Medical Image Segmentation on ISIC 2018 (test)
95.11Dice ScoreLatentFM
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| LatentFMApproach=Generative Segmentation Methods2025.12 | 95.11 | 90.67 | — | — | |
| HyPCA-Net-EMCADnumber of parameters (M)=18.6, GFLOPS=7.65, decoder=EMCAD2026.02 | 93.8 | 86.4 | — | — | |
| HyPCA-Net-Segnumber of parameters (M)=21.7, GFLOPS=8.04, decoder=SegNet2026.02 | 92.7 | 85.3 | — | — | |
| LatentDMApproach=Generative Segmentation Methods2025.12 | 91.3 | 84.1 | — | — | |
| FMApproach=Generative Segmentation Methods2025.12 | 91.01 | 83.99 | — | — | |
| PVT-EMCAD-B2#Params=26.76M, #FLOPs=5.6G, Resolution=256x2562024.05 | 90.96 | — | — | — | |
| EMCADnumber of parameters (M)=26.8, GFLOPS=5.62026.02 | 90.9 | 84.1 | — | — | |
| nnUNetApproach=Deterministic Segmentation Methods2025.12 | 90.8 | 83.6 | — | — | |
| PVT-EMCAD-B0#Params=3.92M, #FLOPs=0.84G, Resolution=256x2562024.05 | 90.7 | — | — | — | |
| DRIFA-Netnumber of parameters (M)=67.3, GFLOPS=19.92026.02 | 90.6 | 83.9 | — | — | |
| PVT-CASCADE#Params=34.12M, #FLOPs=7.62G, Resolution=256x2562024.05 | 90.41 | — | — | — | |
| PolypPVTnumber of parameters (M)=25.1, GFLOPS=5.32026.02 | 90.4 | 83.9 | — | — | |
| PVT-CASCADEnumber of parameters (M)=34.1, GFLOPS=7.62026.02 | 90.4 | 84 | — | — | |
| PolypPVT#Params=25.11M, #FLOPs=5.30G, Resolution=256x2562024.05 | 90.36 | — | — | — | |
| SSFormer-L#Params=66.22M, #FLOPs=17.28G, Resolution=256x2562024.05 | 90.25 | — | — | — | |
| MADGNetnumber of parameters (M)=31, GFLOPS=14.22026.02 | 90.2 | 83.8 | — | — | |
| CaraNet#Params=46.64M, #FLOPs=11.48G, Resolution=256x2562024.05 | 90.18 | — | — | — | |
| S2M-NetLoss=MASL2026.01 | 89.99 | — | — | — | |
| UACANet-L#Params=69.16M, #FLOPs=31.51G, Resolution=256x2562024.05 | 89.76 | — | — | — | |
| DINO-MVRReadout Resolution=512+1024, Metric Resolution=256 × 256, Test-Time Augmentation (TTA)=true, DenseCRF refinement=true2026.05 | 89.76 | 82.7 | 12.4 | — | |
| TransFuse#Params=143.74M, #FLOPs=82.71G, Resolution=256x2562024.05 | 89.62 | — | — | — | |
| TransUNetApproach=Deterministic Segmentation Methods2025.12 | 89.4 | 82.2 | — | — | |
| ACC-UNetparams=16.8M, FLOPS=38G2023.08 | 89.37 | — | — | — | |
| TransFuseLayer attention=×, Params(M)=26.16, FLOPs(G)=11.502026.06 | 89.27 | 80.63 | — | — | |
| SwinUNet#Params=27.17M, #FLOPs=6.2G, Resolution=224x2242024.05 | 89.26 | — | — | — | |
| Swin-Unetparams=27.2M, FLOPS=6.2G2023.08 | 89.24 | — | — | — | |
| KCLA-UNetLayer attention=✓, Params(M)=1.55, FLOPs(G)=3.492026.06 | 89.21 | 80.52 | — | — | |
| MTTU-Netnumber of parameters (M)=71.6, GFLOPS=20.92026.02 | 89.2 | 82.6 | — | — | |
| TransUNet#Params=105.32M, #FLOPs=38.52G, Resolution=256x2562024.05 | 89.16 | — | — | — | |
| MRLA-L-UNetLayer attention=✓, Params(M)=1.54, FLOPs(G)=3.482026.06 | 89.11 | 80.36 | — | — | |
| UCTransnetparams=66.4M, FLOPS=38.8G2023.08 | 89.08 | — | — | — | |
| DLA-L-UNetLayer attention=✓, Params(M)=1.57, FLOPs(G)=3.502026.06 | 89.05 | 80.26 | — | — | |
| DeepLabv3+#Params=39.76M, #FLOPs=14.92G, Resolution=256x2562024.05 | 88.64 | — | — | — | |
| SANetLayer attention=×, Params(M)=23.90, FLOPs(G)=5.962026.06 | 88.59 | 79.52 | — | — | |
| SMESwin-Unetparams=169.8M, FLOPS=6.4G2023.08 | 88.57 | — | — | — | |
| PraNet#Params=32.55M, #FLOPs=6.93G, Resolution=256x2562024.05 | 88.56 | — | — | — | |
