Cardiac Image Segmentation on MMWHS CT → MRI
80.3Average DiceMAPSeg
Evaluation results
| Method | Links | |
|---|---|---|
| MAPSegTraining Paradigm=Self-training, Network Dimension=3D, Backbone Architecture=CNN, Validation Label Availability=Target labels not used2026.02 | 80.3 | |
| FSUDA-V2Training Paradigm=Self-training, Backbone Architecture=Transformer-based2026.02 | 75.5 | |
| FSUDA-V1Training Paradigm=Self-training, Backbone Architecture=Transformer-based2026.02 | 70.6 | |
| SE-ASATraining Paradigm=Adversarial training, Network Dimension=2D, Backbone Architecture=CNN2026.02 | 69.9 | |
| PA+AALP+GLCL (Proposed Method)Training Paradigm=Self-training, Network Dimension=2D, Backbone Architecture=Transformer-based, Validation Label Availability=Target labels not used2026.02 | 69.9 | |
| MA-UDATraining Paradigm=Adversarial training, Backbone Architecture=Transformer-based2026.02 | 68.7 | |
| DAFormerTraining Paradigm=Self-training, Backbone Architecture=Transformer-based2026.02 | 65.9 | |
| SIFA-v2Training Paradigm=Adversarial training, Network Dimension=2D, Backbone Architecture=CNN2026.02 | 63.4 | |
| SIFA-v1Training Paradigm=Adversarial training, Network Dimension=2D, Backbone Architecture=CNN2026.02 | 62.1 | |
| CyCADATraining Paradigm=Adversarial training, Network Dimension=2D, Backbone Architecture=CNN2026.02 | 57.5 |