Medical Image Segmentation on JSRT
85IoURN-DPO
Evaluation results
| Method | Links | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| RN-DPORegime=GT oracle, Base Model=Strong Base (Nseg=6)2026.01 | 85 | — | — | — | — | — | — | — | 0.855 | — | |
| RN-DPORegime=GT oracle, Base Model=Weak Base (Nseg=2)2026.01 | 82 | — | — | — | — | — | — | — | 0.816 | — | |
| DPORegime=GT oracle, Base Model=Strong Base (Nseg=6)2026.01 | 80 | — | — | — | — | — | — | — | 0.796 | — | |
| RN-DPORegime=Strong judge, N_QC=6/9, Base Model=Strong Base (Nseg=6)2026.01 | 79.8 | — | — | — | — | — | — | — | 0.789 | — | |
| RN-DPO (rand.)Regime=Strong judge, N_QC=6/9, Base Model=Strong Base (Nseg=6)2026.01 | 79.1 | — | — | — | — | — | — | — | 0.781 | — | |
| Top-1 PseudolabelingRegime=Strong judge, N_QC=6/9, Base Model=Strong Base (Nseg=6)2026.01 | 79 | — | — | — | — | — | — | — | 0.781 | — | |
| DPORegime=GT oracle, Base Model=Weak Base (Nseg=2)2026.01 | 77.1 | — | — | — | — | — | — | — | 0.758 | — | |
| IPORegime=Strong judge, N_QC=6/9, Base Model=Strong Base (Nseg=6)2026.01 | 76.8 | — | — | — | — | — | — | — | 0.751 | — | |
| DPORegime=Strong judge, N_QC=6/9, Base Model=Strong Base (Nseg=6)2026.01 | 76.6 | — | — | — | — | — | — | — | 0.743 | — | |
| RN-DPO (rand.)Regime=Weak judge, N_QC=3/6, Base Model=Strong Base (Nseg=6)2026.01 | 76.4 | — | — | — | — | — | — | — | 0.767 | — | |
| RN-DPORegime=Weak judge, N_QC=3/6, Base Model=Strong Base (Nseg=6)2026.01 | 76.2 | — | — | — | — | — | — | — | 0.759 | — | |
| DPO (rand.)Regime=Strong judge, N_QC=6/9, Base Model=Strong Base (Nseg=6)2026.01 | 76.2 | — | — | — | — | — | — | — | 0.717 | — | |
| Select-bestRegime=Strong judge, N_QC=6/9, Base Model=Strong Base (Nseg=6)2026.01 | 75.9 | — | — | — | — | — | — | — | — | — | |
| DPO (rand.)Regime=Weak judge, N_QC=3/6, Base Model=Strong Base (Nseg=6)2026.01 | 75.7 | — | — | — | — | — | — | — | 0.73 | — | |
| rDPORegime=Strong judge, N_QC=6/9, Base Model=Strong Base (Nseg=6)2026.01 | 75.5 | — | — | — | — | — | — | — | 0.739 | — | |
| IPORegime=Weak judge, N_QC=3/6, Base Model=Strong Base (Nseg=6)2026.01 | 75 | — | — | — | — | — | — | — | 0.726 | — | |
| Top-1 PseudolabelingRegime=Weak judge, N_QC=3/6, Base Model=Strong Base (Nseg=6)2026.01 | 74.7 | — | — | — | — | — | — | — | 0.74 | — | |
| rDPORegime=Weak judge, N_QC=3/6, Base Model=Strong Base (Nseg=6)2026.01 | 74.6 | — | — | — | — | — | — | — | 0.72 | — | |
| DPORegime=Weak judge, N_QC=3/6, Base Model=Strong Base (Nseg=6)2026.01 | 74.5 | — | — | — | — | — | — | — | 0.729 | — | |
| Select-bestRegime=Weak judge, N_QC=3/6, Base Model=Strong Base (Nseg=6)2026.01 | 74.1 | — | — | — | — | — | — | — | — | — | |
| Top-1 PseudolabelingRegime=Strong judge, N_QC=6/9, Base Model=Weak Base (Nseg=2)2026.01 | 73.2 | — | — | — | — | — | — | — | 0.713 | — | |
| Supervised baselineRegime=Supervised baseline, Base Model=Strong Base (Nseg=6)2026.01 | 73 | — | — | — | — | — | — | — | — | — | |
| RN-DPORegime=Strong judge, N_QC=6/9, Base Model=Weak Base (Nseg=2)2026.01 | 72.1 | — | — | — | — | — | — | — | 0.704 | — | |
| RN-DPO (rand.)Regime=Strong judge, N_QC=6/9, Base Model=Weak Base (Nseg=2)2026.01 | 71.7 | — | — | — | — | — | — | — | 0.694 | — | |
