Lesion Segmentation on Medical Lesion Segmentation Dataset (test)
89.97DicennU-Net (2D)
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| nnU-Net (2D)Params (M)=30.06, GFLOPs=16.09, Training (min)=848*, Epochs=1000, Inf. (ms/slice)=598.5**, Hardware=NVIDIA RTX 3090 Ti GPU, Test-Time Augmentation (TTA)=original + horizontal flip + vertical flip (3 forward passes), Preprocessing/Postprocessing Pipeline=integrated preprocessing/postprocessing pipeline, Training Protocol=default 2D self-configuring protocol2026.07 | 89.97 | 81.77 | 92.05 | 87.98 | 7.98 | 93.76 | 88.78 | |
| EPRA U-NetParams (M)=7.44, GFLOPs=6.02, Training (min)=131.7, Epochs=60, Inf. (ms/slice)=33.97*, Hardware=NVIDIA RTX 3090 Ti GPU2026.07 | 89.84 | 81.55 | 90.82 | 88.87 | 11.62 | 93.78 | 89.25 | |
| UNet++Params (M)=26.28, GFLOPs=18.41, Training (min)=139.1, Epochs=60, Inf. (ms/slice)=37.70, Hardware=NVIDIA RTX 3090 Ti GPU2026.07 | 87.99 | 78.56 | 91.71 | 84.56 | 14 | 92.42 | 85.9 | |
| TransUNetParams (M)=92.84, GFLOPs=7.43, Training (min)=112, Epochs=60, Inf. (ms/slice)=93.71, Hardware=NVIDIA RTX 3090 Ti GPU2026.07 | 86.43 | 76.1 | 89.13 | 83.88 | 13.41 | 90.56 | 84.88 | |
| DeepLabV3+Params (M)=22.44, GFLOPs=7.88, Training (min)=122.5, Epochs=60, Inf. (ms/slice)=25.83, Hardware=NVIDIA RTX 3090 Ti GPU2026.07 | 85.37 | 74.48 | 88.06 | 82.85 | 16.72 | 92.5 | 83.84 |