.853, 0.981) | 0.767 (0.641, 0.893) | 0.837 |
CT–D–g–II | 0.843 (0.753, 0.932) | 0.877 (0.784, 0.971) | 0.908 (0.842, 0.974) | 0.876 |
CT–B–g–II | 0.881 (0.795, 0.967) | 0.891 (0.829, 0.953) | 0.862 (0.791, 0.933) | 0.878 |
PET–C–e–IV | 0.913 (0.863, 0.963) | 0.960 (0.926, 0.995) | 0.859 (0.770, 0.947) | 0.911 |
PET–C–e–I | 0.879 (0.825, 0.932) | 0.924 (0.839, 1.000) | 0.878 (0.815, 0.941) | 0.894 |