Medical Image Lesion Detection and Diagnosis on PI-CAI ideal scenario (fold-0 val)
86.57AUROCMIGFNet-UNet
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| MIGFNet-UNetParams=52.60M2026.04 | 86.57 | 58.58 | 72.57 | 48.55 | 91.19 | 62.13 | |
| MIGFNet-nnUNetParams=9.45M, optimal configuration excludes deep supervision=true2026.04 | 86.36 | 59.72 | 73.04 | 48.69 | 94.29 | 46.85 | |
| MIGFNet-MambaParams=18.67M, optimal configuration excludes deep supervision=true2026.04 | 86.14 | 55.58 | 70.86 | 43.62 | 91.43 | 67.87 | |
| Bare UNetParams=31.80M2026.04 | 86.1 | 55.11 | 70.61 | 47 | 97.14 | 35.28 | |
| Bare nnUNetParams=7.11M2026.04 | 85.23 | 54.4 | 69.81 | 47.22 | 92.38 | 56.02 | |
| Bare MambaParams=9.82M2026.04 | 78.13 | 46.86 | 62.5 | 41.65 | 95 | 26.3 |