| IPORegime=Strong judge, N_QC=6/9, Base Model=Weak Base (Nseg=2)2026.01 | 70.1 | — | — | — | — | — | — | — | 0.685 | — | |
| rDPORegime=Strong judge, N_QC=6/9, Base Model=Weak Base (Nseg=2)2026.01 | 69.7 | — | — | — | — | — | — | — | 0.677 | — | |
| DPORegime=Strong judge, N_QC=6/9, Base Model=Weak Base (Nseg=2)2026.01 | 69.6 | — | — | — | — | — | — | — | 0.679 | — | |
| DPO (rand.)Regime=Strong judge, N_QC=6/9, Base Model=Weak Base (Nseg=2)2026.01 | 67.3 | — | — | — | — | — | — | — | 0.642 | — | |
| RN-DPO (rand.)Regime=Weak judge, N_QC=3/6, Base Model=Weak Base (Nseg=2)2026.01 | 63.4 | — | — | — | — | — | — | — | 0.62 | — | |
| RN-DPORegime=Weak judge, N_QC=3/6, Base Model=Weak Base (Nseg=2)2026.01 | 63.1 | — | — | — | — | — | — | — | 0.619 | — | |
| Top-1 PseudolabelingRegime=Weak judge, N_QC=3/6, Base Model=Weak Base (Nseg=2)2026.01 | 62.3 | — | — | — | — | — | — | — | 0.607 | — | |
| Select-bestRegime=Strong judge, N_QC=6/9, Base Model=Weak Base (Nseg=2)2026.01 | 62.3 | — | — | — | — | — | — | — | — | — | |
| DPO (rand.)Regime=Weak judge, N_QC=3/6, Base Model=Weak Base (Nseg=2)2026.01 | 60.8 | — | — | — | — | — | — | — | 0.594 | — | |
| IPORegime=Weak judge, N_QC=3/6, Base Model=Weak Base (Nseg=2)2026.01 | 60.6 | — | — | — | — | — | — | — | 0.588 | — | |
| DPORegime=Weak judge, N_QC=3/6, Base Model=Weak Base (Nseg=2)2026.01 | 60.4 | — | — | — | — | — | — | — | 0.565 | — | |
| Select-bestRegime=Weak judge, N_QC=3/6, Base Model=Weak Base (Nseg=2)2026.01 | 60.2 | — | — | — | — | — | — | — | — | — | |
| rDPORegime=Weak judge, N_QC=3/6, Base Model=Weak Base (Nseg=2)2026.01 | 60.2 | — | — | — | — | — | — | — | 0.562 | — | |
| Supervised baselineRegime=Supervised baseline, Base Model=Weak Base (Nseg=2)2026.01 | 55.8 | — | — | — | — | — | — | — | — | — | |
| Atlas-RAN_test=375, N_cal=3602025.03 | — | — | — | — | — | — | — | — | — | 91.3 | |
| Certified Medical Image Segmentation with Diffusion ModelsTrained with noise=false, σ=0.25, R=0.172023.10 | — | 95 | 91 | 93 | 87 | 78 | 65 | 0.05 | — | — | |
| Certified Medical Image Segmentation with Diffusion ModelsTrained with noise=false, σ=0.5, R=0.342023.10 | — | 94 | 88 | 91 | 83 | 63 | 48 | 0.08 | — | — | |
| Certified Medical Image Segmentation with Diffusion ModelsTrained with noise=false, σ=1, R=0.672023.10 | — | 90 | 82 | 87 | 77 | 28 | 19 | 0.12 | — | — | |
| PHISegN_test=375, N_cal=3602025.03 | — | — | — | — | — | — | — | — | — | 91.7 | |
| PHISeg (CQR)N_test=375, N_cal=3602025.03 | — | — | — | — | — | — | — | — | — | 89.9 | |
| ResUNet++Trained with noise=false, σ=0, R=02023.10 | — | 97 | 94 | 94 | 91 | 93 | 91 | 0 | — | — | |
| ResUNet++Trained with noise=true, σ=0.25, R=02023.10 | — | 91 | 90 | 89 | 87 | 84 | 79 | 0 | — | — | |
| SAM 2N_test=375, N_cal=3602025.03 | — | — | — | — | — | — | — | — | — | 87.1 | |
| SEGCERTIFYTrained with noise=true, σ=0.25, R=0.172023.10 | — | 96 | 92 | 93 | 88 | 83 | 72 | 0.04 | — | — | |
| SEGCERTIFYTrained with noise=true, σ=0.5, R=0.342023.10 | — | 89 | 84 | 85 | 79 | 58 | 43 | 0.13 | — | — | |
| SEGCERTIFYTrained with noise=true, σ=1, R=0.672023.10 | — | 7 | 4 | 2 | 1 | 0 | 0 | 0.24 | — | — | |
| UniverSegN_test=375, N_cal=3602025.03 | — | — | — | — | — | — | — | — | — | 93.2 |