Table IV.
PREDICTION ERRORS OF ISOLATED, COMBINED AND ENSEMBLE MODELS ON THE UNSEEN AORTIC DISSECTION DATA.
| Metric | Model | Aortic dissection |
|---|---|---|
| RE ↓ | 4DFlowNet-Aorta | 34.70% |
| 4DFlowNet-Cardiac | 40.22% | |
| 4DFlowNet-Cerebro | 75.57% | |
| 4DFlowNet-Combined | 30.53% | |
| Bagging-12 | 25.42% | |
| Stacking Blocks-3 | 24.82% | |
| RMSE ↓ | 4DFlowNet-Aorta | (2.01, 2.30, 4.07) |
| 4DFlowNet-Cardiac | (1.86, 2.56, 5.93) | |
| 4DFlowNet-Cerebro | (54.29, 54.41, 72.42) | |
| 4DFlowNet-Combined | (2.41, 2.75, 5.19) | |
| Bagging-12 | (1.62, 1.95, 4.32) | |
| Stacking Blocks-3 | (1.28, 1.73, 3.49) | |
| k | 4DFlowNet-Aorta | (0.927, 0.819, 0.851) |
| 4DFlowNet-Cardiac | (0.711, 0.569, 0.507) | |
| 4DFlowNet-Cerebro | (0.965, 0.692, 0.903) | |
| 4DFlowNet-Combined | (1.067, 0.994, 0.978) | |
| Bagging-12 | (1.002, 0.982, 0.958) | |
| Stacking Blocks-3 | (0.917, 0.880, 0.893) | |
| R2 ↑ | 4DFlowNet-Aorta | (0.824, 0.805, 0.880) |
| 4DFlowNet-Cardiac | (0.778, 0.739, 0.779) | |
| 4DFlowNet-Cerebro | (0.254, 0.215, 0.612) | |
| 4DFlowNet-Combined | (0.836, 0.847, 0.866) | |
| Bagging-12 | (0.889, 0.904, 0.903) | |
| Stacking Blocks-3 | (0.905, 0.894, 0.921) |
Note: Each metric’s best value is highlighted in bold font. Arrows indicate direction of improvement. RMSE given in cm/s. RMSE, , and given by