Small bowel segmentation on Target domain (test)
83.7DiceOurs
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Oursfeature_disentanglement=true, adaptation_level=feature and output2021.07 | 83.7 | 10.29 | 2.388 | — | |
| Ours + int. feat.feature_disentanglement=true, adaptation_level=feature and output, intensity_features=true2021.07 | 83.7 | 10.559 | 2.536 | 0.86 | |
| Ours w/o output level DAfeature_disentanglement=true, adaptation_level=feature2021.07 | 82.9 | 14.024 | 2.94 | 0.137 | |
| Ours w/o feature level DAfeature_disentanglement=true, adaptation_level=output2021.07 | 82 | 14.745 | 3.085 | 0.006 | |
| multi output level DAadaptation_level=multi-output2021.07 | 81.6 | 14.599 | 3.121 | 0.0001 | |
| feat. & out. level DAadaptation_level=feature and output2021.07 | 81.4 | 12.487 | 2.892 | 0.006 | |
| output level DAadaptation_level=output2021.07 | 81.3 | 15.355 | 3.243 | 0.016 | |
| feat. & out. level DA w/ grad.adaptation_level=feature and output, input_type=gradient2021.07 | 81.1 | 17.948 | 3.527 | 0.002 | |
| tar 5-fold CVdomain_adaptation=none, validation=5-fold cross-validation2021.07 | 80.9 | 13.024 | 2.998 | 0.004 | |
| feature level DAadaptation_level=feature2021.07 | 80.7 | 13.467 | 3.04 | 0 | |
| gradient inputdomain_adaptation=none, input_type=gradient2021.07 | 74.9 | 17.405 | 4.178 | 0 | |
| w/o DAdomain_adaptation=none2021.07 | 72.5 | 15.944 | 3.622 | 0 |