| MultiResUNetparams=7.3M, FLOPS=1.1G2023.08 | 88.55 | — | — | — | |
| UNeXt-SLayer attention=×, Params(M)=0.32, FLOPs(G)=0.102026.06 | 88.33 | 79.09 | — | — | |
| UTNetV2Layer attention=×, Params(M)=12.80, FLOPs(G)=15.502026.06 | 88.25 | 78.97 | — | — | |
| UNetparams=14M, FLOPS=37G2023.08 | 87.97 | — | — | — | |
| UNet++Layer attention=×, Params(M)=9.16, FLOPs(G)=34.862026.06 | 87.83 | 78.31 | — | — | |
| UNeXt#Params=1.47M, #FLOPs=0.57G, Resolution=256x2562024.05 | 87.78 | — | — | — | |
| ResNet-UNetLayer attention=×, Params(M)=1.53, FLOPs(G)=3.452026.06 | 87.73 | 78.14 | — | — | |
| UNetLayer attention=×, Params(M)=7.77, FLOPs(G)=13.762026.06 | 87.55 | 77.86 | — | — | |
| UNet++#Params=9.16M, #FLOPs=34.65G, Resolution=256x2562024.05 | 87.46 | — | — | — | |
| UNetnumber of parameters (M)=34.5, GFLOPS=65.52026.02 | 87.3 | 80.2 | — | — | |
| UMambaLoss=MASL2026.01 | 87.23 | — | — | — | |
| S2M-NetLoss=standard Dice loss2026.01 | 87.16 | — | — | — | |
| DMApproach=Generative Segmentation Methods2025.12 | 87.09 | 77.14 | — | — | |
| AttnUNet#Params=34.88M, #FLOPs=66.64G, Resolution=256x2562024.05 | 87.05 | — | — | — | |
| UNet#Params=24.53M, #FLOPs=65.53G, Resolution=256x2562024.05 | 86.67 | — | — | — | |
| DuckNetLoss=MASL2026.01 | 86.45 | — | — | — | |
| SwinUNetLoss=MASL2026.01 | 86.34 | — | — | — | |
| TransUNetLoss=MASL2026.01 | 85.89 | — | — | — | |
| SegDINOResolution=256 × 2562026.05 | 85.76 | 77.6 | 17.8 | — | |
| UMambaLoss=standard Dice loss2026.01 | 85.68 | — | — | — | |
| U-Net++Loss=MASL2026.01 | 85.67 | — | — | — | |
| SwinUNetLoss=standard Dice loss2026.01 | 84.91 | — | — | — | |
| RAPUNetLoss=MASL2026.01 | 84.78 | — | — | — | |
| DuckNetLoss=standard Dice loss2026.01 | 84.73 | — | — | — | |
| U-NetLoss=MASL2026.01 | 84.38 | — | — | — | |
| TransUNetLoss=standard Dice loss2026.01 | 84.12 | — | — | — | |
| PraNetLoss=MASL2026.01 | 83.92 | — | — | — | |
| UNetApproach=Deterministic Segmentation Methods2025.12 | 83.59 | 75.5 | — | — | |
| U-Net++Loss=standard Dice loss2026.01 | 83.42 | — | — | — | |
| U-KANResolution=256 × 2562026.05 | 83.41 | 74.62 | 23.57 | — | |
| SegNetResolution=256 × 2562026.05 | 83.27 | 74.46 | 21.41 | — | |
| Att-UNetResolution=256 × 2562026.05 | 82.75 | 73.72 | 26.12 | — | |
| RAPUNetLoss=standard Dice loss2026.01 | 82.56 | — | — | — | |
| CM-TTATrainable Params (K)=1.022026.06 | 82.34 | — | — | 9.39 | |
| U-NeXtResolution=256 × 2562026.05 | 82.3 | 73.27 | 23.63 | — | |
| U-NetLoss=standard Dice loss2026.01 | 82.15 | — | — | — | |
| PraNetLoss=standard Dice loss2026.01 | 81.88 | — | — | — | |
| U-NetResolution=256 × 2562026.05 | 81.87 | 72.95 | 25.12 | — | |
| TransUNetResolution=256 × 2562026.05 | 81.86 | 72.3 | 24.76 | — | |
| R2U-NetResolution=256 × 2562026.05 | 81.02 | 71.34 | 25.36 | — | |
| UNet++Approach=Deterministic Segmentation Methods2025.12 | 81 | 72.9 | — | — | |
| ZEROTrainable Params (K)=0.002026.06 | 80.1 | — | — | 17.82 | |
| TPTTrainable Params (K)=1.022026.06 | 79.6 | — | — | 18.08 | |
| HisTPTTrainable Params (K)=1.022026.06 | 78.95 | — | — | 18.51 | |
| TENTTrainable Params (K)=163.842026.06 | 76.45 | — | — | 21.94 | |
| No adaptTrainable Params (K)=0.002026.06 | 76.27 | — | — | 22.45 | |
| TTLTrainable Params (K)=655.362026.06 | 75.61 | — | — | 20.82 |