Table II.
ESTIMATED EVALUATION METRICS ACROSS ISOLATED, COMBINED BASELINE AND ENSEMBLE MODELS WITH TWO BASE LEARNERS EACH.
| Metric | Model | Aorta | Cerebral | Cardiac | Average |
|---|---|---|---|---|---|
| RE ↓ | 4DFlowNet-Aorta | 12.36% | 34.46% | 36.99% | 27.94% |
| 4DFlowNet-Cerebral | 49.65% | 29.58% | 69.81% | 49.68% | |
| 4DFlowNet-Cardiac | 36.22% | 37.15% | 33.02% | 35.47% | |
| Baseline Combined | 10.20% | 27.51% | 31.25% | 22.99% | |
| Bagging-2 | 10.62% | 29.54% | 30.73% | 23.63% | |
| Stacking-2 | 10.07% | 24.46% | 32.20% | 22.25% | |
| RMSE ↓ | 4DFlowNet-Aorta | (1.58, 0.53, 0.73) | (1.65, 1.58, 1.62) | (2.12, 1.59, 1.98) | (1.78, 1.23, 1.44) |
| 4DFlowNet-Cerebral | (34.01, 35.74, 30.80) | (1.04, 0.90, 0.94) | (38.18, 40.06, 35.35) | (24.41, 25.57, 22.36) | |
| 4DFlowNet-Cardiac | (4.74, 1.98, 2.22) | (1.76, 1.63, 1.75) | (2.34, 1.54, 2.13) | (2.95, 1.72, 2.03) | |
| Baseline Combined | (1.22, 0.45, 0.60) | (1.05, 0.83, 0.91) | (2.23, 1.43, 1.94) | (1.50, 0.90, 1.15) | |
| Bagging-2 | (1.42, 0.50, 0.65) | (1.07, 0.92, 0.98) | (2.33, 1.50, 2.00) | (1.61, 0.97, 1.21) | |
| Stacking-2 | (1.29, 0.40, 0.53) | (0.81, 0.80, 0.75) | (2.09, 1.62, 1.87) | (1.39, 0.94, 1.05) | |
| k | 4DFlowNet-Aorta | (0.961, 0.969, 1.012) | (0.822, 0.841, 0.856) | (0.862, 0.816, 0.761) | (0.882, 0.875, 0.876) |
| 4DFlowNet-Cerebral | (0.560, 0.578, 0.587) | (0.902, 0.890, 0.912) | (0.770, 0.668, 0.672) | (0.744, 0.712, 0.724) | |
| 4DFlowNet-Cardiac | (0.610, 0.371, 0.572) | (0.834, 0.758, 0.744) | (0.842, 0.853, 0.726) | (0.762, 0.661, 0.681) | |
| Combined | (0.994, 0.964, 0.948) | (0.917, 0.927, 0.899) | (0.882, 0.863, 0.782) | (0.931, 0.918, 0.876) | |
| Bagging-2 | (1.004, 0.960, 0.984) | (0.916, 0.900, 0.889) | (0.898, 0.872, 0.843) | (0.939, 0.911, 0.905) | |
| Stacking-2 | (0.968, 0.975, 1.003) | (0.923, 0.933, 0.933) | (0.850, 0.856, 0.805) | (0.914, 0.921, 0.914) | |
| R2 ↑ | 4DFlowNet-Aorta | (0.978, 0.967, 0.974) | (0.881, 0.865, 0.852) | (0.858, 0.834, 0.762) | (0.906, 0.889, 0.863) |
| 4DFlowNet-Cerebral | (0.705, 0.492, 0.586) | (0.881, 0.895, 0.880) | (0.405, 0.226, 0.256) | (0.664, 0.538, 0.574) | |
| 4DFlowNet-Cardiac | (0.929, 0.533, 0.831) | (0.861, 0.854, 0.821) | (0.847, 0.868, 0.764) | (0.879, 0.752, 0.805) | |
| Combined | (0.987, 0.977, 0.981) | (0.891, 0.919, 0.889) | (0.874, 0.893, 0.794) | (0.917, 0.930, 0.888) | |
| Bagging-2 | (0.984, 0.972, 0.979) | (0.886, 0.900, 0.881) | (0.874, 0.875, 0.798) | (0.915, 0.916, 0.886) | |
| Stacking-2 | (0.985, 0.981, 0.985) | (0.938, 0.920, 0.925) | (0.884, 0.847, 0.813) | (0.936, 0.916, 0.908) |
Each metric’s best value is highlighted in bold. Arrows indicate direction of improvement. RMSE given in cm/s. RMSE, and given by .