White-Matter Lesion Segmentation on ISBI unseen (test)
67.4DSCnnU-Net SSMSU (ours)
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| nnU-Net SSMSU (ours)Method=Self-Supervised Multi-Stage Unlearning2026.03 | 67.4 | 56 | 57.5 | 20.3 | 36.6 | 1.4 | |
| nnUnet*Trainer=Predefined default trainer in nnU-Net2026.03 | 60.9 | 47.7 | 54.3 | 15.4 | 47.2 | 2.2 | |
| LST-AI***Inputs=T1W and FLAIR scans2026.03 | 60.9 | 54 | 55.2 | — | 24.7 | — | |
| Dinsdale et al.2026.03 | 58.8 | 47.1 | 52.2 | 43.7 | 42.8 | 3.4 | |
| nnU-Net Preprocessing**Trainer=Predefined default trainer in nnU-Net, Preprocessing=denoise and N4 bias correction2026.03 | 54 | 39.3 | 57.4 | 19.1 | 57.8 | 3.4 | |
| nnU-Net No DAData Augmentation=false2026.03 | 53.5 | 39.8 | 50 | 35.3 | 55.4 | 4.6 |