Cardiac MRI Segmentation on ACDC (test)
93.39DSCCardioSAM
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| CardioSAMAdaptation Strategy=Frozen Enc. + Topo. Dec., Params (M)=98.22026.03 | 93.39 | 87.61 | 4.2 | 99.2 | |
| nnU-NetAdaptation Strategy=Auto-Configured, Params (M)=45.12026.03 | 89.5 | 81.2 | 5.9 | 96.3 | |
| TransUNetAdaptation Strategy=Full Training, Params (M)=105.02026.03 | 88.9 | 80.1 | 6.2 | 95.7 | |
| SAM + LoRAAdaptation Strategy=PEFT (Rank=16), Params (M)=88.32026.03 | 88.1 | 79.4 | 6.4 | 95.1 | |
| SAM + Prompt TuningAdaptation Strategy=Soft Prompts, Params (M)=86.52026.03 | 87.5 | 78.2 | 6.8 | 94.4 | |
| U-NetAdaptation Strategy=Full Training, Params (M)=31.22026.03 | 87.2 | 77.3 | 7.1 | 94.1 | |
| SAM-2Adaptation Strategy=Zero-Shot, Params (M)=92.42026.03 | 86.4 | 76.5 | 7.2 | 93.5 | |
| MedSAMAdaptation Strategy=Full Fine-Tuning, Params (M)=89.12026.03 | 85.1 | 74.2 | 7.8 | 92.8 | |
| DeepLab v3+Adaptation Strategy=Full Training, Params (M)=59.82026.03 | 84.3 | 72.8 | 8.4 | 91.9 |